β‐1,4‐Cellobiohydrolase is involved in full expression of phcA, contributing to the feedback loop in quorum sensing of Ralstonia pseudosolanacearum strain OE1‐1

Abstract After infecting roots of tomato plants, the gram‐negative bacterium Ralstonia pseudosolanacearum strain OE1‐1 activates quorum sensing (QS) to induce production of plant cell wall‐degrading enzymes, such as β‐1,4‐endoglucanase (Egl) and β‐1,4‐cellobiohydrolase (CbhA), via the LysR family transcriptional regulator PhcA and then invades xylem vessels to exhibit virulence. The phcA‐deletion mutant (ΔphcA) exhibits neither the ability to infect xylem vessels nor virulence. Compared with strain OE1‐1, the egl‐deletion mutant (Δegl) exhibits lower cellulose degradation activity, lower infectivity in xylem vessels, and reduced virulence. In this study, we analysed functions of CbhA other than cell wall degradation activity that are involved in the virulence of strain OE1‐1. The cbhA‐deletion mutant (ΔcbhA) lacked the ability to infect xylem vessels and displayed loss of virulence, similar to ΔphcA, but exhibited less reduced cellulose degradation activity compared with Δegl. Transcriptome analysis revealed that the phcA expression levels in ΔcbhA were significantly lower than in OE1‐1, with significantly altered expression of more than 50% of PhcA‐regulated genes. Deletion of cbhA led to a significant change in QS‐dependent phenotypes, similar to the effects of phcA deletion. Complementation of ΔcbhA with native cbhA or transformation of this mutant with phcA controlled by a constitutive promoter recovered its QS‐dependent phenotypes. The expression level of phcA in ΔcbhA‐inoculated tomato plants was significantly lower than in strain OE1‐1‐inoculated plants. Our results collectively suggest that CbhA is involved in the full expression of phcA, thereby contributing to the QS feedback loop and virulence of strain OE1‐1.


| INTRODUC TI ON
Bacteria monitor quorum sensing (QS) signals to track changes in abundance and to activate QS for the synchronous control of the expression of genes beneficial for vigorous replication, adaptation to environmental conditions, and virulence (Ham, 2013;Rutherford & Bassler, 2012). The soilborne gram-negative β-proteobacterial Ralstonia solanacearum species complex (RSSC) is globally distributed under diverse environmental conditions. The RSSC, which infects more than 250 plant species belonging to over 50 families, causes potentially devasting bacterial wilt disease and seriously affects plant production worldwide (Mansfield et al., 2012). The phylotype I Ralstonia pseudosolanacearum strain OE1-1 (Kanda et al., 2003) produces methyl 3-hydroxymyristate (3-OH MAME) as a QS signal (Kai et al., 2015;Ujita et al., 2019). This QS signal, which is synthesized by the methyltransferase PhcB, is sensed through the sensor histidine kinase PhcS (Hikichi et al., 2017), leading to the phosphorylation of two regulators, PhcQ and PhcR, that strongly and partially contribute to the regulation of QS-dependent genes, respectively, via the LysR family transcriptional regulator PhcA (Takemura et al., 2021).
The sensor histidine kinase PhcK is required for full expression of phcA independently of 3-OH MAME sensing (Senuma et al., 2020).
RSSC strains in soil invade plant roots through wounds or natural openings from which secondary roots emerge. After colonizing the intercellular spaces of the root cortex and vascular parenchyma, the bacteria eventually enter xylem vessels and spread into the stem and leaves through the xylem (Araud-Razou et al., 1998;Vasse et al., 1995). Liu et al. (2005) have demonstrated that the virulence of the phylotype I R. pseudosolanacearum strain GMI1000 involves six plant cell wall-degrading enzymes (PCWDEs), namely endo-polygalacturonase, exo-polyα-galacturonosidase, galacturan 1,4α-galacturonidase, pectin methyl esterase, β-1,4cellobiohydrolase (CbhA), and β-1,4-endoglucanase (Egl), all secreted through the type II secretion system (T2SS; Liu et al., 2005). Tsujimoto et al. (2008) have shown that the cell wall degradation activity of PCWDEs secreted through the T2SS is required for the virulence of strain OE1-1 and its infection of xylem vessels. Furthermore, PhcA positively regulates the expression of egl and cbhA in strain OE1-1 in the active QS state (Mori et al., 2018).
Strain OE1-1 preferentially attaches to the surface of tomato root elongation zones and then colonizes the intercellular spaces between the epidermis and cortex (Inoue et al., 2023). This colonization leads to the degradation of cell walls of cortical cells adjacent to the epidermis. To infect xylem vessels and achieve virulence, strain OE1-1 then forms mushroom-shaped biofilms in the degraded cortical cells. The phcA-deletion mutant (ΔphcA) and a T2SS-deficient mutant have no cellulose degradation ability and thus cannot infect cortical cells of tomato roots and exhibit virulence in tomato plants.
Compared with strain OE1-1, the egl-deletion mutant (Δegl) exhibits lower cellulose degradation activity and reduced infectivity in cortical cells, in turn leading to lowered infectivity in xylem vessels and significantly reduced virulence. Cellulose degradation activity is thus believed to be positively corelated with the virulence of R. pseudosolanacearum strains. Nevertheless, the function of PCWDEs other than Egl in the infection of tomato roots by strain OE1-1 remains elusive.
In this study, we aimed to elucidate the function of one PCWDE, CbhA, in the infection of tomato roots by strain OE1-1. To achieve this goal, we first analysed the cellulose degradation activity and virulence of the cbhA-deletion mutant (ΔcbhA). In addition, we examined the alternative function of CbhA in the virulence of strain OE1-1.

