Azospirillum brasilense Az39, a model rhizobacterium with AHL quorum‐quenching capacity

The aim of this research was to analyse the quorum‐sensing (QS) and quorum‐quenching (QQ) mechanisms based on N‐acyl‐l‐homoserine lactones (AHLs) in Azospirillum brasilense Az39, a strain with remarkable capacity to benefit a wide range of crops under agronomic conditions.


Introduction
Micro-organisms have the capacity to perceive population density by generating small signalling molecules named autoinducers (Nealson 1977). As a result, a hierarchical response is developed to coordinate social behaviour at the gene level. This process is called quorum sensing (QS) (Fuqua et al. 1994). The most studied QS system is undoubtedly the one that involves N-acyl homoserine lactone (AHL) or AHL-type signals, discovered for the first time in Vibrio fischeri, a seawater symbiont bacterium (Nealson and Hastings 1979). In this bacterium, QS consists of a modulatory protein or transcriptional regulator belonging to the LuxR family and its homologue LuxI, an enzyme that produces the AHL signal molecule. Although a large number of bacteria possess the canonical LuxR/LuxI QS system, it has been found almost exclusively in a-, band c-proteobacteria (Williams 2007). In general, AHLs are small molecules composed of fatty acyl chain linked to a lactonized homoserine through an amide bond. LuxI, more specifically, catalyses the binding of S-adenosylmethionine (SAM) to an acyl carrier protein (acyl-ACP). In other words, LuxI catalyses the binding between a homoserine lactone group derived from the metabolism of amino acids, and an acyl lateral chain derived from fatty acid metabolism, which are the two structural components of the resulting AHL (Fuqua et al. 2001). For their part, LuxR-like proteins (with approximately 250 amino acids) can be subdivided into two functional domains: the amino-terminal region that contains the AHL-binding domain and the carboxyl-terminal region that contains the helix-turn-helix (HTH) of DNA (Whitehead et al. 2001). Once in contact with the AHLs, LuxR joins a 20bp palindrome called the lux box, from the luxI promoter region, in the form of a LuxR-autoinducer complex. This leads to transcriptional activation or repression, thus expressing a particular phenotype.
On the other hand, some bacterial strains present quorum systems with a noncognate LuxR protein (i.e. they lack LuxI) and they thus respond to other signal molecules. These systems are called LuxR orphans or LuxR solos (Patankar and Gonzalez 2009), and in some cases they act in concert with the LuxR/LuxI canonical system. The appearance of LuxR solos regulators indicates that these protein families could be involved in intrakingdom or interkingdom signalling systems through the detection of different compounds produced by other prokaryote or eukaryote organisms (Patankar and Gonzalez 2009;Patel et al. 2013).
In nature, there are also bacterial mechanisms that inactivate quorum signals called quorum quenching (QQ) (Zhang 2003). These can generally act both at the level of signal generation and reception. Although there are several QS mechanisms involving inhibitory proteins and/or AHL antagonist molecules, the mechanisms that involve enzymes are widespread in different environments. Three main enzymatic QQ mechanisms have been clearly described: (i) hydrolysis of the lactone ring (AHL lactonase activity), (ii) hydrolysis of the amide bound (AHL-acylase activity) and (iii) modification of the acyl chain (AHL oxidase and reductase activity) (Uroz et al. 2009), but they have not been studied in depth in soil bacteria. As occurs in the QS system, QQ mechanisms can serve in particular environments to modulate the interaction between a bacterial community and eukaryotic organisms (Tait et al. 2009).
Soil bacteria living in the rhizosphere, or rhizobacteria, have the ability to associate with numerous plant species. If this association is beneficial for plant growth or development, they are called plant growth-promoting rhizobacteria or PGPR (Kloepper et al. 1989). Among the most successful associations and therefore the most studied in nature, are those related to the genus Azospirillum sp. The ability of these rhizobacteria to promote plant growth depends mainly on the presence of one or more mechanisms that might act individually or synchronized on the physiology or metabolism of the colonized plant (Bashan and de-Bashan 2010).
Azospirillum brasilense Az39 was isolated in 1982 from surface-sterilized wheat seedlings in Marcos Juarez, C ordoba, Argentina. It was evaluated under agronomic conditions and selected based on its ability to increase crop yields of maize and wheat under said conditions (D ıaz-Zorita and Fern andez-Canigia 2009). Azospirillum brasilense Az39 has been widely used in American agriculture throughout the last 40 years (Cass an and Diaz-Zorita 2016). The potential mechanisms responsible for growth promotion in this strain have been partially unravelled (Perrig et al. 2007;Cass an et al. 2009). Despite its agro-economic importance and the fact that several genomes from this genus have been sequenced, such as those belonging to Azospirillum sp. B510, Azospirillum lipoferum 4B, Azospirillum brasilense Sp245, CBG497 and Az39 (Kaneko et al. 2010;Wisniewski-Dy e et al. 2011, 2012Rivera et al. 2014), there are few reports related to the bacterial capacity to produce AHL-like molecules and/ or other phenomena associated with quorum mechanisms. Therefore, there is little understanding regarding the Azospirillum-Azospirillum, Azospirillum-bacteria and Azospirillum-plant interactions mediated by quorum mechanisms, highlighting the need for a more exhaustive genomic functional analysis of these bacteria due to their agricultural and economic interest. Considering this background, the main objective of this work was to analyse both in silico and in vitro the QS and QQ phenomenon mediated by AHLs in the model strain A. brasilense Az39.

