Bacterial strains, plasmids and primers used in this study are summarized in Supplementary Tables S1 – S3 , respectively. P. savastanoi strains were grown at 28 °C using lysogeny broth (LB) (Lennox, 1955) without glucose and containing 0.5 % NaCl, King’s B (KB) (King et al., 1954), SSM (Meyer and Abdallah, 1978), super optimal broth (SOB) (Hanahan, 1983) or HIM (Huynh et al., 1989) culture media. Escherichia coli strains were grown using LB at 37 °C. When required, media were supplemented with appropriate antibiotics at the following final concentrations. For P. savastanoi: ampicillin (Ap) (400 μg/mL), gentamicin (Gm) (10 μg/mL), kanamycin (Km) (7 μg/mL), nitrofurantoin (25 μg/mL), and cycloheximide (100 μg/mL). For E. coli: Ap (100 μg/mL), Gm (10 μg/mL) and Km (50 μg/mL).

Gene gacA was removed from Psv NCPPB 3335 using plasmid pgacA-Km, containing the DNA immediately flanking gene gacA in both sides (approximately 1.2 kb on each side) separated by an nptII (Km)-resistance gene ( Supplementary Tables S2 , S6 ), and following previous procedures (Pérez-Martínez et al., 2007; Matas et al., 2014). Afterwards, the Km gene was removed using the pFLP2 plasmid ( Supplementary Table S2 ).

For construction of the complemented strains, the complete coding sequences of gacA and uvrC, together with their predicted ribosomal binding sites, were amplified from Psv NCPPB 3335 by PCR. After verifying the correct sequence of the resulting amplicons, those fragments were individually cloned under the control of P_lac in pBBR1MCS-5 (Gm^R) and pBBR1MCS-2 (Km^R), respectively ( Supplementary Table S2 ).

Promoters were predicted with BPROM (Solovyev and Salamov, 2011), a bacterial sigma 70 promoter recognition program (http://www.softberry.com/).

5’ Rapid amplification of complementary DNA (cDNA) ends (RACE) was used to establish the transcription start point and genomic boundary of uvrC in Psv NCPPB 3335. Transcripts were obtained from bacterial cultures grown in HIM after 6 hours of induction. 5´ RACE strategy was performed using the 5’/3’ RACE Kit 2^nd Generation (Roche Applied Science, Mannheim, Germany). According to the manufacturer’s instructions, three different primers at 357 (SP1), 214 (SP2) and 142 (SP3) base pairs (bp) distance from the ATG of uvrC were used for amplification ( Supplementary Table S3 ).

To purify RNA, pre-inocula of 20 mL of NCPPB 3335 and Psv-ΔgacA were grown overnight in both KB and SSM media at 28 °C. Bacterial cells from each medium were diluted in fresh KB or SSM to a final optical density at 600 nm (OD₆₀₀) of 0.1, to prepare three biological replicates of 110 mL for each medium and grown at 28 °C with gentle shaking to a final OD₆₀₀ of 0.5 (approximately 5x10⁷ CFUs/mL). Next, bacterial cells from these SSM cultures were collected, each biological replicate resuspended in 12 mL of fresh SSM and divided in twelve 1 mL samples that were frozen in liquid nitrogen and stored at -80 °C. Liquid cultures from each biological replicate grown on KB were pelleted, washed twice with one volume of sterile 10 mM MgCl2, and each resuspended in 110 mL of HIM and incubated at 28 °C for 6 hours. After this, bacterial cells were collected, and each biological replicate resuspended in 12 mL of HIM and divided in twelve 1 mL samples. After a centrifugation step, the pellets were flash-frozen in liquid nitrogen and stored at -80 °C. RNA extractions were performed with the RNAeasy Mini Kit (Qiagen; Hilden, Germany) and following the manufacturer’s instructions. Subsequently, the RNA was cleaned up from traces of genomic DNA using the TURBO DNA-free kit (Invitrogen Corp, CA, USA) following the manufacturer’s instructions. For ribosomal RNA degradation, the Illumina Ribo-Zero Plus rRNA Depletion kit Bacteria (Illumina; CA, USA) was used. The resulting RNA was quantified by spectrophotometry, its integrity assessed by agarose gel electrophoresis, and its quality evaluated on an Agilent Bioanalyzer 2100 using a Pico 6000 RNA bioanalyzer chip (Agilent Technologies, Santa Clara, CA, USA). For genotyping and sequencing, the two biological replicates for each medium with the best quality of purified RNA were used. The TruSeq Stranded mRNA kit (Illumina; CA, USA) was used for genotyping and the Illumina NextSeq550 kit (Illumina; CA, USA) was used for sequencing.