| Cellulose degradation activity of ΔcbhA
To elucidate the cellulose degradation activity of CbhA, we first generated the deletion mutant ΔcbhA (Table 1) and compared this mutant with other strains. Two deletion strains, the T2SS-deficient mutant Shin (Table 1; Tsujimito et al., 2008) and ΔphcA (Table 1; Mori et al., 2016), displayed loss of cellulose degradation activity (Figure 1), whereas Δegl (Table 1; Inoue et al., 2023) exhibited significantly reduced cellulose degradation activity compared with strain OE1-1 (Figure 1; p < 0.05). Although ΔcbhA had significantly lower cellulose degradation activity compared with strain OE1-1 (Figure 1; p < 0.05), egl deletion caused a more significant reduction than did chbA deletion (Figure 1; p < 0.05). These results suggest that the contribution of CbhA to the cellulose degradation activity of strain OE1-1 is smaller than that of Egl.

| Virulence of ΔcbhA in tomato plants
We next analysed the effects of cbhA deletion on R. pseudosolanacearum virulence. All tomato plants inoculated with strain OE1-1 exhibited wilt symptoms and were dead by 10 days after inoculation. Δegl exhibited significantly reduced virulence compared with strain OE1-1 (p < 0.05, t test; Figure 2). On the contrary, ΔcbhA had lost its virulence, similar to ΔphcA and Shin. These results suggest that CbhA, similar to PhcA, is a major contributor to the virulence of strain OE1-1 in tomato plants.

| The behaviour of ΔcbhA in tomato roots
To observe the behaviour of ΔcbhA in tomato roots, we placed 4-day-old tomato seedlings with approximately 20-mm-long roots on agar plates and added a suspension of R. pseudosolanacearum strain OE1-1, ΔphcA, or ΔcbhA 10 mm away from the roots (Inoue et al., 2023). Longitudinal semi-thin (800-nm) resin sections of the tomato roots were stained with toluidine blue and observed under an optical microscope. After 36 h of co-incubation (HOI), strain OE1-1 had infected cell wall-denatured cortical cells adjacent to the epidermis in the root elongation zone (Figure 3). At the same time point, ΔphcA and ΔcbhA cells were observed in intercellular spaces between the epidermis and cortex. At 48 HOI, mushroom-shaped biofilms of strain OE1-1 were observed in the cell wall-denatured cortical cells. By contrast, we observed only a few ΔcbhA cells but no ΔphcA cells in intercellular spaces of the cortex at 48 HOI.
Furthermore, neither ΔphcA nor ΔcbhA cells were observed in cortical cells. At 72 HOI, OE1-1 cells were found in intercellular spaces between the cortex and endodermis. At 120 HOI, OE1-1 cells were observed in intercellular spaces of the cortex and endodermis and also in cell wall-denatured pericycle cells without nuclei and xylem vessels, but no ΔphcA or ΔcbhA cells were detected in either location. These observations suggest that cbhA deletion leads to a loss in infectivity in cortical cells and subsequent loss of infection in xylem vessels and virulence, similar to phcA deletion. According to these results, CbhA may have a wider range of roles than just cellulose degradation in OE1-1 virulence.

