Exploring the expression and functionality of the rsm sRNAs in Pseudomonas syringae pv. tomato DC3000

ABSTRACT

The Gac-rsm pathway is a global regulatory network that governs mayor lifestyle and metabolic changes in gamma-proteobacteria. In a previous study, we uncovered the role of CsrA proteins promoting growth and repressing motility, alginate production and virulence in the model phytopathogen Pseudomonas syringae pv. tomato (Pto) DC3000. Here, we focus on the expression and regulation of the rsm regulatory sRNAs, since Pto DC3000 exceptionally has seven variants (rsmX1-5, rsmY and rsmZ). The presented results offer further insights into the functioning of the complex Gac-rsm pathway and the interplay among its components. Overall, rsm expressions reach maximum levels at high cell densities, are unaffected by surface detection, and require GacA for full expression. The rsm levels of expression and GacA-dependence are determined by the sequences found in their −35/-10 promoter regions and GacA binding boxes, respectively. rsmX5 stands out for being the only rsm in Pto DC3000 whose high expression does not require GacA, constituting the main component of the total rsm pool in a gacA mutant. The deletion of rsmY and rsmZ had minor effects on Pto DC3000 motility and virulence phenotypes, indicating that rsmX1-5 can functionally replace them. On the other hand, rsmY or rsmZ overexpression in a gacA mutant did not revert its phenotype. Additionally, a negative feedback regulatory loop in which the CsrA3 protein promotes its own titration by increasing the levels of several rsm RNAs in a GacA-dependent manner has been disclosed as part of this work.

Introduction

Bacterial survival depends on their ability to sense environmental changes and react to them consequently. Global regulatory pathways play a pivotal role in their adaptation because they allow the detection of multiple external stimuli that elicit a coordinated genetic response. In the context of transmembrane signalling, two-component systems (TCS) stand out for their abundance and relevance [1]. A prototypical TCS consists of a membrane-located histidine kinase (HK), which senses the stimulus, and a cytoplasmic response regulator (RR), which mediates the response. After the detection of the signal(s), the sensor protein autophosphorylates on a specific histidine residue, and the phosphoryl group is subsequently transferred to an aspartate residue on the RR’s receiver domain, triggering a conformational change which in turn generates a cellular response, usually by stimulating (or repressing) the expression of target gene(s) [2,3]. The TCS GacS/GacA system was described for the first time in two plant-colonizing strains. HK GacS, initially called LemA (lesion manifestation), was described in Pseudomonas syringae pv. syringae B728a as an essential factor for the production of lesions in bean leaves, since the inactivation of gacS caused decrease in bacterial survival and loss of virulence in plants [4]. The RR GacA was first described in Pseudomonas protegens (formerly fluorescens) CHA0 as a global activator of antibiotic and cyanide production given that its loss decreased the antifungal activity of this strain [5]. Later, this TCS has been found in numerous γ-proteobacteria where it controls pathogenicity, plant growth-promoting activity, survival, motility, biofilm formation and production of secondary metabolites, such as antibiotics, quorum-sensing signals, toxins, extracellular polysaccharides or secreted proteins [2,4–12].

Initially, it was believed that the GacA response regulator controlled gene expression only at transcriptional level, but it was later discovered that the GacS/GacA system exerts its action mainly through the regulatory network of the secondary metabolism, rsm (regulator of secondary metabolism) [13–16]. The Gac-rsm signal transduction cascade is very well studied in P. protegens CHA0 and starts with the reception of the signal(s), hitherto unknown, by the membrane sensor GacS. It phosphorylates itself and the cytoplasmic RR GacA, which is now active and binds to the Gac-box (whose consensus is TGTAAGN₆CTTACA) located in the promoter regions of the small regulatory RNAs rsmX, rsmY and rsmZ, inducing their expression. These RNAs are recognized with high affinity by two RNA-binding proteins, RsmA and RsmE, which belong to the CsrA/RsmA family of small RNA-binding proteins that post-transcriptionally regulate protein synthesis. CsrA (carbon storage regulator) was first described in Escherichia coli, where it plays a major role in controlling the intracellular carbon flux [17]. Thereafter, CsrA homologs were found in many γ-proteobacteria, where they were sometimes called RsmA (repressor of secondary metabolism), as in Pseudomonas aeruginosa [18]. The RsmA/CsrA proteins are usually bound to the 5′ untranslated region (5ʹ-UTR) and/or early coding regions of certain mRNAs, thereby affecting their stability, turnover, transcript elongation and/or translation rates [19–23]. Initially, inhibition of translation initiation was believed their only mode of action, but new ways of repression, and even promotion of transcription and translation, were lately reported [22,24].

The rsm RNAs antagonize RsmA/CsrA binding to mRNAs in a competitive manner, since they contain multiple binding sites that sequester and store RsmA/CsrA. The rsm RNAs contain several stem-loops with the GGA trinucleotide exposed in the loops of hairpins, to which the RsmA/CsrA proteins bind with high affinity [19,20,25,26]. The distinct structures of the diverse rsm species confer different RsmA/CsrA binding properties and affinities [25,27–29]. The expression of the rsm RNAs varies according to culture conditions and increases along the growth curve in different Pseudomonas strains, reaching their maximum in late exponential or stationary phase [28,30–35]. The conservation of the Gac-box sequence determines the GacA-dependence for each rsm RNA [28,36–38]. Other regulators also influence rsm levels, like P. protegens CHA0 PsrA and AlgR, which promote rsmZ transcription [30,39], or P. aeruginosa HptB and MvaT that have a negative effect on rsmZ levels [14,32]. Moreover, the RsmA/CsrA proteins exert a positive action on the transcription of the rsm RNAs, generating a negative feedback regulatory loop [28,30]. This phenomenon has been reported not only in Pseudomonas but also in E. coli, Salmonella enterica and Vibrio cholerae [40–45].

Pseudomonas syringae pv. tomato (Pto) DC3000 is a phytopathogenic bacteria that infects tomato, crucifers and the model plant Arabidopsis thaliana, therefore it has been widely used for the study of bacteria-plant interactions [46,47]. Its virulence relies on a large repertoire of effectors that are secreted through a type III secretion system (T3SS) and the phytotoxin coronatine, which disrupts signalling mediated by jasmonic acid and stimulates stomatal opening, allowing the bacterial entry into the apoplast [48–50]. In addition, Pto possesses other tools that contribute to pathogenicity, as flagella and biosurfactants, which facilitate its movement, or exopolysaccharides, that prevent desiccation. Pto DC3000 produces several polar flagella [51,52] and six lipopeptides with biosurfactant activity: the syringafactins A-F, which are synthesized by a nonribosomal peptide synthetase (NRPS) encoded by two genes, syfA and syfB [53]. Pto DC3000 is able to swim in liquid or viscous media, requiring functional flagella but not syringafactin; in contrast, both flagella and syringafactin are needed for swarming, a rapid and coordinated movement of the bacterial population over solid or semi-solid surfaces [53–56].