Reads obtained from the RNA-Seq analysis were processed in collaboration with the Ultra sequencing Service of the University of Málaga and the Andalusian Bioinformatics Platform, using a workflow of previously described software packages (Moreno-Pérez et al., 2021). SeqTrimNext (v.2.0.60) was used to clean up and process the sequences, and the FastQC software for quality control; then, sequences were aligned to the reference whole chromosome genome (accession no. CP008742.1) and the three native plasmids of Psv NCPPB 3335 (pPsv48A, FR820585.2; pPsv48B, FR820586.1; pPsv48C, FR820587.2) with the Bowtie2 tool (Langmead and Salzberg, 2012). Transcript abundance, measured in fragments per kilobase per million mapped reads (FPKM), and differential gene expression were calculated using Tuxedo together with the Cufflinks suit of tools (Trapnell et al., 2010; Ghosh and Chan, 2016). CuffDiff was used to identify differentially expressed genes (DEGs). Only those genes with an adjusted P-value (q-value) < 0.05 and a log₂ fold change ≤ -0.5 or ≥ 0.5 were considered significant. Graphical representation of the differential expression results was carried out using the CummeRbund package in R (Goff et al., 2012).

For real-time quantitative PCR (RT-qPCR), DNA-free total RNA samples described in the previous section were used. cDNA synthesis was carried out using random hexamers included in the iScriptTM cDNA synthesis kit (BioRAD, CA, USA) and 1 μg of DNA-free total RNA as template. RT-qPCR primers were designed with Primer3Plus (Untergasser et al., 2007) and Bacon Designer Free (Thornton and Basu, 2015) ( Supplementary Table S3 ). Efficiency curve analysis to confirm the amplification specificity of primer pairs was carried out as described (Vargas et al., 2011). The relative transcript abundance was calculated using the ΔΔ cycle-threshold (Ct) method (Livak and Schmittgen, 2001). Transcriptional data was then normalized according to the expression of the housekeeping gene gyrA by calculating the difference between the relative expression of the analysed gene and that of the gyrA gene (ΔΔCt = Ct studied gene – Ct gyrA). Fold change values were calculated as 2^-ΔΔCt (Pfaffl, 2004; Rotenberg et al., 2006). RT-qPCR analyses were carried out in triplicate with three biological replicates. Change rate is represented by the Psv-ΔgacA expression divided by the wild-type strain expression.

Identification and comparison of homologous rsm genes in Psv NCPPB 3335 was carried out by blastp and blastn, using the Geneious 8.1.9 program (Kearse et al., 2012), and the previously described rsm sequences in 1448A (Ramírez-Zapata et al., 2020), and Pto DC3000 (Chatterjee et al., 2003; Ferreiro et al., 2018). Sequence alignments were performed using Multalin (http://multalin.toulouse.inra.fr/multalin/) (Corpet, 1988) and Needle servers (https://www.ebi.ac.uk/Tools/psa/emboss_needle/) (Madeira et al., 2022). Additionally, the different rsm transcripts were identified using the IGV program by aligning the original.fna sequence with the.gtf and bowtie.bam files extracted from the RNA-Seq analysis.

For transmission electron microscopy (TEM) analysis, Psv strains were grown at 28 °C for 48 h on LB. Then, bacteria were collected, washed three times with 10 mM MgCl₂, and resuspended in SSM medium. After incubation at 28 °C overnight, bacterial cultures were resuspended in SSM to an OD₆₀₀ of 0.1 and incubated with shaking at 28 °C until they reached an OD₆₀₀ of 0.6. Grids were deposited over 30 μL of bacterial suspensions and incubated for 1 h at room temperature. Subsequently, grids were treated with 4 % paraformaldehyde for 10 minutes, washed with water for 5 minutes, negatively stained with 1 % uranyl acetate for 30 seconds and washed once with water for 30 seconds. The grids were allowed to dry for 24 h and imaged on a FEI Tecnai G2 20 TWIN TEM at an accelerating voltage of 80 kV. Images were acquired using TIA FEI Imaging Software v.4.14 (Berlanga-Clavero et al., 2022).

