Salicylic acid, a versatile molecule

Each year approximately 40,000 t of aspirin, the acetylated form of the phenolic compound salicylic acid (SA), is produced (Fuster and Sweeny 2011). Acetylsalicylic acid is the most widely used drug worldwide. The history of SA as a cure for aches and fevers goes back thousands of years (Jack 1997). Around 4000 BC the Assyrians used extracts of willow leaves to combat joint pain (Mahdi 2010). The bark of willow contains relatively large quantities of SA. Beavers consume willow bark and the secretion from the castor sac, the SA-containing castoreum, was prescribed for many maladies until the 1700s (Müller-Schwarze and Sun 2003). In animal cells, anti-inflammatory properties of aspirin and SA are partly mediated by inhibition of IκB kinase-β activity (Yin et al. 1998). Aspirin also appears to prevent recurrent cardiovascular events and to reduce incidence of colorectal cancer; however, the underlying mechanisms need further investigations (Fuster and Sweeny 2011).

In plants SA plays a crucial role in responses to pathogen attack (Vlot et al. 2009), but also has a role in responses to abiotic stresses and in growth and development (Rivas-San Vicente and Plasencia 2011). The recently published book edited by Hayat et al. (2013) offers an extensive overview of current knowledge on SA and plant growth and development. The presence of SA in plants is well documented and for some plant species levels in excess of 1 μg per gram fresh weight have been reported (Raskin 1992). SA is required as a signaling molecule in systemic acquired resistance (SAR) that develops in plants after attack by pathogens that cause necrosis (Durrant and Dong 2004). SAR is effective against a broad range of pathogens and the protection can be long-lasting. Manipulation of this induced defense mechanism thus has potential for plant protection. Exogenous application of SA or SA mimics can indeed protect plants against a range of pathogens (Oostendorp et al. 2001). Whereas animals and plants both respond to application of SA, effects on fungi have hardly been studied. In the white button mushroom Agaricus bisporus application of SA did not result in protection against the fungal pathogen Lecanicillium fungicola (Berendsen et al. 2013), suggesting that SA inducible defense pathways are lacking or ineffective.

Production of SA by bacteria has been reported frequently, in many cases related to the production of siderophores under iron limited conditions. Next to playing a role in iron acquisition the effects of SA on plants suggest that the production of this metabolite by bacteria can have a significant impact on plant-microbe interactions in the rhizosphere. However, as yet there is no convincing evidence for such a role of SA. Here we review both the genetics and the influence of physiological conditions on SA biosynthesis in bacteria. Postulated implications of bacterial SA production in induced disease resistance and impact of SA on bacterial ecology are discussed in detail.

Genetics of SA biosynthesis in bacteria

Biosynthesis of SA (2-hydroxybenzoic acid) may proceed either from the phenylpropanoid or the chorismate/isochorismate pathways. In plants, SA can be synthesized from any of these two major enzymatic routes (Lee et al. 1995; Verberne et al. 1999; Wildermuth et al. 2001; Garcion and Métraux 2006; Wildermuth 2006; Chen et al. 2009; Vlot et al. 2009; Dempsey et al. 2011). In SA-producing bacteria, however, the chorismate/isochorismate route is the unique biosynthetic pathway so far identified (Kerbarh et al. 2005).

Production of SA by bacteria is linked to their growth under iron-limiting conditions. SA has been suggested to act as an endogenous siderophore under iron limitation in Mycobacterium spp. (Ratledge and Winder 1962), Pseudomonas spp. (Ankenbauer and Cox 1988; Anthoni et al. 1995; Meyer et al. 1992; Visca et al. 1993), Azospirillum lipoferum (Saxena et al. 1986), and Pseudomonas cepacia (Sokol et al. 1992). However, the role of SA as a siderophore per se was questioned by Chipperfield and Ratledge (2000) who suggested that SA cannot bind Fe3+ at pH above 6.

In general synthesis of SA in bacteria is concomitant to the biosynthesis of SA-based siderophores. SA is an essential precursor or intermediate in the synthesis of the bacterial siderophores mycobactin T in Mycobacterium tuberculosis (Snow 1965), mycobactin S in Mycobacterium smegmatis (Hudson and Bentley 1970; Ratledge and Hall 1972), parabactin in Paracoccus denitrificans (Person and Neilands 1979), pyochelin in Pseudomonas aeruginosa (Cox et al. 1981; Ankenbauer and Cox 1988; Brandel et al. 2012), Pseudomonas fluorescens (Castignetti 1997), Burkholderia arboris (Le Dang et al. 2011) and Burkholderia cenocepacia (Farmer and Thomas 2004), maduraferrin in Acinomadura madurae (Keller-Schierlein et al. 1988), vulnibactin in Vibrio vulnificus (Okujo et al. 1994), aeruginoic acid in P. fluorescens (Carmi et al. 1994), pseudomonine (Psm) in P. fluorescens (Anthoni et al. 1995; Mercado-Blanco et al. 2001) and Pseudomonas entomophila (Matthijs et al. 2009), and yersiniabactin in Yersinia spp. (Drechsel et al. 1995; Chambers et al. 1996; Pelludat et al. 2003; Bultreys et al. 2006) and Pseudomonas syringae pv. tomato (Jones et al. 2007). Therefore, genetics and regulation of SA biosynthesis must be analyzed along with the machinery leading to siderophores having SA as an essential moiety.