| RNA-seq transcriptome analysis of ΔcbhA
To identify QS signalling pathways and QS-dependent genes of strain OE1-1, we previously performed a transcriptome analysis based on RNA sequencing (RNA-seq) of R. pseudosolanacearum strains grown in quarter-strength M63 until OD 600 = 0.3 and then mapped the RNA-seq reads on the GMI1000 genome. Mapping of RNA-seq reads of the OE1-1 strain to the genome of strain GMI1000 (Salanoubat et al., 2002) resulted in the identification of 4370 protein-coding transcripts (Table S1). To extract genes with significant expression changes, the following thresholds were applied: q value < 0.05 and log 2 (fold change) ≥ |2|.
Compared with their expression levels in OE1-1, 261 genes in ΔcbhA exhibited significantly downregulated expression and were TA B L E 1 Strains and plasmids used in this study.

| Gene Ontology enrichment analysis in ΔcbhA
To characterize the functions of genes affected by cbhA deletion, we performed a Gene Ontology (GO) enrichment analysis.
The analysis was applied to three gene categories, namely, those whose expression was affected by phcA deletion, cbhA deletion, or both. Among positively regulated genes, typical GO terms, including "lipopolysaccharide biosynthetic process," were enriched in 255 genes affected by both genotypes (Table 2). Among negatively regulated genes, GO terms related to bacterial motility, such as "chemotaxis" and "bacterial-type flagellum-dependent cell motility," were enriched in 99 genes with significantly reduced expression in both ∆cbhA and ∆phcA. Some terms related to bacterial-type flagella were only enriched in genes with significantly reduced expression in ∆cbhA, whereas terms related to "translation" were enriched in ∆phcA-affected genes but not in ∆cbhA-affected genes.

| QS-dependent phenotypes of ΔcbhA
To assay the influence of CbhA on QS activity, we first examined the in vitro biofilm formation of ΔcbhA and the complemented ΔcbhA strain cbhA-comp (Table 1). Similar to ΔphcA, ΔcbhA exhibited significantly less biofilm formation than did strain OE1-1 (Figure 5a; p < 0.05). In addition, significantly more biofilm formation was observed in strain cbhA-comp than in ΔcbhA (p < 0.05).
The transcriptome analysis revealed that cbhA deletion led to significantly reduced expression levels of epsB, a gene related to EPS I production. We next assayed EPS I production by ΔcbhA and cbhA-comp. cbhA deletion led to significantly reduced production of EPS I (Figure 5b; p < 0.05), similar to the effect of phcA deletion, whereas strain cbhAcomp produced significantly more EPS I than did ΔcbhA (p < 0.05).
The transcriptome analysis also indicated that cbhA deletion led to significantly enhanced expression levels of the flagellar biogenesis-related gene fliC, which is required for the swimming motility of strain OE1-1. Compared with strain OE1-1, ΔcbhA exhibited significantly enhanced swimming motility, similar to ΔphcA