Psv strains were grown at 28 °C for 48 h on KB in the dark. Cells were resuspended in KB to an OD₆₀₀ of 1 (approximately 1x10⁹ CFUs/mL). Two microliters of the bacterial suspension were spotted onto soft KB agar (KB with 0.3 % agar) and plates were incubated for 72 h at 25 °C and 60 % relative humidity in the dark (Antúnez-Lamas et al., 2009). The surface area of swimming colonies was photographed and quantified using the area selection tool of the image software FiJi (Schindelin et al., 2012).

Bacterial suspensions were prepared to an OD₆₀₀ of 1 (approximately 1x10⁹ CFUs/mL) and 100 μL plated on SSM using a Drigalski plating loop. Thirteen-millimetre diameter Whatman paper discs were wetted with 100 μL of 1.5 % H₂O₂ and allowed to dry for 15 minutes. The discs were placed in the centre of each plate and incubated at 28 °C for 48 h before measuring the diameter of the growth inhibition haloes.

To determine the intrinsic oxidative stress of the cells, the Psv strains were grown overnight in 20 mL of LB at 28 °C. Bacterial cells were diluted 1:10 and disaggregated using a 23-gauge needle. Samples of 100 μL each were incubated with 0.2 μL of 2.5 μM CellROX™ Green Reagent (Invitrogen, Corp, CA, USA) for 30 minutes at 37 °C with shaking. Cells were washed three times with Phosphate Buffered Saline (PBS) and then thoroughly resuspended by passing through a needle (Cámara-Almirón et al., 2020). Five microliters of each sample were dropped on a slide and observed using a Leica SP5 confocal microscope equipped with a HCX PL APO lambda blue 63.0x1.40 OIL UV objective. Images were processed using the Leica Application Suite Advance Fluorescence v.2.7.3.9723 (LCS Lite, Leica Microsystems) and the FIJI/ImageJ software. The laser settings, scanning speed, photomultiplier detector gain, and pinhole aperture were constant for the five images per strain acquired for each of the three experiments performed, and were as described (Berlanga-Clavero et al., 2022).

Two-month-old Nicotiana tabacum var. Newdel plants were used for the HR assays. Plants were grown with a photoperiod of 16 h of light, 8 h of darkness at 26 °C. Bacterial suspensions were prepared in 10 mM MgCl₂ with low population density, 1x10⁶ CFUs/mL. Then, a small wound was made on the abaxial side of the leaf with a hypodermic needle and the bacterial suspension was introduced into the periplasmic space using a 1 mL blunt syringe. The development of symptoms was observed 2 days and 3 days post inoculation (dpi). Images were taken with a high-resolution digital camera (Nikon DXM 1200, Nikon Corp).

Pathogenicity was tested on Olea europaea plants, derived from seeds of cv. Arbequina germinated in vitro, grown at 60 % humidity and 26 °C. For inoculation, bacterial suspensions were prepared in 10 mM MgCl₂ to a final OD₆₀₀ of 0.1 (approximately 1x10⁷ CFUs/mL) and two wounds made in the stem of the plant with a scalpel were each inoculated with 5 μL. Four plants were used per strain in each of the three replicate assays. During the 54-day trial, the volume of tumours in each of the inoculated wounds was measured every week. At the end of the trial, symptoms were photographed with a high-resolution digital camera (Nikon DXM 1200). Bacteria were recovered from tumours by grinding them using a mortar and pestle containing 1 mL of sterile 10 mM MgCl₂. Serial dilutions were plated onto LB supplemented with nitrofurantoin, cycloheximide and the corresponding antibiotic, as described (Penyalver et al., 2006; Pérez-Martínez et al., 2007; Matas et al., 2012).

