Characterisation of SalRAB a Salicylic Acid Inducible Positively Regulated Efflux System of Rhizobium leguminosarum bv viciae 3841

Salicylic acid is an important signalling molecule in plant-microbe defence and symbiosis. We analysed the transcriptional responses of the nitrogen fixing plant symbiont, Rhizobium leguminosarum bv viciae 3841 to salicylic acid. Two MFS-type multicomponent efflux systems were induced in response to salicylic acid, rmrAB and the hitherto undescribed system salRAB. Based on sequence similarity salA and salB encode a membrane fusion and inner membrane protein respectively. salAB are positively regulated by the LysR regulator SalR. Disruption of salA significantly increased the sensitivity of the mutant to salicylic acid, while disruption of rmrA did not. A salA/rmrA double mutation did not have increased sensitivity relative to the salA mutant. Pea plants nodulated by salA or rmrA strains did not have altered nodule number or nitrogen fixation rates, consistent with weak expression of salA in the rhizosphere and in nodule bacteria. However, BLAST analysis revealed seventeen putative efflux systems in Rlv3841 and several of these were highly differentially expressed during rhizosphere colonisation, host infection and bacteroid differentiation. This suggests they have an integral role in symbiosis with host plants.


Introduction
Plants produce and secrete a diverse number of compounds into the rhizosphere. These include a myriad of phytoalexins and signalling molecules which not only mediate plant defences but orchestrate plant microbial interactions including symbiosis with nitrogen fixing symbiotic rhizobia [1,2]. One key molecule in the response of plants to microbes is salicylic acid, which is a phenolic hormone with varied roles in plant metabolism and physiology including plant defence [3]. When a plant recognises a biotrophic pathogen, salicylic acid regulates the specific and localised Hypersensitive Response (HR) leading to cell death at the site of infection [4]. Salicylic acid is also involved in the longer lasting systemic protection of the plant against a range of pathogens, termed Systemic Acquired Resistance (SAR) [5]. Plants and microbes have evolved complex signal interactions in order to distinguish friend from foe. In the case of symbioses with rhizobia plants do not usually elicit a defence response [6], although salicylic acid may be important in controlling host range and regulating nodule formation. A number of studies have investigated salicylic acid levels and nodule formation. Exogenous application of salicylic acid has been shown to decrease or inhibit nodule formation when Bradyrhizobium japonicum or Rhizobium leguminosarum are grown on Soybean and Vetch respectively [7,8]. Similarly, decreasing endogenous levels of salicylic acid in Lotus japonicus led to increased nodule numbers when inoculated with Mesorhizobium loti [6].
One mechanism to circumvent the toxic effects of compounds that microbes encounter in the environment is their extrusion via efflux systems. These systems are ubiquitous in bacteria and comprise five superfamilies [9]. The first belong to the ATP Binding Cassette (ABC) family, which are primary transporters using ATP hydrolysis to drive efflux. The other four, The Major Facilitator Superfamily (MFS), the Resistance Nodulation Division (RND), the Multi Antimicrobial Extrusion (MATE) and Drug Metabolite Transporters (DMT)/SMR are secondary H + or Na + antiporters. In gram negative bacteria some ABC, MFS and all RND are multicomponent tripartite systems spanning both inner and outer membranes. They comprise an inner membrane transporter, a membrane fusion protein and a TolC like outer membrane factor [10]. Usually, the inner membrane transport and membrane fusion protein are located together however, cells generally have several TolC like proteins that act with a number of different efflux systems [10].
Successful phytopathogenic bacteria must export or detoxify plant phytoalexins, with disruption of two RND type efflux systems in Erwinia amylovora (ArcAB) and Pseudomonas syringae (MexAB) reducing virulence on apple trees and bean leaves respectively [11,12]. The role of these systems in promoting pathogenesis is clear but efflux systems are also common in plant symbiotic bacteria such as the nitrogen fixing rhizobia. Disruption of the multicomponent MFS efflux pump (rmrAB) in Rhizobium etli CFN42 increased phytoalexin sensitivity and led to impaired nodule formation. [13]. Likewise, nitrogen fixation was impaired and antimicrobial sensitivity increased when the RND family bdeAB efflux system was disrupted in the soybean symbiont Bradyrhizobium japonicum 110spc4 [14]. Nodulation, competiveness and toxin sensitivity were also affected when the RND smeAB pump was disrupted in Sinorhizobium meliloti 1021 [15].
We investigated the transcriptional responses of Rhizobium leguminosarum bv viciae 3841 to salicylic acid, coupled with a genome screen to identify putative multicomponent efflux systems. We investigated the contribution of these systems to salicylic acid resistance and their induction during plant colonisation and nodulation.