By using a Tn5::gacA mutant of Pto DC3000 (AC811), it was shown that the GacS/GacA TCS activated the transcription of several small regulatory RNAs and genes encoding alternative sigma factors controlling carbon metabolism, virulence, motility, production of secondary metabolites, and quorum sensing [52,57,58]. However, the AC811 strain was also defective in uvrC and anmK, which were responsible for the diminished virulence of this mutant in Arabidopsis. The construction of a ΔgacA mutant (JA257) has shown that GacA is required for Pto DC3000 colonization of Arabidopsis leaves but not for virulence in the leaf apoplast [59,60].

The Pto DC3000 Gac-rsm pathway is particularly complex since its genome encodes five CsrA proteins and seven rsm RNAs [8,29,58,61,62]. In a previous study, we uncovered the role of CsrA proteins promoting growth and repressing motility, alginate production and virulence in Pto DC3000 [58], here we will focus on the rsm RNAs. The genes encoding rsm RNAs (rsmX1, rsmX2, rsmX3, rsmX4, rsmX5, rsmY and rsmZ) are spread across the genome, except for rsmX3 and rsmX4, which are located in tandem. The alignment of the rsm promoter regions revealed that the seven sequences present a motif similar to the Gac-box [16,29,61,62], and it has been shown that rsmY and partially rsmZ are controlled by GacA [52,57]. The RNAs have a small size (112 to 132 nt) and the most striking difference with other Pseudomonas is the presence of five rsmX variants (112–120 nt) with significant resemblance at both the nucleotide and structural level [62]. rsmY and rsmZ are the largest rsm RNAs (126 and 132 nt, respectively) and are very similar to those present in other strains. These two RNAs have a greater number of GGA motifs in their sequences (8) than the rsmX (5–7); however, the number of GGA motifs in loops, which would bind CsrA with high affinity, is more similar: 4 for all the RNAs, except for rsmZ that has 5 (Fig. S1). Moreover, it has been shown that rsmX1, rsmX5, rsmY, and rsmZ exhibit similar binding affinity to several CsrA proteins whereas CsrA1, CsrA2, CsrA3 and CsrA4 proteins display distinct binding affinities to each of those RNAs [63]. The aim of this work was to elucidate the physiological meaning of the coexistence of the seven rsm variants in Pto DC3000 by comparing their expression levels according to different environmental conditions and to study the impact of rsmY and rsmZ mutation and overexpression.

Material and methods

Bacterial strains, plasmids, and growth conditions

The bacterial strains used are listed in Table 1. Pto DC3000 strains were grown at 20°C in Luria-Bertani (LB) medium [64], in MMR (7 mM Na-glutamate, 55 mM mannitol, 1.31 mM K₂HPO₄, 2.2 mM KH₂PO₄, 0.61 mM MgSO₄, 0.34 mM CaCl₂, 0.022 mM FeCl₃, 0.85 mM NaCl) minimal medium [65], or in the T3SS-inducing minimal medium MMF (50 mM K-phosphate buffer pH 5.7, 7.6 mM (NH₄)₂SO₄, 1.7 mM MgCl₂, 1.7 mM NaCl, 10 mM fructose) [66]. When required, antibiotics were added to the cultures/plates: gentamicin (5 µg/ml), kanamycin (50 µg/ml), rifampicin (10 µg/ml), streptomycin (30 µg/ml), and tetracycline (10 µg/ml).

Table 1.

Bacterial strains and plasmids used

Strain or Plasmid	Relevant characteristics	Reference
Strains
P. syringae pv. tomato
DC3000	Wild type; RifR	[93–94]
csrA3	ΔcsrA3; RifR	[58]
gacA (AC811)	gacA::mini-Tn5Km; RifR KmR	[57]
gacA (JA257)	ΔgacA	[59]
psrA (AC820)	psrA:: EZ-Tn5(KAN-2); RifR KmR	[95]
ΔhrcQ-U (CUCPB5113)	ΔhrcQbRSTU::Sp; RifR SmR/SpR	[96]
rsmY	ΔrsmY; RifR	This work
rsmZ	ΔrsmZ; RifR	This work
rsmYZ	ΔrsmY ΔrsmZ; RifR	This work
Plasmids
pBBR1-MSC5	GmR; cloning vector	[97]
pBBR1-MCS5::csrA3	GmR; pBBR1-MSC5 with a 1381 pb fragment containing the csrA3 gene flanked by EcoRI restriction sites	[58]
pK18mobsacB	KmR; suicide vector	[70]
pK18ΔrsmY	KmR; pK18mobsacB with a 1410 pb EcoRI fragment bearing a deleted version of rsmY.	This work
pK18ΔrsmZ	KmR; pK18mobsacB with a 1369 pb EcoRI fragment bearing a deleted version of rsmZ.	This work
pME6016	TcR; cloning vector for construction of transcriptional lacZ fusions	[67]
pME6016::PrsmX1	TcR; pME6016 with a 176 pb fragment containing the rsmX1 promoter flanked by the EcoRI and PstI restriction sites. Transcriptional fusion.	This work
pME6016::PrsmX2	TcR; pME6016 with a 231 pb fragment containing the rsmX2 promoter flanked by the EcoRI and PstI restriction sites. Transcriptional fusion.	This work
pME6016::PrsmX3	TcR; pME6016 with a 188 pb fragment containing the rsmX3 promoter flanked by the EcoRI and PstI restriction sites. Transcriptional fusion.	This work
pME6016::PrsmX4	TcR; pME6016 with a 184 pb fragment containing the rsmX4 promoter flanked by the EcoRI and PstI restriction sites. Transcriptional fusion.	This work
pME6016::PrsmX5	TcR; pME6016 with a 237 pb fragment containing the rsmX5 promoter flanked by the EcoRI and PstI restriction sites. Transcriptional fusion.	This work
pME6016::PrsmY	TcR; pME6016 with a 256 pb fragment containing the rsmY promoter flanked by the EcoRI and PstI restriction sites. Transcriptional fusion.	This work
pME6016::PrsmZ	TcR; pME6016 with a 361 pb fragment containing the rsmZ promoter flanked by the EcoRI and BamHI restriction sites. Transcriptional fusion.	This work
pSRK	KmR; expression vector with lac promoter	[68]
pSRK5	GmR; pBBR1-MCS5 with a 1.293 pb EcoRI-SpeI fragment from pSRK_C bearing lacI and the lac promoter	This work
pSRK5::rsmY	GmR; 130 pb BglII-SpeI fragment containing the rsmY gene cloned into BamHI-SpeI digested pSRK5 (expression from the lac promoter)	This work
pSRK5::rsmZ	GmR; 160 pb BamHI-SpeI fragment containing the rsmZ gene cloned into BamHI-SpeI digested pSRK5 (expression from the lac promoter)	This work
GmR, KmR, RifR, SpR,SmR, and TcR stand for resistance to gentamicin, kanamycin, rifampicin, spectinomycin, streptomycin, and tetracycline, respectively.