Competitive index (CI) assays were performed on in vitro olive plants obtained from a germinated seed of O. europaea cv. Arbequina and grown on DKW medium (Driver and Kuniyuki, 1984) supplemented differently in each of the three culture periods (proliferation, channelling, and maintenance). Plants were inoculated with bacterial suspensions in 10 mM MgCl₂ in a 1:1 ratio (wild-type strain: mutant strain). A single wound per plant was inoculated with approximately 5x10³ CFUs of each strain, using six plants per experiment and repeating the experiment three times (a total of 18 inoculated wounds). After 30 dpi under 25 °C, 50-60 % humidity and 16 h light photoperiod, symptoms were visualized using a stereo microscope (Leica MZ FLIII; Leica Microsystems, Wetzlar, Germany). Bacteria were recovered from tumours as above, and serial dilutions were seeded on LB and LB+Km for colony counting after 2 days at 28 °C. The number of CFUs of the mutant strain was obtained from the colonies that grew on the Km plates, while the number of wild-type bacteria was obtained by subtracting the number of mutant bacteria from the total number of bacteria grown on LB plates (Macho et al., 2007; Macho et al., 2016). The CI is defined as the mutant to wild-type ratio in the output sample divided by the mutant to wild-type ratio in the input (inoculum) sample (Freter et al., 1981; Taylor et al., 1987), the input ratio being close to 1.

Statistical analyses were performed using GraphPad Prism version 9. p-values < 0.05 were considered significantly different. Asterisks indicate the level of statistical significance: *p < 0.05, **p < 0.01, ***p < 0.001 and ****p < 0.0001. When needed, data were normalized to the corresponding internal standard of the samples. All experiments were repeated at least three independent times, with similar results.

Pseudomonas species contain two to seven Rsm gene homologues in their genome, with different levels of sequence identity (Sobrero and Valverde, 2020). Genomic analysis of Psv NCPPB 3335 revealed four Rsm homologues (RsmA, RsmC, RsmE and RsmH1) and seven different sRNA genes (rsmX1, rsmX2, rsmX3, rsmX4, rsmX5, rsmY and rsmZ) ( Table 1 ). The Rsm deduced products and the sRNA genes showed a high sequence identity with their respective homologues in Pto DC3000 (Ferreiro et al., 2018) and Pph 1448A (Ramírez-Zapata et al., 2020) ( Table 1 ). Therefore, Psv NCPPB 3335 contains a similar, but distinct complement of Rsm homologues and sRNA genes to other P. syringae strains, which should allow for the full functionality of the GacS/GacA regulatory system.

Total RNA was purified from cultures of Psv NCPPB 3335 and its ΔgacA mutant grown in SSM, a minimal medium, and HIM medium, which simulates conditions within the plant apoplast and causes the induction of the hrp genes (Huynh et al., 1989). Following previous analyses of the role of GacA in P. syringae (Chatterjee et al., 2003; Marutani et al., 2008; Ortiz-Martín et al., 2010; Ferreiro et al., 2018), cells were collected during the log phase (OD₆₀₀ of 0.5) in order to better capture the early induction of virulence genes. To compare transcripts abundance, two biological replicates per strain and media were subjected to RNA Illumina sequencing.

In total, 143.2 and 131.0 million high-quality reads (clean reads) were obtained for the SSM and HIM samples, respectively ( Supplementary Tables S4 , S5 ). By iterative sequence alignment, an average of 99.40 % (SSM) and 99.44 % (HIM) of clean reads were mapped to the Psv NCPPB 3335 genome. Out of the 5508 genes annotated in the GenBank record of the Psv NCPPB 3335 chromosome, a total of 5472 genes (99.35 % of the total) were covered by Illumina sequencing in SSM for strain Psv-ΔgacA ( Supplementary Table S6 ). In HIM medium, Illumina sequencing covered 99.44 % (5477 genes) and 99.44 % (5476 genes) of the genes in strains NCPPB 3335 and Psv-ΔgacA, respectively ( Supplementary Table S6 ). Furthermore, in both media and strains, the Illumina RNA sequencing also covered the 68, 53 and 51 genes annotated in the GenBank records of plasmids pPsv48A, pPsv48B and pPsv48C, respectively ( Supplementary Table S6 ).

O’Malley and Anderson (2021) showed that deletion of the gacA gene in Pto DC3000 did not affect the expression of the downstream genes uvrC and psgA, which is relevant because of the regulatory activity associated to gene uvrC (Humann et al., 2009; O’Malley et al., 2019). This is in contrast with the significant downregulation of uvrC observed in our RNA-Seq analysis of strain Psv-ΔgacA ( Table 2 ). We therefore sought to analyse the transcriptional organization of these genes in strain NCPPB 3335.