The evidence that bacterially-produced SA proceeds via the shikimate pathway came from the pioneering work of Ratledge (1969). In his experiments, high level of [14C7]-shikimic acid was incorporated into SA in M. smegmatis, both alone and in the presence of unlabeled co-substrates (i.e. L-phenylalanine, L-tyrosine, p-aminobenzoic, etc.). Hudson and Bentley (1970) later demonstrated that all seven carbon atoms of shikimic acid are incorporated into SA upon feeding cultures of M. smegmatis with labelled shikimic acid. Then, Marshall and Ratledge (1971) showed that cell extracts of this bacterium were able to convert chorismic acid to isochorismic acid and subsequently to SA. Thus, synthesis of SA took place only with isochorismic acid as substrate, and NAD+ was not required. Finally, Marshall and Ratledge (1972) proposed that SA is synthesized from chorismic acid with isochorismate as the only intermediate, shikimic acid being the precursor of the former. This route was verified not only in M. smegmatis, but also in M. tuberculosis and Mycobacterium fortuitum, and the trivial name of salicylate synthase was suggested for the final enzyme of the pathway (Marshall and Ratledge 1972).

The genetic basis of salicylate biosynthesis in bacteria via the chorismate/isochorismate pathway was first established for P. aeruginosa PAO1 by Serino et al. (1995). SA is a precursor of the siderophore pyochelin in strain PAO1 (Ankenbauer and Cox 1988). Two adjacent genes involved in SA synthesis were cloned and sequenced: pchB (coding for salicylate synthase) and pchA (coding for an isochorismate synthase, ICS) (Gaille et al. 2003). Expression of the pchB gene in an E. coli entB (coding for isochorismatase) mutant resulted in SA production. In addition, when an E. coli entC (ICS) mutant was transformed with pchBA genes from strain PAO1, it became an SA producer (Serino et al. 1995). The pchB product is an isochorismate pyruvate lyase (IPL), proposed to also act as a functional chorismate mutase (Gaille et al. 2002), a suggestion confirmed by Künzler et al. (2005). While SA synthesis by strain PAO1 occurs in a two-step reaction, other SA-producing bacteria harbour the genetic/enzymatic machinery for the direct conversion of chorismate to salicylate. Thus, MbtI of M. tuberculosis (Harrison et al. 2006), YbtS of Yersinia pestis (Gehring et al. 1998), or Irp9 of Yersinia enterocolitica (Pelludat et al. 2003; Kerbarh et al. 2005) are true salicylate synthases, capable of ring isomerization and pyruvate lyase activities in a single enzyme, converting chorismate directly into SA (Pelludat et al. 2003). These proteins are smaller than PchA and bacteria harbouring them do not have pchB homologs nearby ICS-coding genes (Gehring et al. 1998; Quadri et al. 1998).

The two-enzyme process characterized for P. aeruginosa (Serino et al. 1995; Gaille et al. 2003) has been confirmed for other pseudomonads. The first step, chorismate to isochorismate, catalysed by an ICS is encoded by pmsC in P. fluorescens WCS374 (Mercado-Blanco et al. 2001) and P. entomophila L48 (Matthijs et al. 2009), both homologous to the product of pchA in P. aeruginosa PAO1. The second step, conversion of isochorismate to SA, is catalysed by an IPL (Sattely and Walsh 2008; Wuest et al. 2009) encoded by pmsB in strains WCS374 (Mercado-Blanco et al. 2001) and L48 (Matthijs et al. 2009) and homologous to pchB in strain PAO1. Similar to the pyochelin and SA biosynthetic gene cluster in P. aeruginosa PAO1 (Serino et al. 1995, 1997), pmsB and pmsC of strains WCS374 and L48 are part of the operon pmsCEAB involved in the biosynthesis of SA and the siderophore Psm (Mercado-Blanco et al. 2001; Matthijs et al. 2009). Heterologous expression of pmsB and pmsC was accomplished in Pseudomonas putida and Escherichia coli cells. Deletions affecting pmsC reduced SA production in E. coli, whereas deletion of pmsB enterely abolished it (Mercado-Blanco et al. 2001). Final evidence of the involvement of pmsB as the key gene for SA biosynthesis in strain WCS374 came after marker-exchange mutant analysis (Djavaheri et al. 2012). Remarkably, the Psm biosynthesis gene cluster as well as genes putatively involved in the transport of this siderophore have identical organization in P. fluorescens WCS374, a rhizosphere biocontrol strain (Djavaheri et al. 2012), and P. entomophila L48, an entomopathogenic bacterium (Matthijs et al. 2009). Recently, the gene cluster for the biosynthesis and uptake of Psm (or a Psm-like compound) has been reported to be present in the genomes of the biocontrol strains Pseudomonas spp. BG33R and P. fluorescens A506. However, whether this secondary siderophore is biosynthesized by these strains has yet to be confirmed (Loper et al. 2012).