R. leguminosarum 3841 transcriptional responses to salicylic acid
In response to the addition of salicylic acid (0.72 mM) a total of 21 genes were up-regulated more than two fold (t-test p#0.05) compared to free living cells (Table 1). These responses can be broadly classified into those likely to be involved in salicylic acid export or its catabolism. Two genes RL1329 and RL1330 encoding putative efflux pump components were upregulated 18.1 and 2.2 fold respectively, with RL1329 being the most highly elevated gene in Rlv3841 in response to salicylic acid. Upstream of these genes is a putative LysR like transcriptional regulator, RL1328. It is proposed these genes be designated salR (RL1328), salA (RL1329) and salB (RL1330) (Figure 1). Based on homology with known proteins it is proposed salRAB is a MFS family multicomponent efflux system, where salA is a putative membrane fusion protein and salB is an inner membrane transporter. In addition rmrA (pRL90059), which has 85% identity to the membrane fusion protein of the characterised MFS family efflux pump in R. etli CFN42 [13], was also upregulated 4.1 fold.

Mutation of the sal genes
To determine if disruption of the salRAB operon in R. leguminosarum 3841 alters sensitivity to salicylic acid a salA deletion mutant was isolated (LMB519) and growth assayed at varying salicylic acid concentrations. In addition, as loss of rmrA has been shown to increase salicylic acid sensitivity in R. etli CFN42 [13], a Rlv3841 rmrA (pRL90059) deletion mutant previously isolated [16] was also tested. Salicylic acid (2 mM) significantly impaired growth of the salA mutant compared to controls (Figure 2A). In contrast disruption of rmrA led to no detectable difference in growth compared to wild type controls ( Figure 2B). Furthermore a double salA/rmrA (LMB523) mutant was not more sensitive to salicylic acid than the single salA mutant. In this instance the appropriate wild type control for comparison to the salA/rmrA (LMB523) mutant was RU4223 [16] which contains a VSp and pK19mob mutation in genes unrelated salRAB. This was so appropriate antibiotic selection could be used in the media and any differences in growth due to the presence of the spectinomycin cassette or pK19mob insertion could be discounted.
In order to complement the salA gene disruption the whole Sal operon (salRAB) was cloned into the stable low copy number plasmid pJP2 [17] forming pSal. When not induced with salicylic acid carriage of pSal had no effect on growth of either salA::VSp (LMB519) or WT control nifH::VSp (RU3940) compared to pJP2 parent plasmid containing strains (LMB641 and LMB640 respectively). A nifH mutant was used as a control instead of Rlv3841 because it enabled spectinomycin to be included in all media, ensuring growth differences were not due to the presence of a VSp cassette or spectinomycin. When induced with salicylic acid, instead of complementing the salA::VSp mutation (LMB519) pSal reduced growth relative to a strain (LMB641) containing the parent plasmid pJP2 alone. Similarly, pSal in the wild type control RU3940 (nifH::VSp) also reduced growth relative to pJP2 containing strain (LMB640) when induced with salicylic acid ( Figure 3A). Since salicylic acid is required for induction of the sal operon (see below) this suggests over-expression of the salAB operon from a low copy number plasmid is inhibitory and reduces growth. The problem of over expression of a membrane transporter leading to reduced growth is not uncommon. Since complementation proved problematic in order to confirm that the disruption of the salA was responsible for increased sensitivity to salicylic acid, independent pK19mob insertion mutants were isolated for each gene of the Sal operon. In the absence of salicylic acid there was no growth difference between wild type and mutant strains (Doubling times (h 6 standard error of the mean for three triplicates); RU4062 (nifH::pK19mob) 8.360.017, LMB455 (salR::pK19mob) 8.260.06, LMB415 (salA::pK19mob) 860.2, LMB409 (salB::pK19mob) 860.03). With the addition of 1.45 mM salicylic acid all three sal mutants had reduced growth compared to wild type strain (nifH::pK19mob) ( Figure 3B) as was observed with LMB519 (salA::VSp) (Figure 2A).