An RT-qPCR analysis of cells grown in SSM and HIM media showed a significantly reduced expression of uvrC in strain Psv-ΔgacA, in agreement with the RNA-Seq data, whereas the expression of gene psgA was comparable to that of the wild-type strain ( Figure 1A ). Using BPROM, we identified several -35 and -10 regulatory elements typical of Pribnow-type promoters in the gacA DNA region ( Figure 1B ). A 5´ RACE assay with strain NCPPB 3335 showed that the uvrC transcription start site (+1) was located 193 bp upstream from its predicted start codon ( Supplementary Figure S1A ). This result confirms that the coding sequence of gacA includes a functional promoter for uvrC, but upstream of the -35 and -10 regions predicted bioinformatically. However, in strain Psv-ΔgacA we did not observe a complete absence of transcripts covering gene uvrC ( Table 2 , Supplementary Figure S1B ), likely suggesting a low level of uvrC transcription from the promoter upstream of gene gacA ( Figure 1B ).

Nevertheless, the altered transcription of gene uvrC that is associated to certain mutations in gacA was shown to confound the assessment of the actual regulatory role of GacA (O’Malley and Anderson, 2021). Therefore, and considering the irregular expression of uvrC observed in strain Psv-ΔgacA, we decided to construct derivatives of this strain complemented separately with genes uvrC (Psv::uvrC) and gacA (Psv::gacA) in order to perform the following phenotypical assays.

Distribution of the normalized FPKM values did not show significant differences between the biological replicates for any combination of strain and medium ( Supplementary Figure S2 ), suggesting that no technical bias was introduced during library construction or sequencing. The normalized expression levels of wild type and mutant strains were compared to detect DEGs, which were selected considering a fold change of ± 0.5 and a statistical value of q < 0.05 ( Figures 2A, B ). Strain Psv-ΔgacA showed a total of 1020 (SSM) and 1239 (HIM) DEGs ( Figures 2C, D ), with 654 genes downregulated and 366 genes upregulated in SSM, and 623 genes downregulated, and 616 genes upregulated in HIM medium.

As could be expected, the growing conditions determined specific changes in transcript abundance ( Figure 2 ). Thus, genes upregulated in Psv-ΔgacA in SSM medium were mainly related to bacterial motility ( Figure 2E ), but were classified in the functional categories of pathogenesis, host response and programmed cell death in HIM medium, which favours expression of virulence-related genes under the assayed conditions ( Figure 2F ). Likewise, most genes downregulated in SSM were related to the metabolism of carbohydrates ( Figure 2E ), whereas in HIM they belonged to the categories of metabolism of nitrogen and proton transport ( Figure 2F ).

In fact, there appears to be a minor overlap of the pool of genes deregulated in media SSM and HIM. On the one hand, only seven genes showed the same type of differential expression in both media ( Supplementary Figure S3 ; Supplementary Table S7 ). Among these genes is uvrC, which is located immediately downstream of gene gacA and showed the highest downregulation ( Supplementary Table S7 ). On the other hand, we also found opposing regulation for a limited number of genes. Specifically, 58 genes were upregulated in HIM but downregulated in SSM, while 42 genes were downregulated in HIM but upregulated in SSM ( Supplementary Figure S3 ). The functions that exhibited the most notable changes in expression depending on the growth medium were related to metabolic processes or locomotion dependent on flagellum activity. These 100 genes are of considerable interest because they may represent genes that are expressed in free-living conditions (SSM) but are repressed during the interaction with the plant (HIM) and vice versa.

Previous studies reported the involvement of the GacS/GacA system in the motility of Pto DC3000 (Chatterjee et al., 2003; Ferreiro et al., 2018; Ferreiro et al., 2021). The 42 genes for flagellar biosynthesis in diverse Pseudomonas, including NCPPB 3335, are grouped in nine different operons plus one monocistronic gene, fliK ( Figure 3A ) (McCarter, 2006). Our transcriptomic analysis in SSM medium showed that 37 out of the 42 genes were significantly upregulated in strain Psv-ΔgacA ( Figures 3A, B ). In agreement with this, genes fliC, fliE, fliM, flhA and flhF showed a significantly higher expression in strain Psv-ΔgacA than in strain NCPPB 3335 in an RT-qPCR analysis of cells cultivated in SSM ( Figure 3C ). This could suggest that the GacS/GacA system might directly or indirectly regulate motility in Psv by regulating flagellar biogenesis.