The plasmids carrying rsmX1-5, rsmY and rsmZ transcriptional fusions to ‘lacZ were constructed by cloning the promoter fragments generated by PCR (see Table 2 for oligonucleotides) in pME6016 [67]. The PCR fragments were ligated into pGEM-T, sequenced and then cloned into pME6016 after digestion with the appropriate restriction enzymes (Table 1).

Table 2.

Oligonucleotides used in this work

Name	Sequence (5ʹ→3ʹ)
Directed mutants
ΔRsmY.1	aaagaattCCAAGTTCAGCGAAGATGACG
ΔRsmY.2	GGCTTTCCAGACTGCCTGCGCTACGTCC
ΔRsmY.3	GGACGTAGCGCAGGCAGTCTGGAAAGCC
ΔRsmY.4	aaagaattcTTTCTTCCAGGGCGAACTGG
ΔRsmZ.1	aaagaattcATGATGATCTGTCGCAGAGC
ΔRsmZ.2	CCCACATTTTGTCCTGCCTGTCCGTTGG-
ΔRsmZ.3	CCAACGGACAGGCAGGACAAAATGTGGG
ΔRsmZ.4	aaagaattcGCCAGTGCTGCTTTGATTGC
Sequencing and verification
M13U	GTTTTCCCAGTCACGAC
M13R	AACAGCTATGACCATG
pME601*.F2	CTGCGATTCCGACTCGTCC
Z18	GATGTGCTGCAAGGCGAT
Promoter constructions
PrsmX1.F	aaagaattCCTTGGCTCAGGTGACAGC
PrsmX1.R	aaactgcaGGGTAGATTGCGACTTTCC
PrsmX2.F	aaagaATTCATGATCTCACCCC
PrsmX2.R	aaactgcaGAATTGGATCCTACATTG
PrsmX3.F	aaagaattCTCGCCCTGATCGCTC
PrsmX3.R	aaactgcaGGATCCAATATTACGG
PrsmX4.F	aaagaattcGCCTGCGTATTTTGCTG
PrsmX4.R	aaactgcaGGATCCAATCTTATGCG
PrsmX5.F	aaagaaTTCTACTCCTGATGTTCCG
PrsmX5.R	aaactgcaGGATCCAATATTACGAGG
PrsmY.F	aaagaattCAAGGCAGCCTTTG
PrsmY.R	aaactgcaGTACAAAGATTAAC
PrsmZ.F	aaagaaTTCCAGTAAACCTCCC
PrsmZ.R	aaactgcaGAGCAACACCCG
pSRK5 constructions
SrsmY.F	aaagatctCATGGACGTAGCGCAG
SrsmY.R	aaactagtGAAAACCCCGCCTAAG
SrsmZ.F	aaaggatccGCTGTGCCAACGGACAG
SrsmZ.R	aaactagtCACTTTTGCATAGACG

Capital letters: match to the sequence; lower case: added sequence; underlined: newly created restriction sites.

We constructed plasmids pSRK5::rsmY and pSRK5::rsmZ for in trans expression. Plasmid pSRK is a pBBR1-MCS2 derivative with the LacI operator region removed that allows constitutive transcription of regulatory RNAs [68]. A 1293 bp EcoRI-SpeI fragment from pSRK containing the lacI gene and a modified P_lac promoter was cloned into pBBR1-MCS5 digested with the same enzymes giving place to pSRK5. The full-length rsmY and rsmZ genes (i.e. from the transcription start site to the last residue of the Rho-independent terminator) were amplified by PCR using Pto DC3000 genomic DNA as template and the primer pairs SrsmY.F/R and SrsmZ.F/R that incorporate BglII (for rsmY) or BamHI [for rsmZ) and SacI sites to the 5′- and 3′-ends of the fragments, respectively. These PCR products were ligated to pGEM-T, sequenced and the rsmY and rsmZ loci were retrieved and inserted downstream the modified P_lac promoter in pSRK5, yielding plasmids pSRK5::rsmY and pSRK5::rsmZ, which were checked by sequencing.

Transformation of Pto DC3000 strains with the different plasmid constructions was carried out by electroporation. Electro-competent cells were prepared according to ⁶⁹, mixed with DNA (0.3–0.5 µg of DNA per ml of cell suspension], transferred to 0.1 cm cuvettes and submitted to a high-voltage pulse (1.800 V) for 5 ms by using an Eppendorf electroporator 2510. Transformants were selected on LB agar plates supplemented with the appropriate antibiotics.

Construction of directed mutants

We constructed ΔrsmY and ΔrsmZ directed mutants by deleting most of their genes. First, a region with rsmY or rsmZ adjacent sequences but lacking their genes were amplified by PCR with specific oligonucleotides (Table 2) and cloned into pK18mobsacB [70], which does not replicate in P. syringae. The Km^R sucrose^S plasmids were then electroporated into Pto DC3000. Transformants were selected in kanamycin (50 µg/ml), screened for sucrose sensitivity (15% [wt/vol]), and then grown in LB to force plasmid loss. Cells were then plated on LB with sucrose (15% [wt/vol]) and the sucrose^R Km^S colonies, which were expected to be double-recombinants, were selected and checked by PCR and sequencing. The ΔrsmYZ double mutant was obtained by electroporating the pK18ΔrsmZ suicide plasmid into the rsmY mutant.

Motility experiments

For swimming assays, Pto DC3000 and mutants were grown on LB plates for 48 h, resuspended in sterile milliQ water and adjusted to an OD₆₆₀ of 2.0. Two µl aliquots were stabbed into LB plates (0.3% agar) and incubated at least for 48 h at 20°C when the diameters of the swimming halos were measured. For swarming motility assays, the 2 µl aliquots were dropped in the centre of peptone-glucose (PG)-agar plates (0.5% protease peptone No. 3 (Difco 212,693), 0.2% glucose, 0.5% Difco Bacto-Agar) and incubated at 20°C and observed after 24 h. Three motility plates were used for each strain, and the experiment was repeated with three independent cultures (a total of 9 motility plates per strain).

Syringafactin production

Syringafactin was detected with the atomized oil assay previously described [54, 56]. Pto DC3000 and mutants were grown on LB plates for 48 h and resuspended in sterile milliQ water. Ten µl aliquots (OD₆₆₀ = 1.0) were pipetted onto the surface of LB plates, incubated 24 or 48 h at 20°C and then sprayed with a mist of mineral oil (Sigma M5904) using a sprayer with an air flow of 6 l/min. The diameter of the visible halo of brighter oil drops (surfactant halo) and the area of the producing bacterial colony were measured with ImageJ. The normalized halo area was calculated by subtracting the colony area from that of the surfactant halo. In general, larger halos in the atomized oil assay indicate strains with higher levels of surfactant production. Three plates were used for each strain, and the experiment was repeated with four independent cultures.