Plant symbiosis and antimicrobial sensitivity
The disruption of multicomponent efflux systems in other rhizobia has been shown to affect sensitivity to toxins, nodule formation and/or nitrogen fixation activity [13][14][15]. To ascertain if disruption of either the salRAB or rmrAB affected plant nodulation and nitrogen fixation the salA (LMB519), rmrA (RU4314) and double salA/salR (LMB523) mutants were inoculated on pea seeds. After 21 days growth the number of nodules were recorded and nitrogen fixing activity assessed by acetylene reduction assays. Compared to wild type none of the mutants differed in either the number of nodules formed (Rlv3841 In addition, the mutants did not have increased sensitivity to the antibiotics tetracycline and Nalidixic acid. It has been reported that disruption of efflux systems in other rhizobia, including rmrA mutant strains of CFN42, led to increased sensitivity to flavonoids and alkaloids [13,15]. However, neither Rlv3841 (WT), salA (LMB519), rmrA (RU4314) nor salA/rmrA (LMB523) strains of Rlv3841 showed impaired growth on agar plates with filter discs of genistein, naringenin or Berberine.

salRAB regulation
To examine the regulation of the salRAB operon two transcriptional reporter plasmids were constructed. Plasmid pLMB557 contains the putative intergenic promoter region between salR and salA (salAp) (Figure 1) upstream of a promoterless gfpmut3.1 reporter. The second plasmid (pLMB537) contains the promoter region as well as the complete salR gene (salAp and salR) ( Figure 1). The plasmids were introduced into the mutant strain LMB455 (salR::pK19mob) and Rlv3841. In the absence of salicylic acid there was no detectable induction of salA ( Figure 4). When Rlv3841 (WT) carrying pLMB537 (salAp and salR) or pLMB557 (salAp only) were incubated with 0.72 mM salicylic acid salA was induced ( Figure 4A). However, salA was not induced in the salR insertion mutant (LMB455) containing pLMB557 (salAp only) ( Figure 4B). This indicates that the salRAB operon is positively regulated by SalR. This was confirmed by restoration of salA induction in the salR insertion mutant (LMB455) when carrying pLMB537, which contains salAp and a full length copy of salR ( Figure 4B).

Induction specificity of the salRAB operon
It is proposed that salicylic acid is synthesised in plants via two enzymatic pathways from the primary metabolite chorismate, either via phenylalanine, cinnamate and benzoate intermediaries, or by a second pathway from isochorismate [18]. To test the specificity of the Sal operon several of these proposed intermediates and derivatives were used in induction assays with Rlv3841 containing the reporter plasmid pLMB557 (LMB475). These included cinnamate, benzoate and phenylalanine as well as the volatile ester of salicylic acid, methyl salicylate (wintergreen oil), a plant defence signalling molecule [19]. In addition catechol, phydroxybenzoate, protocatechuic acid and acetylsalicylic acid (aspirin) that have a similar structure to salicylic acid were also tested. Apart from salicylic acid itself all of these compounds failed to induce pLMB557, except for acetylsalicylic acid which showed some induction after 24 hours. This slow and weak induction may be the result of spontaneous hydrolysis of acetylsalicylic acid releasing salicylic acid.