| Expression levels of QS-related genes in ΔcbhA in infected tomato roots
To analyse the expression level of phcA in R. pseudosolanacearum strains used to infect tomato roots, 8-week-old tomato plants were inoculated with R. pseudosolanacearum strains using a root-dip inoculation procedure. Reverse transcription-quantitative PCR (RT-qPCR) was then carried out using total RNA isolated from tomato roots 3 days after inoculation. No significant differences in the expression levels of phcB, phcK, or phcQ were detected between strain OE1-1 and ΔcbhA. By contrast, the expression level of phcA was significantly lower in ΔcbhA than in strain OE1-1 (p < 0.05, t test; Figure 6).
TA B L E 2 Gene Ontology enrichment analysis of genes significantly affected by cbhA deletion.
reduced the expression of more than 50% of PhcA-regulated genes, including not only PCWDE genes, such as egl, but also virulence-  (Figure 2). We thus infer that cbhA deletion leads not only to significantly reduced cellulose activity, but also to significantly reduced expression of phcA and subsequent virulence loss.
Transcriptome analysis coupled with RT-qPCR and RNA-seq revealed that cbhA deletion led to approximately 10% reduced expression of phcA. According to the GO enrichment analysis, GO terms "lipopolysaccharide biosynthetic process," "acylphosphatase activity," and "cellular biosynthetic process" included in cluster IV were enriched in 255 genes with significantly reduced expression levels in both ΔphcA and ΔcbhA (Table 2). Furthermore, the GO terms "chemotaxis" and "bacterial-type flagellum-dependent cell motility" included in cluster I were enriched in 99 genes with significantly reduced expression levels in both mutants. We thus conclude that the Our transcriptome analysis showed that cbhA deletion led to significantly reduced expression of phcA. One possible idea to explain how cbhA contributes to phcA regulation is feedback regulation of QS. We previously reported several mechanisms of QS feedback regulation in strain OE1-1. In the active state of QS, PhcA regulates the expression of QS-dependent genes responsible for QS-dependent phenotypes, including virulence, to induce production of virulence-related aryl-furanone secondary metabolites, ralfuranones (Kai et al., 2014(Kai et al., , 2016, and the major exopolysaccharide EPS I (Genin & Denny, 2012;Schell, 2000). These secondary metabolites are associated with the feedback loop of QS-dependent gene regulation by PhcA (Hayashi, Senuma, et al., 2019;Mori et al., 2018). In the QS active state, expression of lecM, encoding the lectin LecM, is induced, and LecM affects the activation of QS by regulating the stability of extracellularly secreted 3-OH MAME . cbhA deletion led to a significant reduction in ralfuranone production-related ralA, EPS I production-related epsB and lecM, and phcA. This phenomenon may inhibit the feedback loop of QS-dependent gene regulation by PhcA. In regard to another pathway regulating phcA, PhcK is reportedly required for full expression of phcA (Senuma et al., 2020). Nevertheless, cbhA deletion did not lead to significantly reduced expression levels of phcK.
Taken together, these results suggest that CbhA contributes to the feedback regulation of QS by positively regulating phcA expression independently of PhcK.
In this study, R. pseudosolanacearum strains were incubatorgrown in quarter-strength M63 medium lacking sugars such as

| General DNA manipulations
We used standard techniques (Sambrook et al., 1989) to manipulate DNA and determined DNA sequences using an Automated DNA Sequencer Model 373 (Applied Biosystems). The resulting DNA sequences were analysed using the program DNASYS-Mac (Hitachi Software Engineering). HindIII-digested pK18mobsacB vector (Kvitko & Collmer, 2011) to produce the recombinant plasmid pdelta-cbhA. The plasmid was electroporated into OE1-1 competent cells, which were prepared as previously described by Mori et al. (2016). The kanamycinsensitive, sucrose-resistant recombinant ΔcbhA (Table 1) was then selected.

| Creation of a cbhA-deletion mutant and complementation constructs
The fragment c-cbhA was amplified by PCR from the genomic DNA of strain OE1-1 using primers delta-cbhA-1FW and delta-cbhA-2RV. The generated fragment was then digested with BamHI and HindIII and ligated into a BamHI-and HindIII-digested pUC18-mini-Tn7-Gm vector (Choi et al., 2005) to create pCcbhA.
To analyse the recovered QS-dependent phenotypes of ΔcbhA after transformation with phcA under the control of a QSindependent, constitutive active promoter of the RSc0480 gene, the recombinant plasmid p0480prophcA (Table 1; Senuma et al., 2020) carrying the promoter of the RSc0480 gene and phcA based on pUC18-mini-Tn7T-Gm was inserted into ΔcbhA to create the transformant phcA-compΔcbhA (Table 1).