β-galactosidase assays

Bacterial cells from LB, MMR and MMF agar plates grown for 3 days at 20°C were resuspended in milliQ water, adjusted to an OD₆₆₀ of 0.5 and the activity was determined after permeabilising the cells [71]. Agar plate assays were performed at least three times with two biological replicates each. β-galactosidase activity was also measured in cells from LB, MMR and MMF liquid cultures. For this, strains grown on LB plates at 28°C were resuspended in sterile milliQ water, centrifuged, washed and resuspended in the corresponding medium. The suspensions were adjusted to OD₆₆₀ ≈ 4.0 − 6.0 and 150 µl were dispensed in a multi-well plate (Greiner Cellstar 96 round bottom, M9436) in which 1:2 serial dilutions until OD₆₆₀ ~ 0.125 were made. Several controls of non-inoculated medium were included at random plate positions. The plates were incubated at 20°C with constant stirring at 600 rpm on an IKA MS 3 Basic shaker with a multi-well plate holder for 5 h (LB), 12 h (MMR) or 48 h (MMF). Each experiment was performed independently at least 3 times and for each condition 2–4 biological replicates were used according to the experiment. The enzymatic reactions were carried out in polypropylene DeepWell plates (VWR 732–3323) kept at 30°C in a water bath. The measurements were performed in an Aeon Biotek plate reader with the Gen5 software after transferring 100 µl of the reaction volume to flat-bottom plates (Greiner Cellstar 96 flat bottom, M0812).

Infection assays

Seeds of Solanum lycopersicum cultivar Moneymaker (a PtoS line, i.e., a compatible host for Pto DC3000) were germinated and grown with 16/8-h light/dark cycles, at 24/16°C day/night and 70% relative humidity in a plant growth chamber. Pto DC3000 strains grown on LB plates for 48 h at 28°C, were suspended in sterile milliQ water, the concentration adjusted to 10⁸ CFU/ml, and the strains were applied to one leaf of three different plants with an airbrush until the leaf (adaxial and abaxial) surface was uniformly wet. The analysis of symptom development and sampling was performed 3 h after inoculation (time 0), when the leaves were dried, and several days after inoculation (3, 6 and 10 dpi) to monitor bacterial growth in plant. Bacteria were recovered from the infected leaves using a 10-mm-diameter cork-borer, sampling 20 disks per plant. Five disks (3.9 cm²) were homogenized by mechanical disruption into 1 ml of milliQ water and counted by plating serial dilutions onto LB plates with the corresponding antibiotics. The severity of symptoms was evaluated as the percentage of necrotic area per leaflet induced by the inoculated strains at 10 dpi. The necrotic areas were digitally measured using ImageJ [72] on five inoculated leaflets of, at least, three different plants.

HR assays

Hypersensitive response assays were carried out by infiltration of Phaseolus vulgaris cv. Canadian Wonder (a PtoR line, i.e., a resistant host that limits Pto DC3000 infection with a hypersensitive response) leaves with Pto DC3000. Bean seeds were germinated and grown in pots with 16/8 h light/dark cycles, at 24/16°C day/night and 70% relative humidity in a plant growth chamber. Pto DC3000 strains were suspended in milliQ water (10⁷ CFU/ml) from LB plates incubated for 48 h at 28°C. A small area of the abaxial leaf surface was inoculated using a blunt syringe and the symptoms were visualized every day until 8 dpi. Up to six different strains were inoculated in the same leaf for comparison and at least 10 plants were used for each experiment.

Statistical analysis

Statistical treatment of data was performed using R or Graphpad Prism 6 software. Comparison among different strains or conditions was performed by one-way ANOVA with post-hoc Tukey HSD test and differences between two strains were assessed by t-test using Welch′s correction when unequal variances existed.

When performing ANOVA for the evaluation of significant differences among diverse datasets, the different categories were named using a letter code. This means that the differences between groups, indicated with different letters, are statistically significant, whereas the groups that share a letter do not exhibit significant differences with each other. When ANOVA was applied to two different data sets (obtained, for instance, at two different times), single letters were used to distinguish the categories generated at one time and the apostrophe (‘) for the categories obtained at the other time.

Results

In vivo expression of the rsm RNAs under different growth conditions

The fact that rsmX1-5 sequences are almost identical makes it difficult to independently quantify them by RT-qPCR, so measuring the β-galactosidase activity of the different rsm promoter (P_rsm) transcriptional fusions to ‘lacZ was the alternative used. In P. aeruginosa the rsm RNAs are differentially regulated according to the growth conditions, varying between planktonically and surface-grown cells [32]. To investigate that in Pto DC3000, we monitored the in vivo expression of rsmX1-5, rsmY and rsmZ in planktonically and surface-grown bacteria. In addition, the JA257 gacA mutant was used to determine the level of dependence on this activator for each rsm.

We started measuring Pto DC3000 rsm expression after growing the wild type and the JA257 gacA mutant on LB, MMR (mannitol/glutamate minimal medium) and MMF (minimal medium simulating the plant apoplast composition that induces the expression of the T3SS) plates for 3 days (Fig. 1). Overall, similar expression profiles were obtained in all the media tested, but three categories can be distinguished depending on the P_rsm expression levels. The first includes rsmX3, rsmX5 and rsmY, with high expression; the second comprises rsmX1, rsmX2 and rsmX4, with medium expression, and, the third rsmZ, with low expression. Most of the rsm promoters showed a drastic decrease in their expression in the absence of gacA, except for rsmX5 and rsmZ, which exhibited small or no differences with respect to the wild type in most of the media tested (Fig. 1). Interestingly, the effect of GacA on the different rsm promoters varied with the culture medium. Thus, rsmX4 in LB, rsmX1 in MMR and rsmY in MMF were the most dependent on GacA, whereas rsmX5 is expressed at high levels in the JA257 gacA mutant, regardless of the culture medium.

Figure 1.

P_rsm expression in LB, MMR and MMF solid media

β‐Galactosidase activities of transcriptional fusions from Pto DC3000 rsmX1-X5, Y, and Z promoters to ’lacZ were determined in the wild type (dark grey) and in the JA257 gacA mutant (light grey). The graph shows the average activity in Miller units (MU), and error bars correspond to the standard deviation of three independent experiments with two biological replicates each. Promoter expressions are statistically different (p < 0.01) between strains, except for the cases marked with ns (non-significant), according to a t-test.