Bacterial strains and growth conditions
Azospirillum brasilense Az39 was obtained from the bacterial culture collection at the INTA-IMYZA, Castelar, Buenos Aires, Argentina (WDCM31). Pure cultures of A. brasilense Az39 were obtained in Petri dishes containing Luria-Bertani medium (Miller 1972) modified by the addition of 15 ml l À1 Congo Red (LB-RC) or MMAB minimal medium (Vanstockem et al. 1987). Typical colonies from such media were used to inoculate LB liquid medium in 100-ml flasks and cultured at 37°C with 240 rev min À1 shaking until late exponential growth phase was reached. Chromobacterium violaceum CV026 (McClean et al. 1997) grew in LB medium supplemented with 25 lg ml À1 kanamycin (Km). Agrobacterium tumefaciens NTL4/pZLR4 (Cha et al. 1998) was cultured in AT medium (Morton and Fuqua 2013) supplemented with 50 lg ml À1 gentamicin (Gm). These two strains were used as reporter strains in the bioassays described below.

In silico analysis of quorum mechanisms in A. brasilense Az39
We determined the presence of coding sequences for proteins involved in QS mechanisms in the genome of A. brasilense Az39, and compared them with available sequences from other strains belonging to the genus Azospirillum. For the analysis, the comparative tools KEGG (Kanehisa et al. 2004), RAST (Aziz et al. 2008) and MaGe (Vallenet et al. 2006) were used, as well as the bioinformatic tools UniProt (Apweiler et al. 2004) and InterPro (Mulder et al. 2005). Our work focused on the identification of coding sequences related with: (i) enzymes and transcriptional regulators involved in QS detection/response, (ii) AHL synthases homologues, (iii) homologous LuxR-type regulatory proteins, (iv) LuxR orphans or LuxR solos and (v) enzymes and transcriptional regulators involved in QQ detection/response, including lactonases, acylases and oxidoreductases. In order to predict subcellular localization of a protein, CELLO database (http://cello.life.nctu.edu.tw.) was used.

In vitro analysis of quorum mechanisms in A. brasilense Az39
Quorum sensing Evaluation of production of AHLs by bioassays. The presence of AHLs in Az39 cultures was validated using the reporter strains C. violaceum CV026 and A. tumefaciens NTL4/pZLR4 which are specific for AHLs with a short and long acyl chains respectively. A 500-ll aliquot of a A. tumefaciens NTL4/pZLR4 or C. violaceum CV026 exponential cultures were individually transferred into 10-ml capacity glass tubes containing 4500 ll of semisolid AT medium 0Á7% (w/v agar), modified by the addition of 50 lg ml À1 X-gal at 45°C or semisolid LB 0Á7% (w/v agar) respectively. The suspensions were plated out on Petri dishes containing solidified media under aseptic conditions. In both cases, small holes were made in the Petri dish containing AT-or LB-solidified culture medium, using a 5-mm cylindrical punch. A 10-ll aliquot of filtered supernatants obtained from 50 ml LB culture medium at 6, 12, 24 and 48 h after inoculation with 50 ll of Az39 were individually dispensed into the holes and evaluated. The plates were incubated at 30°C for 24 h to reveal the presence of AHL molecules by a colorimetric reaction. In addition, some experimental conditions, such as incubation temperature, pH and AHL concentration, were previously evaluated to analyse the reproducibility of the methodology and the stability of the AHL molecules in the Petri dish during incubation. Experiments were carried out in triplicate.
Evaluation of production of AHLs by Az39 using liquid chromatography coupled to mass-mass spectrometry analysis. Extraction of AHLs from Az39 cultures-Typically, A. brasilense AZ39 colonies grown on LB-RC medium were used to inoculate 250 ml of LB medium and incubated at 37°C, with shaking (200 rev min À1 ) until the stationary growth phase had been reached. Aliquots (100 ml) of centrifuged (5 min at 10 000 rev min À1 ), and sterile filtered supernatant (0Á22 lm, Millipore Express PLUS) were acidified to pH 2 with the addition of HCl. Supernatant samples were extracted three times by liquid-liquid extraction using an equal volume of acidified ethyl acetate (1% (v/v) AcOH in EtOAc), as previously described (Ortori et al. 2011). Combined extracts were dried under vacuum and stored at À80°C prior to analysis.
Liquid chromatography coupled with mass-mass spectrometry analysis-The liquid chromatography coupled with mass-mass spectrometry (LC-MS/MS) analysis of extracted samples was conducted as previously described (Ortori et al. 2011) with minor modification. Dried extracts were redissolved in 50 ll of 0Á1% (v/v) formic acid in MeOH. The chromatography column used was a Phenomenex Gemini C18 (3Á0 lm, 150 9 3Á0 mm), and the mobile phases used were 0Á1% (v/v) formic acid and 0Á1% (v/v) formic acid in methanol. The analysis was conducted with the MS operating in multiple reaction monitoring mode, simultaneously screening the LC eluent for all specific AHLs, comparing the retention time of detected analytes with authentic synthetic standards. For each detected chromatographic peak, the mean peak area was calculated from three biological replicates.