Other multicomponent efflux systems of Rlv3841
To put the MFS type salRAB and rmrAB in context of the total multicomponent efflux systems encoded by Rlv3841, the genome was screened for proteins with sequence similarity to characterised systems. In total 17 systems were identified, including salRAB and rmrAB ( Table 2). Eleven of the systems belong to the RND family, four to the MFS family and two to the ABC families. Fifteen of the systems were encoded on the chromosome and one each on pRL9  and pRL10. In addition several TolC like homologs were also identified (RL3876, pR100291 and pR100178).
We have collected a large dataset (73 growth conditions) of the transcriptional response of Rlv3841 to different environments and inducers, most of which has been published previously, including colonisation of different plant rhizospheres [20] and in bacteroid development [16]. This allowed analysis of the expression of the efflux transporters when interacting with host plants, as well as in response to salicylic acid ( Figure 5). In early seven day bacteroid development many of the systems were highly upregulated compared to free living cells and in comparison to 21 day bacteroids ( Figure 5B). In total five systems show an above fivefold elevation at seven days with rmrA the highest at 27.5 fold up. In contrast salA has a modest 1.5 fold change. On colonisation of the rhizosphere of seven-day old pea, alfalfa and sugar beet ( Figure 5C) both rmrA and salA showed transcriptional increases of 5.5,3.5,2.5-and 1.8,2,2.7-fold, respectively. By far the largest increase at 135-fold was that of RL4274, a putative RND type efflux pump. How the efflux systems are differentially expressed when Rlv3841 colonises Pea plants of different ages is given in Figure 5D. Thus salRAB is only weakly induced in the plant rhizosphere and during nodule formation explaining the lack of a phenotype in a salA mutant.