The expression of the rsm RNAs fluctuates with the culture conditions and the growth phase in various Pseudomonas strains, usually reaching maximum levels at the stationary phase [16,28,30–35,73]. To test for differential expression, the activity of all the rsm promoters was measured in cells grown in liquid media at different optical densities. In MMF, the promoter activities increased linearly until they reached high cell densities (greater than OD₆₆₀ = 4.0) (Fig. 2), and they behaved similarly in other laboratory media, like LB and MMR (Fig. S2 and S3). Therefore, the rsm levels seem to be an indicator of the culture population density, as previously proposed [16,34,43,73,74]. Nonetheless, gene categories based on that expression can be established: high for rsmX3, rsmY and rsmX5, medium for rsmX2 and rsmX4, and low for rsmX1 and rsmZ. Moreover, the deletion of gacA caused similar effects in all the media tested but more clearly in MMF (Fig. 2, S2 and S3): the expression from rsmX2, rsmX3, rsmY and rsmX4 promoters considerably dropped, whereas that of rsmX5, rsmZ and rsmX1 diminished moderately. Hence, the fact that P_rsm activity levels are similar in both solid and liquid cultures suggests that surface detection is not involved in the expression of these RNAs in Pto DC3000.

Figure 2.

P_rsm expression varies throughout the growth curve in MMF liquid cultures

Pto DC3000 rsmX1-X5, Y, and Z expression was assessed fusing their promoters to ’lacZ and measuring β‐galactosidase activity. The wild type (closed circles, full line) and the JA257 gacA mutant (open circles, dashed line) were grown for 48 h at 20°C in a microtiter plate including different dilutions of starter cultures. Measurements are represented in scatter plots and each strain’s data are fitted to a nonlinear regression.

Overall, rsm RNAs reached maximum levels at high cell densities, but were unaffected by surface detection. The most remarkable observation is that, independently of the culture medium, rsmX5 was expressed at high level in both the wild type and the JA257 gacA mutant, suggesting that it is the only rsm in Pto DC3000 whose high expression does not require GacA. Therefore, independently of the culture medium, rsmX5 is the main component of the total rsm pool in a gacA mutant. The other rsm promoters were activated by GacA to different degrees, whereas P_rsmX1 and P_rsmZ were the most independent.

Effects of rsmY and rsmZ levels on Pto DC3000 phenotypes

rsmY and rsmZ are the primary sRNAs that sequester RsmA/RsmF in P. aeruginosa, whereas the other rsm RNAs appear to be less physiologically important [36,37]. For this reason, we decided to start by determining the role of rsmY and rsmZ in Pto DC3000, for which single and double mutants were constructed and characterized. Since it was previously shown that rsmY and rsmZ mutations had reduced effect in P. protegens CHA0 and P. fluorescens F113, whereas their overexpression overrode the pleiotropic effects of gacS and gacA mutations [16,28,29,75,76], we decided to clone rsmY and rsmZ in plasmids and study their influence on Pto DC3000 phenotypes when expressed in trans. First, we assessed the growth of the rsmY, rsmZ and the double rsmYZ mutant strains in MMF at 20°C, observing that their growth rates were similar to that of the wild type and different from that of the JA257 gacA mutant, which exhibited diminished growth (Fig. S4A). It should be noted that rsmY and rsmZ overexpression in the wild type and the JA257 gacA mutant had a minor impact on their respective behaviours (Fig. S4B).

Motility

Since the Gac-rsm pathway regulates motility traits in Pto DC3000 [52,57,58], swimming and swarming motility assays were carried out with JA257 gacA and rsm mutants to establish whether rsmY and rsmZ contributed to that phenotype. The swimming of the three rsm mutants was similar to that of the wild type, but the gacA halo was significantly smaller than that of the wild type (Fig. 3A). In trans expression of rsmY and rsmZ did not change the gacA defective phenotype even after 5 days (not shown), but it significantly increased the swimming halo of the wild type (Fig. 3B).

Figure 3.

rsmY and rsmZ have little effect on swimming motility

Swimming assay comparing the wild type strain with JA257 gacA, rsmY, rsmZ, rsmYZ mutants (A) and the wild type strain with the JA257 gacA mutant both carrying the constitutive expression plasmids pSRK5 (white), pSRK5::rsmY (light grey) and pSRK5::rsmZ (dark grey) (B). Cell suspensions were punctured in the centre of Luria-Bertani (0.3% agar) plates and were incubated 48 h at 20°C, when pictures were taken and swimming halos were measured. The graphs show the average diameter of the halos and error bars correspond to the standard deviation of three independent experiments with three biological replicates each; letters a to c denote analysis of variance (ANOVA) categories with significant differences (p < 0.01).

Regarding the swarming motility, the behaviour of the rsmZ mutant was similar to the wild type, whereas the rsmY and the double rsmYZ mutants were slower (Fig. 4A). Conversely, JA257 gacA cells remained at the site of inoculation. In trans expression of rsmY and rsmZ caused a significant increase of the wild type swarming, but the gacA mutant’s swarming area only increased in the presence of rsmZ, becoming noticeable after 48 h of incubation (Fig. 4B). It has been shown that the Pto DC3000 swarming phenotype is positively regulated by GacA and antagonized by the CsrA3 and CsrA2 proteins [52,57,58,63]. Taken together, these results support a model where rsmY and rsmZ promote swarming motility by sequestering the CsrA regulatory proteins.

Figure 4.

rsmY and rsmZ promote swarming motility

Swarming behaviour of the wild type strain along with JA257 gacA, rsmY, rsmZ, rsmYZ mutants (A) and the wild type strain together with the JA257 gacA mutant both carrying the constitutive expression plasmids pSRK5 (white), pSRK5::rsmY (light grey) and pSRK5::rsmZ (dark grey) (B). Bacterial suspensions were deposited on the surface of PG (0.5% agar) plates and were monitored every 24 h after incubating them at 20°C. The graph shows the average swarming surface and error bars correspond to the standard deviation of three independent experiments with three biological replicates each; letters a to d or a’ to b’ denote ANOVA categories with significant differences (p < 0.01). The dotted line represents the total area of a Petri dish.

Given that CsrA3 and CsrA2 inhibited swarming by decreasing biosurfactant synthesis [58], we assessed syringafactin production, observing that the surfactant halo generated by the rsmY/Z mutants was similar to that of the wild type, whereas the JA257 gacA mutant was severely impaired in syringafactin production (Fig. 5A), as seen before with the AC811 strain [58]. In trans expression of rsmY and rsmZ significantly increased syringafactin production in the wild type after 24 h, but it did not have any effect on the gacA mutant. However, 48 h later rsmZ significantly increased the syringafactin production halo in the gacA mutant (Fig. 5B). Overall, syringafactin levels matched the swarming phenotypes and the higher production of syringafactin by the JA257 gacA mutant overexpressing rsmZ seems to be the cause of its greater swarming motility.

Figure 5.

rsmY and rsmZ induce syringafactin production

Comparison of surfactant-induced halos around bacterial colonies among the wild type strain together with JA257 gacA, rsmY, rsmZ, rsmYZ mutants (A) and the wild type strain along the JA257 gacA mutant both carrying the constitutive expression plasmids pSRK5 (white), pSRK5::rsmY (light grey) and pSRK5::rsmZ (dark grey) (B). The strains were grown on Luria-Bertani (1% agar) plates for 24 h or 48 h at 20°C and visualized with atomized oil. Biosurfactant areas were calculated and the means with the standard deviations of three experiments with three biological replicates each were plotted, lower case letters a to d or a’ to d’ denoting analysis of variance (ANOVA) categories with significant differences (p < 0.01).