Quorum quenching
Evaluation of degradation of AHLs by Az39 by LC-MS/MS analysis. A set of nine glass flasks of 50 ml capacity containing 20 ml of LB medium was prepared. Only six were inoculated with 20 ll of Az39 culture obtained from liquid LB medium in late exponential growth phase (OD 595 1Á0), and three remained without inoculation (controls). The nine flasks were then incubated overnight at 37°C with 200 rev min À1 orbital shaking. After a 12-h incubation period, the tubes containing the Az39 cultures and the noninoculated control tubes were modified by the exogenous addition of 100 ll of a methanolic solution containing C4, C6, C8, C10, C12, C14, Oxo-C4, Oxo-C6, Oxo-C8, Oxo-C10, Oxo-C12, Oxo-C14, OH-C4, OH-C6, OH-C8, OH-C10, OH-C12 and OH-C14, each at a concentration of 100 lmol l À1 for a final concentration of 500 nmol l À1 for each individually added AHL molecule. A 100 ll methanol control treatment was used to evaluate bacterial growth inhibition. The glass flasks were incubated for 6 h, and at 1-, 3-and 6-h intervals. One millilitre samples were collected and stored at À20°C until processing, extraction of the AHL and analysis by liquid chromatography-mass spectrometry, as described above. The degradation of each AHL across three time points was indicated by a significantly reduced chromatographic peak area from cultures of Az39 with endogenously added AHLs compared with uninoculated control samples.
Enzymatic activity associated with the AHL degradation. A 50-ll aliquot of A. brasilense Az39 exponential growth culture (OD 595 1Á0) obtained in liquid LB medium was used to inoculate 100-ml capacity glass flask containing 50 ml of MMAB medium. When the cultures reached OD 595 0Á8-1Á0, corresponding to the exponential growth phase, they were fractionated into 5-ml portions, placed in sterile 10-ml tubes and treated individually with 10 lmol l À1 C6-HSL, hexanoyl-homoserine lactone or 10 lmol l À1 C10-HSL, decanoyl-homoserine lactone (University of Nottingham, Nottingham, UK). Tubes were then incubated for 12 h at 37°C with 240 rev min À1 shaking. After incubation, the presence of AHLs in the culture medium was evaluated by bioassays using the reporter strains as described in section "Evaluation of production of AHLs by bioassays". In a second experiment under similar conditions, a 1-ml aliquot of the AHL-treated Az39 culture was transferred to sterile microtubes and heated at 100°C for 10 min with the aim of inactivating the bacterial cells and denaturing the proteins in the culture. An additional tube without heat treatment was used as nondenaturing control. Once heating finished, 10 lmol l À1 of C6-AHL or C10-AHL were individually added and the tubes were incubated at 37°C with 240 rev min À1 orbital shaking. After different incubation times (0Á5, 1, 3, 6, 12 and 24 h), 30 ll samples were taken to be analysed in bioassays as described above. To check the cellular localization of the putative enzyme (or enzymes) involved in this activity we performed a second analysis considering an induction stage according to Uroz et al. (2007). For that purpose, Az39 was grown in MMAB medium supplemented by the exogenous addition of 10 lmol l À1 individual AHL (C6-AHL or C10-AHL), and this was defined as a preinduced Az39 culture (Az39-pi). All the treatments performed after induction are detailed as follows: T1: Noninoculated LB supplemented with 10 lmol l À1 AHL (control); T2: Filtered supernatant of Az39pi + 10 lmol l À1 AHL; T3: Heated and filtered supernatant of Az39-pi + 10 lmol l À1 AHL; T4: Culture of Az39-pi + 10 lmol l À1 AHL and T5: Heated culture of Az39-pi + 10 lmol l À1 AHL. The addition of individual AHL to each treatment depended on the reporter strain used: C6-AHL for C. violaceum and C10-AHL for A. tumefaciens.