Environmental and physiological factors influencing bacterial SA biosynthesis

As already mentioned, bacterial SA biosynthesis is linked to the assembly line of SA-based siderophores. Considering that siderophores are produced to cope with Fe3+ limiting conditions, the major factor influencing biosynthesis of SA and SA-based siderophores is Fe3+ availability (Ratledge and Winder 1962; Cox and Graham 1979). Indeed, SA concentration increases when available iron in the medium is low (Ratledge and Hall 1971; Leeman et al. 1996).

The promoter of the pchDCBA operon, involved in the biosynthesis of SA and pyochelin in P. aeruginosa PAO1 (Serino et al. 1995, 1997), is positively controlled by the PchR protein in the presence of pyochelin (Reimmann et al. 1998) and negatively by the Fur (Ferric Uptake Regulation) repressor (Bagg and Neilands 1987; de Lorenzo et al. 1987) in the presence of iron (Serino et al. 1997). Iron availability, and pyochelin acting as an autoinducer, are major signals determining the yield of the pyochelin pathway by regulating the expression of the pchA (ICS) gene (Gaille et al. 2003). The PchA concentration was proposed as an important factor influencing this siderophore biosynthetic pathway: pyochelin acting as a positive signal and iron as the negative one (Gaille et al. 2003). Whether this scenario can also be portrayed for other SA-based siderophores such as Psm is not known. Nevertheless, presence of two potential Fur boxes in the promoter region of the pmsCEAB operon as well as repression of this gene cluster by iron has been confirmed (Mercado-Blanco et al. 2001). Availability of iron and diverse substrates, as well as temperature affect bacterial production of SA and SA-based siderophores (Leeman et al. 1996; De Meyer and Höfte 1997; Press et al. 1997; Audenaert et al. 2002; Ran et al. 2005a; Djavaheri et al. 2012). Despite that production of SA by bacteria can be easily confirmed in vitro, SA synthesis in vivo, in the ecological niche where they naturally live or can be artificially introduced, is more difficult to assess. For instance, while production of SA by strain 7NSK2 is measurable in vitro (De Meyer and Höfte 1997), production in vivo (tomato roots) could not be demonstrated (Audenaert et al. 2002). This could be due to the detection limit of the methodological approach used or to the possibility that SA could be entirely channeled to pyochelin synthesis when bacteria are grown in a natural environment. Indeed, these authors have demonstrated that when strain 7NSK2 is grown in the presence of L-cysteine, production of pyochelin, composed of one molecule of SA and two molecules of cysteine, is enhanced (Cox et al. 1981). Since cysteine is a component of plant roots exudates, including tomato (Solanum lycopersicum Mill.) (Gamliel and Katan 1992), it is plausible to assume that SA is incorporated into pyochelin in the tomato rhizosphere (Audenaert et al. 2002). This could also operate for P. fluorescens WCS374, which produces relatively high amounts of SA in vitro (see below) (Leeman et al. 1996; Mercado-Blanco et al. 2001; Ran et al. 2005a; Djavaheri et al. 2012). However, in standard succinate medium (SSM) (Meyer and Abdallah 1978), that is typically used for SA determination, L-histidine and L-threonine, the other two building blocks required for Psm biosynthesis (Mercado-Blanco et al. 2001; Sattely and Walsh 2008; Djavaheri et al. 2012) are lacking. Thus similar to the situation for strain 7NSK2, SA production by WCS374 in the rhizosphere may be strongly reduced, or completely abolished, due to the presence of histidine and threonine in the root exudates. Excretion of SA has also been reported in M. smegmatis as an early metabolic event which is then followed by a re-absorption by the cells to be incorporated into the mycobactin biosynthetic route (Hudson and Bentley 1970; Marshall and Ratledge 1972).

Production of SA is also influenced by temperature. Ran et al. (2005a) reported that the relatively high production of SA in vitro by strain WCS374 is even enhanced at supra-optimal temperatures (i.e. 31–33 ºC). This increase was also observed for P. fluorescens CHA0r, but not for P. fluorescens WCS417r or P. aeruginosa 7NSK2 (Ran et al. 2005a).