Virulence

To study the impact of rsmY and rsmZ on Pto DC3000 virulence, we carried out infection assays in tomato plants comparing the abilities of the wild type and the rsmY, rsmZ and rsmYZ mutants to infect and multiply in tomato leaf tissues by monitoring bacterial populations and development of disease symptoms for 10 days after inoculation by spray (Fig. 6A and C). All the mutants entered the tomato leaves and grew in the apoplast where their populations reached a maximum at 3 dpi. The disease symptoms caused by the wild type and all the assayed mutants were similar: small water-soaked lesions that appeared 2–3 days after inoculation and soon turned brown with the surrounding tissue turning yellow (Fig. 6C). The severity of the symptoms (quantified as the extension of the necrotic areas) caused by all the mutants was similar to that of the wild type (not shown). These results indicate that the loss of rsmY and rsmZ does not modify plant growth or virulence.

Figure 6.

Bacterial growth and symptom development on tomato leaves

Time course of growth of the indicated strains in the primary leaves of tomato plants inoculated with approximately 10⁸ CFU/ml by spray. CFU were quantified at 0, 3, 6, and 10 days postinoculation (dpi) and represented in a graph as the average of four experiments with their standard error (A and B). An analysis of variance (ANOVA) test was made for each sampling time, finding only significant differences (p < 0.01) at 3 dpi between the strains bearing the empty plasmid and both JA257 rsmY and rsmZ expressing strains (B). Development of symptoms induced on tomato leaves 10 days after inoculation with the wild type and different mutants at 10⁸ CFU/ml by spray (C). Hypersensitive response caused by different Pto DC3000 mutants on bean leaves at 8 dpi (D). All the strains were infiltrated at 10⁷ CFU/ml, and the hrcQ-U mutant was used as a negative control.

The JA257 gacA mutant behaved similarly to the wild type both at internal leaf colonization and symptom development in tomato (Fig. 6B and C), different to what was previously shown for Arabidopsis thaliana, where GacA was required for colonization of leaves but not for virulence in the apoplast [59,60]. To study the effect of rsmY or rsmZ overexpression in the JA257 gacA mutant, infection assays were performed. In trans expression of rsmY and rsmZ in the mutant did not significantly modify in planta growth or symptom development (Fig. 6B and C), probably due to plasmid loss. When monitoring plasmid stability throughout the assay, it was observed that it was only present in 40–50% of the population at 3 days and 30–35% at 6 days. This suggests that increasing levels of these RNAs did not confer an adaptive advantage for the bacteria colonizing the apoplast.

Subsequently, we evaluated the hypersensitive response (HR) caused by the same strains when infiltrated in bean leaves to infer their T3SS functionality (Fig. 6D). The rsmY, rsmZ and rsmYZ mutants elicited an HR equivalent to that of the wild type, whereas it was absent in the T3SS-deficient hrcQ-U mutant (negative control). Moreover, the expression of rsmY and rsmZ affected neither the wild type nor the JA257 gacA mutant phenotype.

In summary, rsmY and rsmZ do not seem to be required for Pto DC3000 pathogenicity under the assayed conditions, probably because the rsmX1-5 small RNAs can compensate for their absence.

Transcriptional regulation of rsmY, rsmZ and rsmX5

The transcriptional regulator PsrA, known to be a transcriptional activator of the rpoS gene and a repressor of a fatty acid degradation operon [77,78], was shown to be an activator of the rsmZ promoter but not of the rsmY one in P. protegens CHA0 [30]. In order to verify if it was the same in Pto DC3000, the β-galactosidase activity from P_rsmY and P_rsmZ was determined in a psrA mutant, observing that the psrA mutation did not significantly modify the expression of any of them along the growth curve (Fig. S5).

Negative-feedback regulation, a mechanism by which RsmA/CsrA increases the levels of the csr/rsm sRNAs, has been observed in E. coli, S. enterica, V. cholerae, P. aeruginosa, P. fluorescens and P. protegens, although the mechanisms remain unclear in most of the cases [28,30,40–45,79,80]. To determine whether transcription of the rsmY and rsmZ genes was altered by csrA3 in Pto DC3000, we measured β-galactosidase activity in the wild type overexpressing csrA3 and compared it with the strain bearing the empty plasmid (Fig. 7A). The specific β-galactosidase activity from both rsmY and rsmZ fusions was strongly induced by csrA3 and increased with cell density. Next, we measured β-galactosidase activity in the csrA3 mutant and its complemented strain (Fig. 7B). The expression of P_rsmY exhibited a noteworthy decrease in the csrA3 mutant, whereas the activity of P_rsmZ did not substantially change. However, both promoters were strongly induced by csrA3, reaching lower levels in the csrA3 mutant than in the wild type strain. Finally, we also studied the effect of csrA3 overexpression in the JA257 gacA mutant, observing that the expression of rsmY significantly diminished in gacA, whereas that of rsmZ remained the same. However, unlike in the wild type and the csrA3 mutant, the expression from both promoters was not modified by csrA3 overexpression (Fig. 7C). Therefore, the activation of rsmY and rsmZ expression by CsrA3 is GacA-dependent, as it occurred in P. protegens CHA0 [28,30].

Figure 7.

CsrA3 promotes rsmX5, rsmY and rsmZ expression in a GacA-dependent manner

β‐Galactosidase activities of transcriptional fusions from Pto DC3000 rsmX5, Y, and Z promoters to ’lacZ were determined in the wild type (A), csrA3 mutant (B) and JA257 gacA mutant (C) carrying pBBR1-MSC5 (closed squares) and pBBR1-MCS5::csrA3 (open squares). Different dilutions of MMF cultures were grown for 48 h at 20°C in a microtiter plate, measured and represented in scatter plots, each strain’s data are fitted to a nonlinear regression.

Since rsmX5 reached high levels in the absence of GacA (Fig. 2), we examined whether CsrA3 was able to induce the expression from this rsm promoter by measuring β-galactosidase activity in the wild type, csrA3 and gacA strains overexpressing csrA3 (Fig. 7). We observed that csrA3 affected P_rsmX5 in the same way as P_rsmY and P_rsmZ, that is, increasing its expression in a GacA-dependent manner. (Fig. 7). However, the increase was subtler for rsmX5, probably due to the high levels already present in the cell.

