Plant pathogenic bacteria cause fatal diseases in their hosts, leading to reduced agricultural productivity, which can eventually result in massive economic losses. Of these plant pathogenic bacteria, most belong to the following genera within the phylum Proteobacteria: Pseudomonas, Ralstonia, Burkholderia, Agrobacterium, Xanthomonas, and Pectobacterium. Within a distinct environment of hosts, the pathogenicity of plant pathogens is a complex activity achieved through a myriad of virulence factors that enable the acquisition of resources, overcoming of host defenses, and production of toxic molecules (Kim et al., 2004; Gudesblat et al., 2009; Dalsing et al., 2015). Until now, most studies have been limited to the functional validation of only one or a few virulence factors, in a single pathogen, as opposed to revealing an entire landscape of plant pathogenesis or common virulence factors across phytopathogenic species (Sundin et al., 2016). However, to develop practical strategies for the control of bacterial diseases, the convergence and/or divergence of virulence mechanisms should be fully understood. Advances in next-generation sequencing technology have accelerated comparative genomic studies thanks to the elucidation of the genomes of various organisms. The genomic architecture of pathogens is an important resource of adaptive evolution for specialized lifestyles in each host niche (Jones et al., 2021). However, only genome-dependent analyses may not define individual characteristics among pathogens due to the deviation of unique gene repertoires by the population size. It is also difficult to consider the importance of regulatory genes that affect multiple systems, including infection mechanisms or adaptation processes within plant host tissues. Hence, a universal model based on the expression patterns will help to elucidate the complexity of global plant-pathogen interactions and allow for the identification of novel virulence factors.

Comparative analysis of in planta transcriptomes across plant pathogens may be one of the best solutions to fill this knowledge gap. In planta transcriptome analysis involves monitoring the gene expression profiles of pathogens during infection. The genome-wide modification of transcriptome during in planta infection is the inevitable outcome of bacterial adaptation to a hostile environment (Yadeta and Thomma, 2013). Although several studies have investigated the transcriptome of pathogens, either in plant-mimicking media or in whole plant extracts (Ciraulo et al., 2010; Yazawa et al., 2013), the quality or quantity of the RNAs only partially reflect the in planta conditions. Obtaining a high-quality transcriptome of pathogenic bacteria directly from the infected tissues is a crucial step in any such analysis (Azhikina et al., 2010; Skvortsov and Azhikina, 2010). Some previous studies reported high-quality in planta transcriptomes of plant pathogenic bacteria (Jacobs et al., 2012; Kim et al., 2014).

As a destructive pathogenic bacterium, Ralstonia solanacearum causes bacterial wilt in most solanaceous plants, including tomato, tobacco, and potato (Gerlin et al., 2021; Nion and Toyota, 2015). Jacobs et al. (2012), characterized the in planta transcriptome of R. solanacearum GMI1000 within tomato stems. Another extensively studied plant pathogen is Burkholderia glumae, which causes grain and seedling rot in rice and bacterial wilt in many field crops, including tomato, eggplant, and pepper (Ham et al., 2011; Riera-Ruiz et al., 2014). We previously reported the in planta transcriptional profiles of B. glumae BGR1 in rice stems (Kim et al., 2014). A comparison of independently acquired in planta transcriptomes with in vitro transcriptomes revealed substantial changes in the expression profiles of genes involved in various metabolic and signaling pathways in both R. solanacearum and B. glumae (Jacobs et al., 2012; Kim et al., 2014). In the present study, to increase the number of available bacterial transcriptomes, we constructed in planta transcriptomes of Xanthomonas oryzae pv. Oryzae (Xoo) KACC10331 within rice leaves, using high-throughput RNA sequencing (RNA-seq). Xoo is the causative agent of bacterial leaf blight, which is responsible for losses in rice yield of up to 50% worldwide (Niño-Liu et al., 2006).