Due to the significant number of deregulated genes encoding structural proteins of the bacterial flagella, we decided to analyse the morphology and motility of strain Psv-ΔgacA. Transmission electron microscopy ( Figure 4A ) showed a characteristic rod-shaped morphology with several amphitrichous flagella for the wild-type strain, whereas no regular flagella could be seen in cells from strain Psv-ΔgacA, which showed an abnormal conical morphology at their poles ( Figure 4A ).

We then tested the swimming capacity of strain Psv-ΔgacA in comparison with the wild-type strain and the gacA and uvrC complemented strains ( Figure 4B ). After 72 h, both strain Psv-ΔgacA and the complemented strain Psv::uvrC exhibited a significantly reduced swimming halo when compared to strain NCPPB 3335 or the complemented strain Psv::gacA ( Figure 4B ). These results indicate that the overexpression of flagellar genes observed in strain Psv-ΔgacA does not correlate with an increased motility or a higher abundance of flagella in the experimental conditions analysed. Additionally, the complete absence of flagella in strain Psv-ΔgacA suggest that they are not essential for swimming motility in Psv NCPPB 3335, as occurs with other bacteria (McCarren and Brahamsha, 2009; Wadhwa and Berg, 2022).

Excessive levels of reactive oxygen species (ROS) or oxidants can cause cellular damage and disrupt cellular signalling processes. Therefore, a critical cell defence mechanism against oxidative stress involves the regulation of genes maintaining ROS homeostasis (Ray et al., 2012). In this sense, previous studies have shown that the GacS/GacA system enhances resistance to H₂O₂ and participates in the regulation of diverse genes involved in resistance to oxidative stress in biocontrol strains of Pseudomonas (Heeb et al., 2005; Hassan et al., 2010; Kim et al., 2014).

The RNA-Seq analysis showed a differential expression of several genes involved in the oxidative stress response, although the expression patterns were dependent on the growth medium ( Supplementary Table S8 ). In cells of strain Psv-ΔgacA incubated in HIM, all the genes encoding the different subunits of the ATP synthase complex, as well as five out of 13 genes encoding the NADH oxidoreductase complex, were significantly downregulated ( Supplementary Table S8 ). In contrast, 19 genes encoding catalases, peroxidases, aquaporins and reductases showed significantly higher expression levels, likely because of a differential expression of transcriptional regulator encoding genes ( Supplementary Table S8 ). This pattern of regulation is essentially reversed in the ΔgacA mutant grown in SSM medium, where many of the genes that were downregulated in HIM medium were either upregulated or did not show a differential expression.

To evaluate the role of GacA on the response of Psv to an external oxidative stress, we measured growth inhibition around a filter paper disc soaked in H₂O₂. After 48 hours, both strains Psv-ΔgacA and Psv::uvrC showed a reduced growth inhibition halo compared to strain NCPPB 3335 ( Figure 5A ). Notably, this phenotype was reversed in strain Psv::gacA, complemented with the gacA gene. Additionally, the oxidative stress state of the different mutants and the wild-type strain was measured using the cell permeable dye provided with the CellROX™ Green Reagent (Thermofisher). This dye penetrates through the plasma membrane and remains in a non-fluorescent version in a reduced state but emits fluorescence upon oxidation by ROS. The emission of green fluorescence detected by confocal microscopy was significantly reduced for both strains Psv-ΔgacA and Psv::uvrC, suggesting a lower abundance of ROS in their cells when compared with strains NCPPB 3335 or Psv::gacA ( Figures 5B, C ). These results would indicate that the activity of GacA not only contributes to a decreased tolerance to H₂O₂, but also to a reduction in the amount of intracellular ROS in free-living bacterial cells not exposed to oxidative stresses.

The T3SS is one of the most important pathogenicity factors in plant pathogenic bacteria, including Psv NCPBB 3335 (Pérez-Martínez et al., 2010), facilitating the delivery of T3Es into the host cell cytoplasm (Hueck, 1998; Göhre and Robatzek, 2008). However, the influence of the GacS/GacA pathway over the T3SS can vary among different phytopathogenic Pseudomonas strains (Valentini et al., 2018; O’Malley et al., 2019). Our RNA-Seq data in HIM, revealed that 25 out of the 29 genes from the hrp/hrc cluster were significantly upregulated in strain Psv-ΔgacA compared to the wild-type strain ( Figure 6A ). Additionally, nearly two thirds of the T3E genes were also upregulated in the ΔgacA mutant exclusively in HIM medium ( Supplementary Table S9 ), suggesting that GacA also participates, directly or indirectly, in ensuring the correct level of transcription of effector genes during pathogenesis. Furthermore, RT-qPCR analyses showed that the expression levels of two T3SS regulatory genes (hrpA and hrpL) and two T3Es genes (hopAZ1 and avrRpm2) were significantly increased in Psv-ΔgacA compared to the wild-type strain ( Figure 6B ).