Production of SA by rhizosphere bacteria in situ is difficult to confirm. So is the unravelling of its potential function(s) in a dynamic environment where multiple trophic interactions take place which, in addition, are influenced by diverse physicochemical factors. SA is produced at iron limiting conditions, a situation commonly found in the rhizosphere (Loper and Henkels 1997; Duijff et al. 1999). While this primary condition induces SA biosynthesis, its functioning as a true siderophore would only be expected at the specific situation of pH < 6 (Chipperfield and Ratledge 2000). Considering that pH in the rhizosphere may rapidly change due to diverse factors (Hinsinger et al. 2003), bacterial SA working as a siderophore in the rhizosphere or in the root interior seems to be a highly-conditioned event. Moreover, presence of root exudates at specific rhizosphere spots would provide important substrates to bacteria needed for siderophore biosynthesis, thus favouring the channeling of free SA towards SA-based siderophore assembly. On the other hand, some bacteria can overproduce SA at high temperatures, pointing to scenarios (i.e. high soil temperature during hot summers and/or tropical areas) where biosynthesis of SA and specific SA-based siderophores could be enhanced, triggering responses in plants which otherwise would remain silent under low-temperature conditions. Finally, SA may pose ecological advantages to SA-producing bacteria, due to its antibacterial and antifungal properties (Gershon and Parmegiani 1962; Himejina and Kubo 1991). Whereas the potential of SA as an antibiotic in the rhizosphere has hardly been investigated, production of a salicylate containing antibiotic, promysalin, by P. putida RW10S1 has been described (Li et al. 2011).

Changes in the production of SA and SA-based siderophores will largely depend on the dominating environmental factors at any given moment. In fact, the same SA-based siderophore can be produced under different trophic scenarios, for instance by saprophytic (i.e. mycobactin in M. smegmatis or Psm in P. fluorescens AH2) (Anthoni et al. 1995; Adilakshmi et al. 2000), entomopathogenic (Psm in P. entomophila L48) (Matthijs et al. 2009), opportunistic pathogenic (i.e. vulnibactin in V. vulnificus or pyochelin in P. aeruginosa) (Okujo et al. 1994; Serino et al. 1997), animal pathogenic (i.e. mycobactin in M. tuberculosis or yersiniabactin in Y. enterocolitica) (Pelludat et al. 2003; Harrison et al. 2006), plant pathogenic (i.e. yersiniabactin in P. syringae pv. tomato DC3000) (Jones et al. 2007), or plant beneficial (i.e. Psm in P. fluorescens WCS374) (Mercado-Blanco et al. 2001) bacteria. In pseudomonads, the production of SA-containing siderophores may be an adaptation to environmental conditions under which the production of their fluorescent siderophore pyoverdine is shut down, for example high temperatures.

SA production by plant growth promoting rhizobacteria

Plant growth-promoting rhizobacteria (PGPR) are root-colonizing bacteria that exert beneficial effects on plant growth (Bakker et al. 2007b). Several PGPR strains have been reported to have the ability to produce SA (Table 1). Here we summarize well documented examples of SA production by PGPR.

Table 1 SA producing plant growth-promoting rhizobacteria strains

Pseudomonas is the best studied genus for SA production and SA-producing species include P. aeruginosa, Pseudomonas aureofaciens, Pseudomonas corrugata and P. fluorescens (Table 1). P. fluorescens holds the largest number of SA-producers studied, including strains WCS374 and WCS417 (Geels and Schippers 1983; Lamers et al. 1988), CHA0 (Stutz et al. 1986), Pf4–92, Pf12–94, Pf151–94 and Pf179–94 (Saikia et al. 2003, 2005), or PICF3, PICF4 and PICF7 (Mercado-Blanco et al. 2004). The latter strains were reported to produce only minor amounts of SA when grown in SSM (Mercado-Blanco et al. 2004). Among them, the olive (Olea europaea L.) root endophytic strain PICF7 (Prieto and Mercado-Blanco 2008) has been shown to be an efficient biological control agent (BCA) against Verticillium wilt of olive (Verticillium dahliae) (López-Escudero and Mercado-Blanco 2011; Prieto et al. 2009), and able to trigger a broad range of defense responses in olive root tissues (Schilirò et al. 2012).

P. fluorescens strains WCS374 and WCS417, isolated from respectively potato (Solanum tuberosum L.) and wheat (Triticum aestivum L.) rhizospheres, have been investigated for plant growth promotion and biocontrol activities in several plant species and against diverse pathogens (Bakker et al. 2007a; De Vleesschauwer and Höfte 2009; Van Loon and Bakker 2005). In vitro production of SA by strains WCS374 and WCS417 was measured in SSM with low iron availability. Strain WCS374 can be considered as a SA ‘super-producer’ (up to 55 μg per ml) in vitro. This amount is approximately 10 times higher than that detected for WCS417 as well as for other SA producers under similar culturing conditions (Leeman et al. 1996; Mercado-Blanco et al. 2001; Ran et al. 2005a).