Therefore, expanding our understanding of the transcriptomes of plant pathogenic B. glumae, R. solanacearum, and Xoo will support the biological interpretation of pathogenicity and provide a valuable reference for comparative analyses involving multi-species of phytopathogenic bacteria. Given a sessile and autotrophic lifestyle that limits resource availability, land plants face a fundamental dilemma: either efficiently allocate limited resources to optimal growth or economically invest these resources in defense strategies against invaders (Pageau et al., 2006; Schultz et al., 2013; Veillet et al., 2016). Unlike the immune system of animals, plant immunity consists of broad and slightly less complex strategies that have developed through evolution and are shared among plant species (Tao et al., 2003; Maekawa et al., 2011). Accordingly, comparing the in planta transcriptomes of different pathogens enables the identification of fundamental principles for their survival in response to integrated plant conditions that transcend different hosts, infection sites, or environments. In addition, B. glumae shares a broad host range with Xoo in rice and with R. solanacearum in vegetable crops. B. glumae can attack diverse infection sites, including grains, stems, and seedlings. Similar to the other two pathogens, it can also infect vascular systems. The genome organization of B. glumae shows a high degree of plasticity to cope with such a variety of environments (Seo et al., 2015). From this perspective, the genomic information of B. glumae can be applied as a central guide to identify the common factors employed by three plant pathogens.

This study was conducted to compare the in planta transcriptomes across three major bacterial phytopathogens in an attempt to infer a universal model of their dynamic interactions with plant hosts. A pan-genome guided map was utilized to characterize the highly conserved in planta co-expression patterns among the three phytopathogens. In addition, we aim to identify the key functions of genes comprising the common virulence of the genomes of studied species and validate their involvement in virulence through mutagenesis experiments. The provided methodology represents an applicable strategy for investigating in planta interactions across plant pathogens and identifying common virulence factors expressed during infection.

In recent years, the comparative analysis of transcriptomes across species has emerged as a framework for studying similarities in gene expression and could reveal novel unique and common genetic factors (Gerstein et al., 2014; Czaban et al., 2015; Sudmant et al., 2015). This concept is based on a universal model, containing a single set of organism-independent parameters that despite millions of years of evolutionary divergence, can illustrate the fundamental principles of conserved transcription (Gerstein et al., 2014; Sudmant et al., 2015). In this study, cross-species in planta transcriptome analysis, was used to characterize the gene expression pattern and to identify the common factors utilized by various three different phytopathogenic bacteria (B. glumae BGR1, R. solanacearum GMI1000, and Xoo KACC10331) to establish dynamic interactions with their hosts. The selected bacterial phytopathogen species in this study are causing serious diseases to their host and B. glumae shares common hosts with both R. solanacearum GMI1000, and Xoo. Therefore, the genome sequence of B. glumae served as cornerstone for comparisons with the other two species. Moreover, the identified common up-regulated 4 key functional genes from the in planta transcriptome were first tested in B. glumae then the influential gene locus on the virulence was verified in following mutagenesis experiments in the other two species. Ultimately the PCD was identified as a common cross-species virulence-related gene in the tested bacterial species following the workflow illustrated in Figure 7 .

Despite the evolutionary distance of these pathogens, many genetic elements were highly conserved at a genome-wide scale ( Figure 1 ). The microbial core genome supports the basic functional aspects of organisms associated with essential phenotypes (Tettelin et al., 2005). We discovered that the orthologous groups of genes included processes such as energy production and the flow of genetic information, which are crucial for bacterial life. In contrast, non-orthologous groups, dominated by proteins of unknown function and transposons, reflected the characteristics of the unique bacterial genome. Dispensable and unique genes are responsible for the overall microbial fitness in environmental niches (Tettelin et al., 2005). Indeed, the in planta transcriptome analysis of the unique genomes emphasized the optimized mechanisms that each pathogen developed to fully trigger plant disease ( Figure 3 ). It should be noted that B. glumae BGR1 was originally isolated from the diseased rice grain (Lim et al., 2009). In comparison with other tissues, the rice stem provides an efficient waterway for plant pathogens to the upper plant parts (Yadeta and Thomma, 2013). B. glumae would attempt to reach the grain to obtain abundant nutrients by activating most of the motility-related systems. In R. solanacearum, nitrogen has been reported to play an important role in ATP production, detoxification, and bacterial virulence (Dalsing et al., 2015). Dalsing et al. demonstrated that nitrogen source limitation significantly decreases bacterial virulence in the tomato host (Dalsing et al., 2015). Also, the bacterial secretion system uniquely induced in Xoo is a major mechanism for the delivery of virulence factor to the host and environment (Furutani et al., 2009). Although the function of the type VI secretion system in Xoo has less been characterized, many studies recently revealed that the type VI secretion system directly affects the virulence and survival within hostile environments (Ma et al., 2014; Shyntum et al., 2015). Therefore, the in planta-dependent activation of non-orthologous genes as well as orthologous genes would be an essential element of convergent strategies to dominate plant hosts.