Comparison between AC811 and JA257 gacA mutants

All the experiments shown in this work were performed with the AC811 gacA mutant [57]. When J. Anderson’s lab papers came out showing that some of the phenotypes reported for the AC811 mutant were caused by a polar effect on uvrC and a point mutation in anmK [59,60], we obtained the JA257 gacA mutant and conducted all the experiments again. We have observed that JA257 and AC811 growth curves in LB, MMR and MMF (Fig. S4 and S6A), swarming phenotype (Figs. 4 and ⁵⁸) and syringafactin production were similar (Figs. 5 and ⁵⁸). Also, the activity of the different rsm promoters was comparable (Fig. 1 and S6B). However, they behaved differently in swimming, since the AC811 halo [58] was significantly smaller than that of JA257 (Fig. 3), tomato infection assays, as AC811 was severely impaired for growth in the apoplast (Fig. S6C) and provoked less symptoms (Fig. S6D), and HR assays, in which AC811 phenotype was milder than both the JA257 gacA mutant and the wild type (Fig. S6E).

Discussion

In order to switch efficiently between its two natural niches, the water and the plant host, the model phytopathogenic bacterium Pto DC3000 must sense changes in its environment and respond by rapidly altering gene expression to promote adaptation. Sensing through the TCS GacS/GacA has shown to be required for colonization of the plant host [60], and is likely needed for the bacterium to successfully transition from the host back into the aqueous environment. This TCS induces the transcription of the rsm sRNAs, which sequester the CsrA regulatory proteins. Pto DC3000 has a complex Gac-rsm pathway with five CsrA proteins and seven rsm RNAs which control virulence, motility, production of secondary metabolites, carbon metabolism and quorum sensing [52,57–60,63]. Homologous ncRNAs exert their regulatory effects by having redundant, additive or independent functions [81].

To investigate the expression and regulation of the different Pto DC3000 rsm RNAs and elucidate their implications in the physiology of this bacterium, we determined their promoter expression patterns under different conditions. They vary with the culture medium and the cell density, but not between planktonically and surface-grown cells, suggesting that surface detection is not involved in the expression of these RNAs, at least under the conditions assayed. Similarly to what occurs in other Pseudomonas, where rsm expression fluctuates with the culture conditions but is usually maximum at stationary phase [16,28,30,32,34,36], the expression of all the Pto DC3000 rsm RNAs is remarkably upregulated by cell density, as previously observed for most of them [73]. It has been shown that the signal detected by GacS/GacA accumulates in stationary phase and creates a positive feedback loop in the Gac-rsm cascade of P. protegens CHA0, P. aeruginosa, E. coli, Salmonella and Vibrio cholerae [28,30,33,40,41]. The perception of those signal molecules that lead to activation of the Gac-rsm cascade in P. protegens CHA0 depends on GacA, the three small RNAs, the RsmA and RsmE proteins, and a functional GacS [16,28,29,82–83]. We believe that this mechanism also operates in Pto DC3000 and the Gac-rsm cascade positively regulates its activity as a function of increasing cell population densities with an autoinduction pattern similar to quorum-sensing regulation depending on N-acyl-homoserine lactone signalling [28].

In the present study, we show that Pto DC3000 rsm sRNAs are differentially expressed and regulated. First, we observed variances in the promoter activities of the seven RNAs, both in the wild type and the gacA mutant, that correlate with the degree of conservation of the corresponding promoter sequences for σ⁷⁰ binding (Table 3 and Fig. S7). The rsmX3, rsmX5 and rsmY promoters, with the highest expression, have A/T-rich −10 regions with sequences highly similar to the consensus (TAATCT in P_rsmY and TAATAT in P_rsmX3 and P_rsmX5). The lower expression of rsmX2 and rsmX4 may be due to the presence of Gs in their −10 regions. In fact, the P_rsmX4 − 35 region is identical to that of P_rsmX5 but its −10 region differ in a G. rsmX1 and rsmZ, the RNAs with the lowest levels of expression, have poorly conserved sequences, both at the −10 and −35 regions.

Table 3.

Comparison of the seven Pto DC3000 rsm promoter regions

*Gac-box: similar to the consensus proposed for Pto DC3000 (bold).

**Promoter: nucleotides in −35 and −10 positions (bold).

^§From the first nucleotide (5ʹ) of the Gac-box.

^#Consensus sequence proposed for P. aeruginosa Gac-box [33] and P. aeruginosa and P. putida σ⁷⁰ promoters, respectively [98].

The conserved palindromic UAS, also called Gac-box, is essential for the high expression of the rsm genes in γ-proteobacteria and their positive regulation by GacA [14,28,29,61,74,84]. We have analysed the seven rsm promoter sequences observing that the in vivo hierarchy of GacA-dependence (rsmX4≈ rsmY>rsmX3≈ rsmX2> rsmX1> rsmZ = rsmX5) correlates with the conservation of the respective binding sites upstream of these genes. The sequence of the Gac-box in the P_rsmY is identical to the TGTAAGN₆CTTACA consensus proposed for P. aeruginosa PAO1 and, accordingly, its dependence on GacA was high. Moreover, other conserved nucleotides may also contribute to the Pto DC3000 Gac-box and we propose the palindromic sequence GTGTAAGCA(a motif)-N₂-TGCTTACAC(b motif) as the GacA binding site in this strain (Table 3). The Gac-box present in the rest of the rsm promoters is an imperfect palindrome and includes an array of variations from the consensus, though the high expression of some of them still largely relies on GacA activation. Interestingly, all the P_rsmX have the same Gac-box-b motif (GGCTTACAC), except for P_rsmX1 (GGCTTACGC) and P_rsmX2 (GGCTTACAT). P_rsmZ exhibited lower GacA-dependence than P_rsmX1-4, even when its Gac-box-a motif (TTGTAAAGTC) is conserved, matching the P. aeruginosa consensus. However, its Gac-box-b motif is very divergent so its low GacA dependence is probably caused by the presence of the −172 G. It should be noticed that the Gac-box in this promoter is located further upstream, at −190, and a third Gac-box-c motif can be found downstream the Gac-box-b (Table 3). However, the P_rsmZ low GacA-dependence may not be due to the Gac-box position, since the Gac boxes of E. coli csrB or P. aeruginosa rsmZ promoters, highly dependent on GacA activation, have similar sequences and are located at the same distance [14,85]. P_rsmX5, the most GacA-independent promoter, has a Gac-box-a motif significantly divergent from that of P_rsmY and it is no longer palindromic with the b motif (−75 C instead of T), which suggests that this nucleotide is key for GacA binding. In summary, GacA-dependent transcriptional activation of P_rsm promoters is governed by the conservation of the Gac-box sequences, and nucleotides T2 and T4 in the Gac-box-a motif together with their corresponding nucleotides (A8 and A6) in Gac-box-b seem to be essential for GacA recognition. The degree of GacA-dependence is variable among rsm promoters of several genera such as Escherichia, Vibrio and Pseudomonas [28,34,40,74,85] and it allows the integration of other signals in the pathway. For instance, transcription of P. aeruginosa rsmY and rsmZ exhibited strong GacA-dependence whereas P_rsmV and P_rsmW activity showed no difference between the wild type and the gacA mutant [14,31,37]. Interestingly, rsmW can be transcribed from a σ³²-dependent promoter and its expression is regulated by temperature [37]. P_rsmY, P_rsmX2, P_rsmX3 and P_rsmX4 are highly dependent on GacA activation, whereas P_rsmX1 and P_rsmZ are partially dependent, and P_rsmX5 is the most GacA-independent promoter. It remains to be determined if they respond to other transcription factors.