In silico analysis
Quorum sensing Different bioinformatic tools were used to identify putative proteins related to canonical and noncanonical QS systems in these bacteria. When the genome of several strains belonging to the genus Azospirillum was analysed, the presence of a coding sequence for an AHL synthase (LuxI) (EC 2.3.1.184) could be confirmed in only three of them: A. lipoferum TVV3, Azospirillum sp. B510 and Azospirillum sp. RU38E. This protein is formed by a typical domain defined as IPR001690 (autoinducer synthase) that refers to the autoinducer synthase family of proteins according to InterPro database. The genes encoding the AHL synthases in these Azospirillum strains have been annotated in the UniProt database as alpI, AZL_a05890, luxI AZA_90644, SAMN05880556_102381 and SAMN 05880556_11440 for A. lipoferum TVV3 (Q19U13_ AZOLI), Azospirillum sp. B510 (D3P0E1_AZOS), the only strain containing the domain IPR018311 and Azospirillum sp. RU38E (A0A239I230) respectively. For A. brasilense Az39, no homologues of LuxI or another AHL synthase (LuxS, CqsA, HdtS and LuxM) involved in QS were identified.

Quorum quenching
Although N-acyl-homoserine lactonases (EC: 3.1.1.81) were not found in the genome of the Azospirillum strains analysed, there are several N-acyl-homoserine lactone acylases (EC: 3.5.1.97) annotated for this bacterial genus in the UniProt database: A. brasilense Sp7 (AMK58_19595), A. brasilense Sp245 (AZOBR_p1130068), Azospirillum sp. B510 (AZL_013430), A. lipoferum 4B (AZOLI_p40482) and A. thiophilum DSM 21654 (VY88_13715), and in particular for A. brasilense Az39 (ABAZ39_22635). In the RAST server, a protein annotated as penicillin acylase (fig 192.31.peg.4511) was identified in plasmid 1 of the Az39 genome (Fig. S4). Its sequence has 100% identity and homology with the sequence identified through the Uni-Prot database. In addition to penicillin acylase, an aliphatic amidase AmiE (EC. 3.5.1.4) was found in the genome of Az39 (fig 192.31.peg.3259) and both enzymes have been described as AHL-acylases in some databases and literature (Ochiai et al. 2014). Results found through BRENDA (http://www.brenda-enzymes.org) depended on the organism studied. In the case of AmiE, there are 13 recorded entries, distributed in 4 cellular locations (cytoplasmic, extracellular, lysosomal and in the membrane). On the other hand, 23 entries were registered for penicillin acylase, associated with five cellular locations in different bacteria (cytosolic, extracellular, intracellular, periplasmic and in the membrane). While it is evident that there are AHL-acylase enzymes with different substrate specificities, there are records of an aculeacin-A acylase, a putative N-acyl-homoserine lactone acylase with quorum-quenching activity (EC: 3.5.1.-) from the Gramnegative Ralstonia solanacearum with the same ability as Az39 to degrade AHLs (Chen et al. 2009). A more detailed analysis of the aculeacin-A acylase using both UniProt and InterPro revealed a structural organization of 786 amino acids distributed in 6 protein regions: signal peptide, propeptide, aculeacin-A acylase itself, the small subunit of aculeacin-A acylase, peptide spacer and the large subunit of aculeacin-A acylase (Inokoshi et al. 1992). Subsequently, a BLASTP analysis was made in block with these sequences against the Az39 genome, to determine if all these regions were present. Interestingly, the absent region in Az39 is the signal peptide responsible for releasing the enzyme into the extracellular space, in agreement with the analysis by CELLO (http://cello.life.nc tu.edu.tw/), which probabilistically locates this enzyme in the cytoplasm or associated with the internal membrane and periplasmic space rather than with the extracellular space or external membranes.