Overlaying in planta expression profiles onto the pan-genome revealed the co-expression patterns associated with global interactions with a hostile environment. The dominant feature of co-downregulated genes was their disproportionate involvement in energy metabolism with ATP binding ( Figure 2 ). Nutrient acquisition within plant hosts poses a fundamental challenge to energy production by pathogens because of the extremely low solubility and availability of major nutrients (Jones and Wildermuth, 2011; Yadeta and Thomma, 2013). Moreover, the plant defense mechanism of nutrient allocation from infected sites contributes to the difficulty associated with the acquisition of sufficient nutrients by pathogens (Pageau et al., 2006; Schultz et al., 2013; Veillet et al., 2016). In contrast to nutrient-rich in vitro conditions, a lack of ATP-linked energy in hostile environments may result in the reduction of the proportion of cellular components identified in the orthologous groups, as well as in the inhibition of the associated essential functions.

The three pathogens also employed alternative strategies to overcome their survival limitations and establish pathogenicity in a hostile environment. We found that the co-expression patterns were involved in a signal transduction cascade ( Figure 2 ). The ability of pathogens to dominate plant hosts depends on the functional products that mediate nutrient acquisition and resistance to defense mechanisms, as well as on the signal transduction that allows induction of the expression of these products when and where they are needed. The first step in any signal transduction is signal detection by a membrane protein. In this study, 20 MCPs, acting as membrane sensors, were found to be commonly upregulated among three pathogens. Signal recognition by MCPs provides insights into the molecular events governing the related pathways, including flagellum biosynthesis (Berleman and Bauer, 2005), degradation of xenobiotic compounds (Luu et al., 2015), and toxin production (Harkey et al., 1994). Therefore, our results revealed that the three pathogens consistently use signal transduction to respond to different hosts, infection sites, or environments. In silico analysis of the bacterial genome showed that 54% of 450 genomes possess signal transduction genes (Wuichet and Zhulin, 2010). The abundance of MCPs in bacteria varies with the metabolic versatility, habitat stability, and interaction with other species. Interestingly, Matilla and Krell (2018) revealed that the abundance of these genes in plant pathogens is higher than that of overall bacteria and that of animal/human pathogens. The plant condition is a complex environment with both low nutrient availability, defense mechanisms, abiotic stresses, and competition from native microorganisms. The signal transduction of plant pathogens in this hostile environment is essential to promote the colonization of more favorable niches within host tissues.