F I G U R E 6
Expression of quorum sensing-related genes phcB, phcK, phcQ, phcR, and phcA in Ralstonia pseudosolanacearum OE1-1 and a cbhA-deletion mutant (ΔcbhA) after inoculation of 8-week-old tomato plants using a root-dip inoculation procedure. RNA was extracted from the roots of tomato plants 3 days after inoculation as described in the Experimental Procedures section. The rpoD gene was used as an internal control for reverse transcription-quantitative PCR. The RNA levels of the analysed genes are expressed relative to the rpoD expression level. The experiments were performed with three biological replicates and two technical replicates. Bars represent maximum and minimum values of three biological replicates. Crosses represent average values of three biological replicates. Asterisks indicate values that are significantly different from those following inoculation with strain OE1-1 (p < 0.05, t test).

| Cellulose degradation activity
Plate assays for extracellular cellulase were conducted as described by Tsujimoto et al. (2008). In brief, bacterial strains were grown in PY medium (Kanda et al., 2003) until an OD 600 of 1.0 was reached, and after lysis, 1 mL of lyzed cell solution was centrifuged at 15,000 × g for 5 min to remove cell debris.

| Differential gene expression analysis
Statistical analysis of the RNA-seq data was performed in the R environment. Genes with zero counts in at least one OE1-1 sample were excluded. The RNA-seq read counts of the remaining genes were normalized using the function calcNormFactors (trimmed mean of M value normalization) in the package edgeR (Robinson et al., 2010). To extract genes with significant expression changes, the following thresholds were applied: q < 0.05 and log 2 (fold change) ≥ |2|. The false discovery rate (Storey's q value) was calculated in edgeR from Benjamini-Hochberg-corrected p values. Hierarchical clustering of all normalized mean expression values based on their relative expression (counts per million) was performed using Cluster 3.0 software (de Hoon et al., 2004). The average value of three replicates per strain was used. Heatmaps were created in TreeView (Eisen et al., 1998). For GO enrichment analysis, GO terms were obtained from QuickGO (https://www. ebi.ac.uk/Quick GO/), and the analysis was performed with the R package GoSeq (Young et al., 2010).

| QS-dependent phenotypes
We examined the in vitro biofilm formation of R. pseudosolanacearum strains grown without shaking in quarter-strength M63 as previously described (Mori et al., 2016). Biofilm formation was quantified based on the absorbance at 550 nm. The resulting value was normalized according to the number of cells (OD 600 ). Three biological replicates were included, with five technical replicates per biological replicate. Means were analysed for significant differences between R. pseudosolanacearum strains by ANOVA followed by Tukey-Kramer's HSD test.
EPS I production by R. pseudosolanacearum cells grown on quarter-strength M63 solidified with 1.5% agar was quantitatively analysed by ELISA (Agdia) as previously described (Mori et al., 2016

| RT-qPCR-based quantification of expression levels of QS-related genes in R. pseudosolanacearum strains in infected tomato plants
First, 8-week-old tomato (S. lycopersicum 'Ohgata-Fukuju') plants were inoculated with R. pseudosolanacearum strains (10 8 cfu/mL) using a root-dip inoculation procedure as previously described . Three days after inoculation, total RNA was isolated from tomato roots with RNAiso (Takara), and RNA samples were treated with DNase I (RNase-free; Takara). Reverse transcription was carried out with 1 μg total RNA using a PrimeScript RT reagent kit (Takara). qPCR amplifications were carried out in 20-μL reaction mixtures containing 1 μL of the cDNA stock and 0.4 μL of the appropriate primers (10 pM; Table S4) using the SYBR GreenER qPCR reagent system (Invitrogen) on an Applied Biosystems 7300 real-time PCR system. All values were normalized against the expression level of rpoD, which was used as an internal standard for each cDNA sample. No significant differences in rpoD expression levels were observed among R. pseudosolanacearum strains. The experiments were performed with three biological replicates and two technical replicates. Means of the assays were analysed for significant differences between R. pseudosolanacearum strains using Student's t test in Microsoft Excel.