P. fluorescens CHA0, a well-studied PGPR strain, was originally isolated from roots of tobacco (Nicotiana tabacum L.) plants grown in soil naturally suppressive to black root rot, a disease caused by Thielaviopsis basicola (Stutz et al. 1986). For this strain in vitro SA production was detected under low iron conditions (Reimmann et al. 1997), and effects of carbon sources and minerals on production have been investigated (Duffy and Défago 1999). At elevated temperature, SA production by CHA0 is enhanced (Ran et al. 2005a).

Besides SA-producing rhizosphere pseudomonads, other bacterial genera have been demonstrated to produce SA. Achromobacter sp. SF2 and Bacillus spp. SF3 and SF4, originating from sunflower (Helianthus annuus L.) roots, can enhance growth of sunflower seedlings under water stress. Moreover, these bacteria produced SA and, interestingly, for all three strains SA production was significantly increased under water stress (Forchetti et al. 2010). Serratia marcescens strain 90–166 was also characterized as an SA-producing rhizobacterium (Press et al. 1997). SA biosynthesis by strain 90–166 is affected by the culture medium with the highest production in Kings medium B that has low iron availability (Zhang et al. 2002).

Whereas measuring bacterial SA production in vitro is rather straightforward, detection of bacterial SA in the rhizosphere is challenging. On cucumber roots colonized by P. aureofaciens 63–28 or P. corrugata 13, Chen et al. (1999) measured elevated levels of SA as compared to control roots. However, the elevated levels were magnitudes higher than those produced by the bacteria in vitro, and the authors concluded that the bacteria stimulated the plant itself to accumulate SA (Chen et al. 1999). For the SA and pyochelin, a siderophore containing a SA moiety, producing P. aeruginosa strain 7NSK2 (Buysens et al. 1996; De Meyer and Höfte 1997), elevated levels of SA on tomato roots colonized by the wild type strain could not be detected (Audenaert et al. 2002), suggesting that the bacteria do not produce SA in the rhizosphere. However, a mutant of 7NSK2 that cannot produce pyochelin but that does produce SA, appeared to produce significant amounts of SA in the rhizosphere (Audenaert et al. 2002).

In summary, many PGPR have the ability to produce SA in an iron availability dependent way and SA is detected on plant roots (Hayat et al. 2013), although likely originating from plant root tissues upon interaction with rhizobacteria. Given the fact that several PGPR can elicit induced systemic resistance (ISR) in plants (Van Loon et al. 1998; Kloepper et al. 2004; Bakker et al. 2003, 2007a) and that application of SA to plants leads to induced resistance against a range of pathogens (White 1979; Oostendorp et al. 2001; An and Mou 2011), attempts to study the involvement of bacterial SA in PGPR-mediated ISR are reviewed below.

Are SA and SA-based siderophores involved in induced resistance?

The involvement of SA and SA-based siderophores, i.e. Psm and pyochelin, in PGPR-mediated ISR has been studied using knock-out mutants of bacteria that no longer produce these metabolites. Another approach has been to use mutant and transgenic plants that are defective in the SA signaling pathway, for example NahG-transformed plants that cannot accumulate SA, thereby unable to express induced resistance (Gaffney et al. 1993).

P. fluorescens WCS374 and WCS417 and P. putida WCS358 can all suppress Fusarium wilt, caused by Fusarium oxysporum f. sp. raphani (For), in radish (Raphanus sativus L.) (Leeman et al. 1996). Whereas WCS374 and WCS417 can elicit ISR in this pathosystem, P. putida WCS358 cannot and this strain suppresses disease by siderophore-mediated competition for iron (Leeman et al. 1996; Raaijmakers et al. 1995). Moreover, ISR by the P. fluorescens strains is enhanced under iron-limiting conditions, a situation favoring production of iron-regulated metabolites including SA. Since strain WCS358 (a non SA producer) does not elicit ISR in radish, Leeman et al. (1996) suggested that rhizobacteria-produced SA is essential for ISR. This suggestion was sustained by the fact that application of as low as 100 fg of SA to radish roots induced resistance against For. However, in the model system Arabidopsis thaliana-Pseudomonas syringae pv. tomato (Pst) WCS358 can elicit ISR, so can WCS417 (a moderate SA producer), whereas WCS374 (a SA ‘super producer’) cannot (Van Wees et al. 1997). Moreover, ISR by WCS417 was still effective in NahG-transformed Arabidopsis plants in which exogenous application of SA does not induce resistance (Pieterse et al. 1996). These results argue against SA being involved in ISR by the SA-producing strain WCS417.