Therefore, there is a need to determine common factors among plant pathogens. Co-upregulated DEGs in the studied bacterial genomes could represent key functions within the in planta common virulence model ( Table 1 ). The onset of bacterial motility through the production of a flagellar structure is the primary strategy for coping with environmental changes. Three pathogens showed high flhA expression, which encodes the export apparatus for flagellum assembly. Bacterial motility enables movement towards favorable environments, as well as harm avoidance, in response to diverse signals (Kim et al., 2007). Given nutrient limitations or considerable energy requirements, abundant carbon sources in plant hosts are likely to be attractive signals for pathogens. As central metabolites, C4-dicarboxylates, including succinate, fumarate, and malate, have been identified in various plant tissues (Janausch et al., 2002; Mellgren et al., 2009; Tamir-Ariel et al., 2011). Thus, the activation of C4-dicarboxylate transporters for the uptake of plant-derived carbon sources has been suggested as an essential strategy for the survival of pathogens in hostile environments. For example, some pathogens of the genera Pseudomonas and Xanthomonas prefer C4-dicarboxylates to other carbon sources within plants (Mellgren et al., 2009; Tamir-Ariel et al., 2011). In addition, Pseudomonas syringae pv. tomato DC3000 mutants lacking the dicarboxylate transporter do not grow to the level of the wild-type strain in planta (Mellgren et al., 2009). We revealed that the mutation of a C4-dicarboxylate transporter, as well as flagellar biosynthesis protein, is directly related to the virulence of B. glumae, indicating their importance for both survival and colonization ( Figure 4 ).

In hostile environments, MICL may play another key function related to the conversion of external compounds into available resources. MICL, a signature enzyme of the methylcitrate cycle, protects microorganisms by detoxifying propionate, a cell-toxic compound abundant in soil (Lee et al., 2009). The methylcitrate cycle supplies succinate and pyruvate, which are used by pathogens to produce energy and biomass in a nutrient-limited environment (Dubey et al., 2013). Studies have revealed that MICL mutants show no significant differences in phenotype with the wild-type strain (Lee et al., 2009; Goo et al., 2017), similar to our virulence assay results ( Figure 4 ). This phenomenon is due to the cross-activity of isocitrate lyase (ICL), which has evolved from an ancestral gene performing a different function by gene duplication. The mutation of MICL in Fusarium graminearum did not affect growth or virulence; however, the double mutation of MICL and ICL resulted in reduced virulence against plant hosts (Lee et al., 2009). Similarly, Goo et al. (2017), reported that MICL and ICL double mutants of B. glumae exhibited a significantly reduced oxalate production, resulting in decreased plant-mediated alkaline toxicity. Indeed, ICL of B. glumae (BGLU_RS06215) and R. solanacearum (RS_RS06810) showed high expression levels, with ≥5-fold changes in planta ( Table S4 ). In contrast, the absence of ICL in Xoo led to the highest in planta expression of MICL among three pathogens for an alternative requirement ( Table 1 ). Although the role of MICL has been studied less extensively in pathogenic bacteria than in fungi, the 50% sequence identity between fungal and bacterial MICL genes suggests the importance of further studying this gene as a virulence factor.

The other gene of interest encodes PCD, an enzyme responsible for PCA ring cleavage. Defects in PCD resulted in a significant reduction in virulence at the major infection site in all three pathogens ( Figure 5 ). PCA is regarded as a beneficial phenolic compound because of its antibacterial activity (Chao and Yin, 2009); it is widely distributed across edible plants and human foods (Kakkar and Bais, 2014). Here, we demonstrated an intimate relationship between PCD and bacterial resistance to PCA by measuring the effect of PCA on bacterial growth ( Figures 6A , B ). The expression of PCD, a key enzyme in the β-ketoadipate pathway, may represent an essential countermeasure of pathogens against plant defense mechanisms. Thus, plant hosts synthesize PCA to eliminate invaders from their tissues (Kakkar and Bais, 2014), whereas pathogens may express PCD to protect themselves from antibacterial activity. Although the β-ketoadipate pathway is present in most bacteria that recognize aromatic compounds, it is mainly distributed among soil microorganisms (Harwood and Parales, 1996; Li et al., 2010). Further, a BLASTp-based survey (>30% identity, >30% coverage, and <1.0 × 10^-5 E value cutoffs to define a match) of the distribution of PCD across 30 representative plant pathogens deposited in the NCBI genome database highlighted the ubiquity of this gene in most pathogens (22 matches, 73.3%) ( Table S9 ). At the genus level, plant pathogens were grouped into the same clade, respectively ( Figure S8 ).