Discussion
Multidrug efflux systems are important to plant pathogens and symbionts alike. Regardless of lifestyle soil organisms must overcome both abiotic and biotic stresses, including antimicrobials derived from plants and microbial competitors. In this study we investigated the global transcriptional response of Rlv3841 to salicylic acid, and identified two efflux systems significantly unregulated. The first of these was the hitherto uncharacterised salRAB and the second rmrAB, which has 85% (rmrA) and 89% (rrmB) amino acid identity to the characterised efflux system of R. etli CFN42 [13].
Disruption of salRAB in Rlv3841 led to significantly increased sensitivity to salicylic acid. However, it did not increase sensitivity to other antimicrobials tested or affect nodulation and nitrogen fixation. In R. etli CFN42, a close relative of Rlv3841 which also shows a high degree of sequence similarity to the salRAB system of Rlv3841 (RHE_CH01191, 90%; RHE_CH01192, 82% and RHE_CH01193 93% amino acid similarity to salRAB respec- tively), it has been demonstrated that disruption of rmrA leads to a 40% decrease in nodule number on bean (Phaseolus vulgaris) and five-fold increase in sensitivity to salicylic acid [13]. In contrast, disruption of rmrA in Rlv3841 did not increase sensitivity to salicylic acid or other antimicrobials. Furthermore, a double rmrA salA mutant did not have greater sensitivity to salicylic acid than the salA mutant. One explanation for the difference between the two organisms could be functional redundancy where efflux systems and/or other translocation mechanisms might overlap in substrate specificity. This is particularly true of soil organisms, which contain a disproportionally high number of efflux systems [21]. Thus it is possible the resilience of Rlv3841 to salicylic acid could suggest an as of yet unknown determinant that partially compensates the loss of both salRAB and rmrAB.
In addition to functional redundancy there are inherent difficulties in functional characterisation of efflux systems. Due to the wide diversity of compounds translocated by these systems, identifying the appropriate compounds is challenging. Efflux systems can also be host specific, for example disruption of the RND type efflux system bdeAB of Bradyrhizobium japonicum 110spc4 led to a decrease in the number of bacteroids in nodules and in nitrogen fixation when grown on soybean, but not on other plant hosts, mungbean and cowpea [14]. In addition, not all efflux may be responsible for removal of antimicrobials from cells, for example in Pseudomonas aeruginosa the MexAB-OprM system is involved in transport of compounds for quorum sensing [22], therefore their disruption is unlikely to affect antimicrobial resistance. In a recent study 14 multicomponent efflux systems were identified in S. meliloti 1021 via screening the genome for genes similar to known efflux pump components [15]. Each of these systems was deleted individually. Of the 14 mutants only one, smeAB, led to an increase in antimicrobial sensitivity and loss of competitive colonisation fitness compared to wild type. Double mutations of smeAB and systems smeCD and smeEF led to further antimicrobial sensitivity, however for 11 systems no difference in antimicrobial susceptibility was detected.
In this study we identified 17 putative efflux systems in Rlv3841 and interrogated transcriptional data previously produced from our lab to gain insight into the induction of these systems during rhizosphere colonisation, host infection and bacteroid differentiation during nodulation. In seven day bacteroids over one third of all the putative encoded efflux systems were induced above threefold, after 21 days only one system was induced above three-fold. The involvement of so many systems is testament to the complex and temporal physiological conditions encountered during nodule development. It is also possible that these systems overlap in specificity and thus offer an explanation why disruption of rmrA, the most uprgeulated system in seven-day bacteroids, had no effect on nodulation and nitrogen fixation. Furthermore, many of these systems may be expressed constitutively resulting in significant background resistance to antimicrobials.
Similarly, during plant colonisation of pea, alfalfa and sugarbeet a large number of different efflux systems were up regulated including salRAB and rmrAB. In addition most respond similarly to the different rhizospheres i.e. are generalists. Although, one system (RND type RL4274/4275), while upregulated in all three rhizospheres, was most highly induced by pea ( Figure 5C), suggesting specific induction for this system. Indeed in a previous study mutation of RL4274 decreased the competiveness compared to Rlv3841 in the pea rhizosphere [20].
In this study we have demonstrated that salAB is positively regulated by a LysR family transcriptional type regulator SalR. Induction of this operon is highly specific to salicylic acid. Salicylic acid is known to be instrumental in nodule development such that in Rhizobium leguminosarum bv viciae strains RBL 5523 and 248 exogenous application of 10 24 M salicylic acid completely inhibited nodule formation on vetch [8]. Moreover, Stacey et al., 2006 [6] demonstrated that reducing the endogenous levels of salicylic acid by transgenic expression of salicylate hydroxylase (NahG) in Lotus japonicus correlated with an increase in nodule number when inoculated with Mesorhizobium loti. As salRAB confers increased resistance at high levels of salicylic acid (above 1.45 mM) it can be hypothesised salRAB confers a fitness advantage to Rlv3841 by eliminating the inhibitory effects of salicylic acid through expulsion from the cell. However, the Sal system was only weakly expressed in the rhizosphere and in nodule bacteria explaining the lack of effect of mutation in salA on nodulation and N 2 -fixation.

Strains, plasmids and culture conditions
The strains and plasmids used in this study are detailed in Table 3. All Escherichia coli strains were grown at 37uC in Lennox (L) broth or L agar. R. leguminosarum strains were grown at 28uC in Tryptone-yeast extract [23] or Acid Minimal Salts (AMS) [24] supplemented with 30 mM Pyruvate and 10 mM ammonium chloride and agitated at 220 rpm. Antibiotics were used in the following concentrations (mg ml 21 ) unless otherwise stated. Gentamicin, 20 (10 in E. coli); Kanamycin, 20; Neomycin, 20; Spectinomycin, 50; Streptomycin 500; Tetracycline (2 in AMS, 5 in TY).

Molecular Microbiology
All general DNA cloning was as described [25]. PCR amplification, unless otherwise stated, was performed in 100 ml volumes using 2.5 units GoTaq (Promega), 1 mM primer and 0.2 mM dNTPs. Cycling conditions were: One cycle of 95uC for 2 minutes, 30 cycles of 95uC for 45 s, 57uC for 45 s, 72uC for 3 minutes with a final extension of 10 minutes at 72uC. All constructs were confirmed using Sanger sequencing. Plasmids were conjugated from E. coli into R. leguminosarum strains using the helper plasmid pRK2013 as previously described [26]. PCR primers used in this study are given in Table 4.