WCS374 is an intriguing fluorescent pseudomonad strain because of its ability to produce large amount of SA in vitro (see above). This strain can trigger ISR against Fusarium in radish (Leeman et al. 1996), and against Ralstonia solanacearum in Eucalyptus (Eucalyptus urophylla S. T. Blake) (Ran et al. 2005b), but appeared to be unable to induce resistance against Pst in A. thaliana (Ran et al. 2005a; Van Wees et al. 1997). However, when applied at low population densities, P. fluorescens WCS374r did elicit ISR in A. thaliana against the bacterial pathogen Pst, with no requirement of SA or siderophore production (Djavaheri et al. 2012). Also, WCS374r-mediated ISR against the fungus Magnaporthe oryzae in rice (Oryza sativa L.) does not require bacterial SA production (De Vleesschauwer et al. 2008). Overall, these results suggest that involvement of WCS374-produced SA in ISR is unlikely.

Besides SA, strain WCS374 produces the SA-based siderophore Psm under iron-limited conditions (Djavaheri et al. 2012; Mercado-Blanco et al. 2001). A possible role of Psm in ISR in the pathosystem Eucalyptus-R. solanacearum was investigated using transformant JM218(pMB374-07), a pyoverdin mutant of WCS358 (Marugg et al. 1985) that is transformed with the Psm biosynthesis genes of WCS374 (Mercado-Blanco et al. 2001). Reduction of bacterial wilt was not observed upon treatment with JM218(pMB374-07), discarding Psm as an ISR triggering factor in this system (Ran et al. 2005b). Moreover, the possible role of Psm in ISR was evaluated in the rice-M. oryzae pathosystem by using mutants of WCS374 deficient in Psm production (De Vleesschauwer et al. 2008). It was concluded that Psm is not essential for ISR against the leaf blast pathogen in rice (De Vleesschauwer et al. 2008). Finally, the potential involvement of Psm in ISR against Pst was also tested in Arabidopsis. The Pms mutants behaved similar to the wild type in their ability to elicit ISR against Pst, again suggesting that Psm production by WCS374r is not required for eliciting ISR in Arabidopsis (Djavaheri et al. 2012).

P. aeruginosa strain 7NSK2 can induce resistance in several plant species, including bean, rice, tobacco and tomato (De Vleesschauwer and Höfte 2009). The ability of 7NSK2 to elicit ISR appeared to be linked to SA production by the bacteria (De Meyer and Höfte 1997; De Meyer et al. 1999b), and nanogram amounts of SA produced by this strain appeared to be sufficient (De Meyer et al. 1999a). Audenaert et al. (2002) postulated that SA is not the inducing compound of 7NSK2, but ISR depends on the production of both pyochelin and pyocyanin. Apparently in the rhizosphere, 7NSK2 does not produce SA, but instead the SA is channeled into the SA-containing siderophore pyochelin (Audenaert et al. 2002). However, a mutant of 7NSK2 that cannot produce pyochelin, but that does produce SA (mutant KMPCH) was demonstrated to elicit induced resistance through the production of SA (Audenaert et al. 2002).

Pseudomonas fluorescens strain CHA0 is a SA producer under iron-starvation conditions (Meyer et al. 1992) and it can induce systemic resistance against Tobacco necrosis virus (TNV) (Maurhofer et al. 1994). CHA400, a pyoverdin-negative but SA-positive mutant of CHA0, partially maintained the ability to induce resistance against TNV. Furthermore, introduction of the SA biosynthetic genes pchBA from P. aeruginosa PAO1 (Serino et al. 1997) into P. fluorescens strain P3, which does not produce SA, rendered this strain capable of SA production in vitro. Furthermore, SA production by P3 significantly improved its ability to induce systemic resistance in tobacco against TNV (Maurhofer et al. 1998).

To further understand the role of bacterially-produced SA, several of the above-mentioned SA-producing Pseudomonas strains, including P. fluorescens WCS374r, WCS417r, CHA0r, and P. aeruginosa 7NSK2, were compared for their abilities to induce systemic resistance in A. thaliana against bacterial speck caused by Pst (Ran et al. 2005a). The SA ‘superproducer’ P. fluorescens WCS374r could not induced resistance in A. thaliana, whereas strains WCS417r, CHA0r, and 7NSK2 could, and this phenotype was apparently associated with their ability to produce SA. Conversely, a SA mutant of 7NSK2, MPFM1-569, still was able to trigger ISR. The possible involvement of SA in the induction of resistance was also evaluated using non SA-accumulating NahG plants, transformants that express the bacterial NahG gene that encodes salicylate hydroxylase converting SA into cathechol. Strains WCS417r, CHA0r, and 7NSK2 all induced resistance in NahG Arabidopsis. Likewise, WCS374r triggered ISR in these plants when grown at 33 or 36 °C. Thus, even though WCS374r can be manipulated to elicit ISR against bacterial speck in Arabidopsis, SA is not the primary determinant. Also for the other SA-producing strains, WCS417r, CHA0r and 7NSK2, bacterial determinants other than SA must be responsible for inducing resistance (Ran et al. 2005a).