Interestingly, we confirmed that the production of toxoflavin by the WBI23755 mutant was reduced in the presence of PCA ( Figure 6D ). Toxoflavin causes chlorotic damage to the panicles and inhibits the growth of leaves and roots of rice, leading to severe crop losses (Jeong et al., 2003). The gene cluster of the tox operon and the genetic regulation for toxoflavin biosynthesis and transport have been well-characterized in B. glumae. The major regulators, a LysR-type, ToxR and ToxJ regulators, were shown to be involved in the activation of tox operons for biosynthesis and transport. The process is also regulated by the quorum sensing system through tofI, and tofR, via N-acyl homoserine lactone signals (Mannaa et al., 2018). In this study we have shown that PCD defects resulted in down-regulation of the tofI, tofR, toxJ, toxA and toxR which could explain the reduction of toxoflavin production. Nevertheless, the relationship between PCD and toxoflavin production remains unclear. This enzyme may be involved in the metabolism of pigments and toxoflavin in B. glumae. The qPCR results revealed a negative effect of PCA on the expression of major genes involved in toxoflavin production, depending on the presence of PCD ( Figure 6E ). We are currently investigating the nature of the brown pigment produced by the WBI23755 mutant under PCA supplementation ( Figure 6C ). Karki et al. (2012), reported that plant pathogenic B. glumae 411gr-6 produces various pigments (brown, purple, and fluorescent compounds) different from toxoflavin under specific conditions. It has been suggested that these alternative pigments are likely to be involved in resistance to environmental stresses (Karki et al., 2012).

Although the exact roles of PCD in relation to phytopathogenic bacterial virulence is yet unknown, the fact that it is involved in the degradation of aromatic compounds supports the suggestion that it could be employing bacteria with ecological advantage. Previous studies have also shown that another important phytopathogen, the Agrobacterium tumefaciens could utilize the same function to promote its pathogenicity allowing the bacteria to overcome the toxicity of the aromatic compounds present around wounded plant tissues that might be detrimental to bacterial survival (Parke, 1997; Parke, 2000). However, further work is required to unravel the exact link between PCD and the bacterial phytopathogen as our investigation is suggesting its ubiquity as a virulence factor.

Our study attempted to construct a universal virulence model for representative plant pathogenic bacteria, namely B. glumae, R. solanacearum, and Xoo. Comparative analysis of in planta transcriptomes among species allowed us to identify putative common factors related to global plant-pathogen interactions, which could not be found by independent analysis of each pathogen. Co-expression patterns showed consistent activation of MCP-based signal transduction to recognize the external conditions within hosts. The following key functions in the core genome showed alternative strategies for the survival and colonization of hostile environments: (i) bacterial movement to preferred niches; (ii) uptake of carbon sources to overcome nutrient limitation; (iii) conversion of external compounds into available resources; and (iv) degradation of plant-derived antibacterial substances. Among these, the disruption of PCD, a novel virulence factor, led to mild disease symptoms in plants infected with all pathogens, and in particular, it had a critical effect on toxoflavin production by B. glumae. In conclusion, comparative in planta transcriptome analysis across plant pathogens can be a useful approach for understanding the pathogenic lifestyle within plant hosts. As such, it can contribute to unraveling key virulence factors, thereby aiding in the development of novel plant defense strategies.

The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found in the article/ Supplementary Material .

JP and Y-SS conceived and designed the project. JP, SL, NK, and GH conducted the experiments. D-SP, Se-WL, and Sa-WL helped conduct some of the experiments. JP, MM, HJ, and H-HL analyzed the data. JP, MM, and Y-SS wrote the manuscript with contributions from other authors. All authors read and approved the manuscript.

This research was supported by the Basic Science Research Program of the National Research Foundation (NRF), funded by the Ministry of Education (2022R1A2C1003190), (2020R1A6A1A03047729), Republic of Korea and by the National Institute of Fisheries Science (NIFS), funded by the Ministry of Oceans and Fisheries of the South Korea (R2022045).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.