RNA isolation and microarray analysis
R. leguminosarum 3841 was grown in triplicate in 100 ml AMS to mid-log growth phase before the cells where harvested and washed twice in AMS and resuspended in AMS with or without 0.72 mM salicylic acid (i.e. 100 mg ml 21 ). Cultures were induced for three hours before RNA was extracted, amplified and hybridised as previously described [16]. Microarray data are available in the ArrayExpress database (www.ebi.ac.uk/ arrayexpress) under accession number E-MTAB-2187.  Table 2. Fold changes are given for the membrane fusion protein (MFP) of each efflux system. A, Addition of 0.72 mM salicylic acid. B, Bacteroids isolated from 7 and 21 day nodules [16]. C, Rlv3841 isolated 7 day post inoculation from 7 day old pea (Pisum sativum), alfalfa (Medicago sativa) and sugar beet (Beta vulgaris) after 7 day post inoculation [20]. D, Rlv3841 isolated 1 day post inoculation from 7, 14 and 21 day old peas [20]. doi:10.1371/journal.pone.0103647.g005

Induction reporter assays
Induction assays were performed in 100 ml volumes. Cells were grown to mid log growth phase before candidate inducers were added to the media to a final concentration of 0.72 mM. Assays were performed in triplicate and GFP fluorescence was detected with a Tecan GENios fluorometer (excitation 485 nm, emission 510 nm).
To construct the complementing plasmid pSal, salRAB was amplified using primers pr1394/pr1395 (containing KpnI and BamHI respectively on the 59 ends) and blunt cloned into pJET1.2 blunt (Thermo Scientific). Digestion with KpnI/BamHI enabled ligation into KpnI/BamHI digested pJP2. Transcriptional reporter plasmids pLMB557 (containing the salR-salA intergenic region, salAp) and pLMB537 (containing salAp as well as salR) were isolated by PCR amplification of fragments from Rlv3841 using primers E6_P_for/E6_P_rev and E6_RL1328_for/E6_RL1328_rev respectively. HindIII and BamHI sequences to the 59 end of the forward and reverse primers enabled ligation of the fragments into HindIII/BamHI digested pRU1097. The correct insertions were confirmed by sequencing and conjugated into R. leguminosarum strains.

Sensitivity assays
Cells were harvested from TY slopes and adjusted to an OD 600 of 0.2. Aliquots of 100 ml were mixed into 3 ml soft TY agar (0.7%) and overlaid on TY plates. Sterile filter discs (6 mm diameter) were placed on the top agar and 5 ml of the test compound was added per disc. A two fold dilution series was used for each compound giving test concentration ranges of; Tetracycline (0.31-5 mg ml 21 ), Nalidixic acid (0.13-2 mg ml 21 ), Naringenin (0.013-5 mg ml 21 ), Geinstein (0.013-5 mg ml 21 ), Berberine (0.31-5 mg ml 21 ). Each compound at each different concentration was tested in triplicate. After 72 hours growth at 28uC the size of the zone of inhibition was measured.

Plant growth conditions and acetylene reduction assays
Pea seeds (Pisum sativum cv Avola) were surface sterilised and grown in one litre pots filled with autoclaved vermiculite containing 400 ml of N-free rooting solution [24]. Plants were inoculated with 10 6 c.f.u. and grown for 21 days in a controlled environment at 22uC under sonti-agro lights with a 16 : 8 h Day : Night cycle. In total there were 10 plant replicates for each strain. At harvest nodules were counted and acetylene reductions performed as previously described [30]. For each strain tested, 12 nodules were picked, surface sterilised, crushed and plated on TY medium plates. For each plate 10 individual colonies were subcultured onto Neomycin (80 mg ml 21 ) and or Spectinomycin (100 mg ml 21 ) plates to confirm the mutant genotype.
Identification of putative multicomponent efflux pumps