Unquestionably, GacS/GacA signalling is essential for achieving high rsm levels in Pto DC3000, but in a gacA mutant some rsm RNAs are still expressing: rsmX5 at high level, rsmY and rsmZ at a medium level and rsmX1 at a low level. That would explain why a strain overexpressing csrA3 has a more extreme phenotype than the gacA mutant, since CsrA2 and CsrA3 are able to bind to all the rsm [63]. Remarkably, the expression of P_rsmX5 is different from the rest of the rsm promoters and the highest in the gacA mutant (Fig. 2, S2 and S3). This means that rsmX5 constitutes the main fraction of the total rsm pool in the gacA mutant, regardless of the culture medium. We speculate that the biological meaning of the high rsmX5 basal levels is conferring stability to the system by keeping the regulatory feedback loops active (see below).

Independently of the number and expression levels of the rsm RNAs, they exert the same function of antagonizing the CsrA/RsmA proteins. This leads to the question of why multiple sRNAs and CsrA proteins are produced in Pto DC3000 when they have the same function. In principle, this arrangement allows a more efficient regulatory response via a gene dosage effect, and the rsm RNAs seem to be functionally redundant under certain conditions. Thus, the rsmY, rsmZ and rsmYZ mutants were indistinguishable from the wild type in all the phenotypes analysed, except swarming. Consequently, Pto DC3000 rsmX1-5 are sufficient for keeping a wild type-like behaviour in the absence of rsmY and/or rsmZ. Likewise, only the deletion of all rsm/csr sRNA genes in E. coli, P. entomophila, P. fluorescens, S. enterica and V. cholerae produced a distinguishable phenotype from the wild type, whereas artificial overexpression of one of those sRNAs suppressed the negative effects of gacS and gacA mutations on target gene expression [16,28,38,42,45,74,76,86,87]. However, Pto DC3000 rsmZ, but not rsmY, is able to partially complement the gacA defective phenotype in swarming and syringafactin production, whereas in trans expression of rsmY and rsmZ in the gacA mutant did not significantly change in planta populations or symptoms. The different effects of the rsm RNAs depending on the phenotypes have also been observed in other systems. In P. aeruginosa, rsmZ, but not rsmY, overexpression impaired biofilm formation [88], constitutive expression of rsmZ, but not rsmX, rescued phenazine and AHL production in P. chlororaphis gac mutants [87], and rsmY from A. vinelandii regulates alginate and alkylresorcinols but not poly-β-hydroxybutyrate synthesis [89]. Therefore, the existence of multiple rsm sRNAs and CsrA proteins allows differential regulation of the phenotypes controlled by the Gac-rsm pathway. It has been shown that Pto DC3000 rsmX1, rsmX5, rsmY, and rsmZ exhibited similar in vitro binding affinity for the different CsrA ortologs, whereas CsrA1, CsrA2, CsrA3 and CsrA4 proteins displayed distinct binding affinities to each of those sRNAs [63]. Thus, an important functional difference lies in the ability of the CsrA proteins to bind to the diverse rsm molecules. In fact, the seven rsm RNA molecules have diverse 3D structures (Fig. S1), which probably determine distinct stabilities, affinities and sequestration properties of the regulatory proteins and, therefore, in vivo outcomes [25,27,62,90]. In summary, our results show that Pto DC3000 rsm sRNAs seem to be functionally redundant, but their differential expression together with their specific stabilities and sequestration characteristics, most likely confers different CsrA binding properties and affinities, as it occurs in other bacteria [25,27–29].

The complexity of the system is even greater as the expression of at least three rsm sRNAs (rsmY, rsmZ and rsmX5) significantly decreased when csrA3 was knocked out, and dramatically improved when overexpressed, indicating a feedback mechanism by which CsrA3 regulates the level of its antagonistic sRNAs in Pto DC3000. This regulation has been also observed in other γ-proteobacteria, like P. protegens CHA0, E. coli and S. enterica or V. cholerae [28,30,40–44]. RsmA and RsmE have the same positive effect on P. protegens CHA0 rsmX, rsmY and rsmZ levels [28,30]. In E. coli, UvrY (GacA) protein levels decreased in a csrA mutant, and CsrA was found to indirectly activate csrB/C transcription. This indirect activation occurs via positive effects of CsrA on uvrY expression and the ability to cause BarA (GacS) to switch from its phosphatase to kinase activity [91]. In V. cholerae, CsrA positively regulates the expression of the csr sRNAs through its effects on VarA (GacA) since it positively regulates VarA protein levels by binding directly to the varA mRNA promoting its expression [40]. This negative feedback loop can cause a number of regulatory outcomes, including acceleration of response times and decreased cell-cell variability. In E. coli, negative feedback has been shown to reduce response times since sequestration of CsrA by csrB/C allows rapid reduction of CsrA activity without the need for its dilution via growth [92]. The presence of such feedback mechanism in Pto DC3000 is not surprising, given the importance of controlling the level of active CsrA3, which is functionally the most important CsrA protein followed by CsrA2 [58,63]. Our previous work indicated that a complete deletion of the csrA3 gene caused significant growth defects, and its overexpression allowed growth but not virulence in tomato [58]. One way to prevent drastic fluctuations in CsrA3 activity is to regulate the level of rsm sRNAs produced in response to the levels of available CsrA3. When unsequestered CsrA3 exceeds a threshold, it indirectly induces GacA activity, by a mechanism yet unknown, and this leads to an increased production of rsm sRNAs, which in turn bind to CsrA3 reducing its availability. This ensures optimal levels of active CsrA within the cell.

Overall, the existence of multiple rsm sRNAs in several P. syringae pathovars allows differential regulation of their synthesis and/or stability in response to diverse environmental stimuli. Having several CsrA proteins as well helps to respond rapidly and efficiently to changing conditions, which is extremely important for the success of this pathogen that needs to adopt different lifestyles depending on the presence or the absence of its host and also has to deal with the selection pressure exerted by the plant.

Funding Statement

This work was supported by the ERDF/Spanish Ministry of Science, Innovation and Universities - State Research Agency under Grants BIO2014-55075-P and BIO2017-83533-P;Ministerio de Educación, Cultura y Deporte [ECD/1619/2013];Deutscher Akademischer Austauschdienst (DE) [RISE-Worldwide scholarship];

Disclosure statement

No potential conflict of interest was reported by the authors.

Supplementary material

Supplemental data for this article can be accessed here.