SA-producing rhizobacterial strain S. marcescens 90166 induces resistance against P. syringae pv. tabaci in both wild-type and NahG tobacco plants (Press et al. 1997). Moreover, a SA mutant retained ISR-eliciting activity in cucumber against Colletotrichum orbiculare, and an ISR mutant, 90-166-2882, still produced SA. Also in tobacco against blue mold disease caused by Peronospora tabacina S. marcescens 90166 significantly reduced disease severity in both wild type and NahG tobacco, indicating that systemically-induced resistance in tobacco to blue mold by strain 90–166 is SA-independent (Zhang et al. 2002). In similar experiments it was demonstrated that protection of Arabidopsis against CMV by strain 90–166 is independent of SA production (Ryu et al. 2004). In summary, SA produced by strain 90–166 does not appear to be involved in the resistance induced by this strain in Arabidopsis, tobacco and cucumber (Press et al. 1997; Ryu et al. 2004; Zhang et al. 2002).

In conclusion, while PGPR strains producing SA in vitro have been identified frequently and the involvement of SA in both root colonization and elicitation of ISR response has been investigated, a clear role for PGPR-produced SA in ISR has not been established. In most cases the bacteria do not seem to produce SA in the rhizosphere, but instead they produce SA-containing siderophores (Fig. 1a). However, when SA is delivered in the rhizosphere by modified bacteria, like mutant KMPCH of P. aeruginosa 7NSK2 (Audenaert et al. 2002) or the SA producing transformant of P. fluorescens P3 (Maurhofer et al. 1998), it is bioactive in eliciting induced resistance. Moreover, root colonization by SA producing pseudomonads can elicit endogenous SA production in the root (Chen et al. 1999) (Fig. 1b). Thus, the presence of SA in the rhizosphere must be anticipated. Indeed, SA has been detected in root exudates of for example Arabidopsis (Badri et al. 2013), sugar beet (Khorassani et al. 2011), sunflower (Park et al. 1992) and watermelon (Ling et al. 2013). It thus appears that studies on SA in rhizosphere microbiology should not be restricted to SA produced by bacteria.

Fig. 1
figure 1

a Although many bacteria that inhabit plant roots have the potential to produce salicylic acid (SA) in vitro, in the rhizosphere they will most likely not produce SA but instead excrete siderophores containing a SA moiety. This is postulated to be due to the presence of compounds in the root exudates that are precursors for such siderophores (Audenaert et al. 2002). b Colonization of the rhizosphere by certain bacteria (but also infection of the plant by pathogens) can lead to SA production in the plant and subsequent deposition of SA in the rhizosphere. SA can influence different microbial activities including antibiotic production, biofilm formation and motility, all important for rhizosphere colonization, but also can modulate resistance to antibiotics. Thus SA can influence microbial interactions and play a significant role in shaping microbial community structure

SA and modulation of microbial interactions

Whereas we have just seen that rhizobacterial produced SA can provoke headaches in rhizosphere scientists, exposure of bacteria to SA does result in more intriguing phenotypes. SA can induce a multiple antibiotic resistance phenotype, or reduce resistance to some antibiotics, and affect production of bacterial virulence factors (Price et al. 2000). However, such studies on effects of SA have focused on bacteria that are pathogenic in mammals. In that context SA may interfere with effective treatment of bacterial infections with antibiotics in patients treated with salicylates, or it can have a direct effect on pathogenesis. In a plant-microbe interactions context, we may have to consider that SA production by the plant influences antibiosis driven antagonism among members of the microbiome. Indeed, bacterial competition in natural environments seems to be fierce (Hibbing et al. 2010; Foster and Bell 2012). In an elegant study, Garbeva et al. (2011) demonstrated that P. fluorescens Pf0-1 responds differentially to different bacterial competitors. In the rhizosphere, inhibition of microbial plant pathogens by certain PGPR is based on production of antibiotics, and in some cases development of resistance in the pathogen can interfere with the protective effect of the PGPR (Raaijmakers et al. 2002, 2009). Antibiosis thus seems a common phenomenon in rhizosphere microbial communities. Antibiotics are not only produced as weapons against other microbes but they are also important in signaling networks (Raaijmakers and Mazzola 2012; Baquero et al. 2013). A modulating effect of SA on these interactions may thus have significant impact on the composition and functioning of the rhizosphere microbiome.