Since genes for the biosynthesis and delivery of T3Es are overexpressed in strain Psv-ΔgacA, we evaluated its ability to induce an HR in tobacco leaves. In line with this overexpression, after 48 h of infiltration, strain Psv-ΔgacA showed a more intense HR than that observed for the wild-type strain or the gacA complemented strain ( Figure 7 ). The Psv-ΔgacA strain complemented with gene uvrC consistently displayed an intermediate phenotype ( Figure 7 ), which was difficult to interpret. It is possible that this is the consequence of a partial complementation in our experimental conditions, as occurs with other systems (e.g., Añorga et al., 2020; Ramírez-Zapata et al., 2020), or a possible contribution of uvrC to the regulation of the secretion process.

Besides the genes presented above, several other genes involved in virulence were overexpressed in the RNA-Seq analysis of strain Psv-ΔgacA ( Supplementary Table S10 ). Among these, five genes located within the ant and cat operons of the WHOP region (Caballo-Ponce et al., 2017b), a genomic island involved in the degradation of aromatic compounds, were significantly upregulated in HIM medium. Additionally, genes involved in the biosynthesis of the phytohormones indole-3-acetic acid (iaaH1 and iaaM1) and cytokinins (ptz) were also significantly upregulated in SSM medium.

To assess the possible impact on virulence of the lack of gene gacA, we carried out virulence assays in olive seedlings under controlled conditions. In these conditions, we measured virulence on the basis of tumour volume and total population within the tumours, as has been done previously (Rodríguez-Moreno et al., 2008; Añorga et al., 2020; Moreno-Pérez et al., 2021). After 54 dpi, tumours induced by strain Psv-ΔgacA and the complemented strain Psv::uvrC were noticeably bigger than tumours induced by strains NCPPB 3335 and Psv::gacA ( Figure 8A ). As expected, the average tumour volume was significantly higher for strain Psv-ΔgacA and its uvrC-complemented derivative ( Figure 8B ), while they reached significantly higher cell densities within the tumours than the wild-type and the gacA-complemented strains ( Figure 8C ). Nevertheless, relative bacterial populations were not significantly different for strains NCPPB 3335 and Psv-gacA, reaching 6.1x10⁵ and 7.4x10⁵ CFU per gram of tumour tissue, respectively. Therefore, the higher virulence of the gacA mutant resulted in larger tumours that, because of a higher volume, supported larger total bacterial populations. However, these tumours are not representative of typical disease symptoms because tumours induced by the wild-type strain showed a more compact and necrotized tissue, whereas those induced by strain Psv-ΔgacA had a softer and more greenish aspect and showed significant parts of the tissue occupied by aggregations of dedifferentiated cells ( Figure 8D ).

To investigate competitiveness, micropropagated olive plants were inoculated with a total of 5x10³ CFUs of a 1:1 ratio mixture of the wild-type strain and either 1) strain Psv-ΔgacA, 2) strain Psv-ΔgacA complemented with gene gacA (Psv::gacA), or 3) strain Psv-ΔgacA complemented with gene uvrC (Psv::uvrC). The analysis of the bacterial populations of tumours showed no significant difference in competitiveness between the wild-type strain and either one of the two complemented strains ( Figure 8E ). However, population counts of strain Psv-ΔgacA were significantly lower than those of the wild-type strain, indicating a lower competitiveness of strain Psv-ΔgacA. This indicates that genes gacA and uvrC act redundantly to increase the competitive fitness of the bacterium.

In summary, these results indicate that GacA is necessary for the development of typical disease symptoms, characterized by tumours with increased cavity formation and necrotic tissue. Conversely, the absence of this gene increases virulence, inducing significantly larger tumours with increased bacterial populations, but reduces the competitive advantage of the strain in planta.