The phenomenon that SA mediates resistance to antibiotics in E. coli has been known for decades (Rosner 1985; Cohen et al. 1993). Also for other pathogenic bacteria such effects have been reported. Exposure of S. marcescens to SA induced a phenotypic resistance to nalidixic acid and ciprofloxacin (Berlanga and Viñas 2000). In Salmonella choleraesuis multiple antibiotic resistance was induced by SA in some isolates, but for others SA had no effect (Tibbetts et al. 2003). SA also reduces sensitivity of Salmonella enterica to the antibiotic ciprofloxacin (Hartog et al. 2010). Similarly, antibiotic resistance is increased in Staphylococcus aureus in the presence of SA (Gustafson et al. 1999) due to specific up- and down-regulation of genes involved in resistance to antimicrobials (Riordan et al. 2007). In M. tuberculosis SA reduced sensitivity to isoniazid, rifampin, ethambutol, streptomycin and p-aminosalicylate (Schaller et al. 2002). In this respect it is interesting to note that sensitivity to p-aminosalicylate in M. smegmatis is increased in knock-out mutants that no longer produce SA (Nagachar and Ratledge 2010). For the B. cepacia complex SA induces antibiotic resistance (Burns and Clark 1992) due to the induction of an antibiotic efflux pump (Nair et al. 2004). Induction of the CmeABC multidrug efflux pump is also the mechanism of SA-induced antibiotic resistance in Campylobacter jejuni (Shen et al. 2011). In a study on responses of Bacillus subtilis to SA, proteomic and transcriptomic analysis revealed changes that seemed related to resistance to SA itself (Duy et al. 2007). It was postulated that salicylate production by P. aeruginosa during infection may lead to a phenotype of antibiotic resistance (Martínez and Baquero 2002). For Helicobacter pylori growth was inhibited by SA, and in this case it increased susceptibility to the antibiotics amoxicillin, clarithromycin and metronidazole (Wang et al. 2003).

Also other traits of bacteria that are of importance in their environmental ecology are influenced by SA. Flagellar motility is an important trait for colonization of roots (De Weger et al. 1987; Lugtenberg and Kamilova 2009). SA inhibits motility in E. coli, Proteus mirabilis, Providencia rettgeri, Providencia stuartii and P. cepacia, but not that of P. aeruginosa (Kunin et al. 1995). Biofilm formation is important in the rhizosphere (Morris and Monier 2003). For Staphylococcus epidermidis adding SA to the growth medium resulted in inhibition of biofilm formation (Müller et al. 1998). In a study on biofilm formation by E. coli, Klebsiella pneumoniae, P. mirabilis, and P. aeruginosa, SA was shown to reduce biofilm production and to disrupt preformed biofilms (El-Banna et al. 2012). Quorum sensing, regulation of microbial gene expression in response to cell density, is another important driver of microbial activities in the rhizosphere (Pierson and Pierson 2007). Such activities include antibiotic production, biofilm formation, conjugation, motility, symbiosis, and virulence (Miller and Bassler 2001). SA was identified as a compound that can inhibit quorum sensing activity of the opportunistic human pathogen P. aeruginosa in a dose dependent manner (Yang et al. 2009). Also for the plant pathogen Agrobacterium tumefaciens SA was demonstrated to interfere with quorum sensing (Yuan et al. 2008). Other effects of SA on microbial functioning have been reported. In P. fluorescens CHA0, SA represses the production of 2,4-diacetylphloroglucinol, an antimicrobial metabolite that is important for control of soil borne disease by this PGPR (Schnider-Keel et al. 2000). Virulence of P. aeruginosa on A. thaliana is down-regulated by SA (Prithiviraj et al. 2005b). Also resistance of Arabidopsis to infection by S. aureus may be mediated by a direct effect of SA on this gram-positive bacterium (Prithiviraj et al. 2005a). Again such an impact of SA can greatly influence microbial and interkingdom interactions in the rhizosphere microbiome. Thus, whereas it seems that the rhizobacterial strains studied do not produce SA in the rhizosphere, specific activities of the rhizosphere microbiome may well be influenced by SA produced by the plant (Fig. 1b).

Concluding remarks

Despite the vast knowledge on effects of SA on both animals and plants, relatively little is known about its effect on micro-organisms, including fungi and bacteria. A number of plant associated bacteria have been demonstrated to have the ability to produce significant amounts of SA, but mostly co-regulated with the biosynthesis of siderophores that contain a salicylate moiety. Mutants that no longer produce SA and/or the SA containing siderophore do not differ significantly from the wild-type with regard to their ecology in the rhizosphere. Li et al. (2011) describe the production of the SA containing antibiotic promysalin that is produced by P. putida RW10S1. The involvement of promysalin in the ecology of the producer is intriguing, since it promotes swarming and biofilm formation by the producing strain and at the same time shows toxicity specific to Pseudomonas spp. Thus the interspecies activity of microbially-produced SA containing compounds is not limited to siderophores. The modulation of bacterial motility, biofilm formation, quorum sensing, and resistance to antibiotics by SA suggests that this molecule could be one of the key components involved in assembling the rhizosphere microbiome. Given the recent attention for the rhizosphere microbiome and how it is influenced by plant signals (Bakker et al. 2013; Berendsen et al. 2012; Bulgarelli et al. 2013; Doornbos et al. 2011, 2012; Mendes et al. 2011, 2013), the role of plant derived SA in microbiome assembly deserves to be studied in detail.