Lux R transcriptional regulators
A total of 28 LuxR transcriptional regulators were found in A. brasilense Az39 genome (Table 1). These sequences belong to the superfamily of LuxR regulators and share between them the InterPro IPR000792, HTH binding to the DNA C-terminal domain that is characteristic of this large superfamily. Although these proteins are annotated as LuxR regulators in A. brasilense Az39, only one of them corresponds to a typical LuxR with an N-terminal domain binding to the autoinducer and could be a putative LuxR solo since it lacks an AHL synthase cognate enzyme. It is annotated as an uncharacterized protein ABAZ39_30865 under accession UniProtKB-A0A060DZ Q2 and as an autoinducer-binding transcriptional regulator of the LuxR family (fig 192.31.peg.6164) in the Uni-Prot database and RAST server respectively. Azospirillum brasilense Az39 genome contains also coding sequences associated with the biosynthesis of 8 GroEL/ES-type chaperone proteins, which are fundamental for folding and stability in this type of receptors. Table 1 summarizes the findings of the in silico analysis of LuxR-type regulators from several strains belonging to the genus Azospirillum.

Evaluation of the biosynthesis of AHLs by Az39 using reporter strains
The presence of AHL molecules in filtered supernatants of A. brasilense Az39 was evaluated in bioassays using C. violaceum CV026 and A. tumefaciens NTL4/pZLR4, reporters for short-and long-chain AHLs, respectively, and is summarized in Fig. 1. The evaluation was performed at different time points in the typical growth curve using two liquid culture media and synthetic AHLs as control. According to the absence of an AHL synthase in the genome of Az39, this bacteria is unable to biosynthesize this type of molecules, something that was clearly evidenced in the bioassays using C. violaceum CV026 (Fig. 1a) and A. tumefaciens NTL4/pZLR4 (Fig. 1b). Additional extractions with organic solvents were made from larger volumes of culture medium in order to increase the concentration of possible metabolites at different time points in the growth curve. None of the analysed samples presented a reporter activity due to the absence of AHL-type molecules (Fig. S1).

Evaluation of AHL degradation by Az39 using reporter strains
The degradation of exogenous AHLs in cultures of A. brasilense Az39 was assessed using the bioassay system as described before. The evaluation was performed at different time points of the typical growth curve using uninoculated liquid culture media modified by the addition of synthetic AHLs as control (Fig. 2a,c). To determine whether the inactivation by Az39 was of enzymatic origin,  a simple experiment of induction and denaturation was carried out. Figure 2b,d clearly shows that degradation of AHLs by Az39 has an enzymatic origin, because the denaturation of the supernatant at 100°C revealed the presence of both short-chain and long-chain AHLs in the supernatants respectively.

Evaluation of AHL degradation by LC-MS/MS analysis
In order to validate the results obtained by the use of reporter strains regarding the ability of A. brasilense Az39 to produce or degrade AHLs (4-14 carbon atoms), a confirmation procedure was performed by the use of LC-MS/MS. As seen in Fig. 3, no AHLs were detected in the samples obtained from Az39 cultures (Az39-AHL). In the samples of Az39 cultures pre-incubated with unsubstituted AHLs in C3 (Az39 + AHL), AHL levels were lower than in noninoculated LB incubated with 500 nmol l À1 of each AHL (LB + AHL) under similar experimental conditions. A similar behaviour was observed in experiments by the addition of AHLs substituted with the hydroxy and keto (oxo-) groups in carbon 3 (Figs S2 and S3). These results not only demonstrate the ability of Az39 to degrade AHLs, they also highlight the wide spectrum of molecules it can degrade, making this strain a putative regulator of bacterial quorum activity in the rhizosphere of higher plants.
QQ activity is associated with Az39 cells As seen in Fig. 4, the activity of reporter strain C. violaceum CV026 and synthetic short-chain AHLs confirmed the influence of the denaturation process (100°C) on the loss of degradation activity in Az39 cultures. This phenomenon was visualized as a strong decrease in violacein production at increasing incubation times (Fig 4, Treatment 5). Because the inactivation of AHLs was not observed in the denaturized supernatants of Az39, we assume that quenching activity must be associated with the bacterial cell. In other words, the enzyme/s responsible for AHL degradation is/are not secreted into the culture medium by A. brasilense Az39. Similar results were obtained in the case of long-chain AHLs and A. tumefacines (data not shown). In summary, these results support the notion that AHL degradation by Az39 is of enzymatic character and limited to a specific cellular compartment, since the enzymes do not seem to be released into the external environment, which suggests that the activity could be linked to the plasma membrane or periplasm.

Discussion
Despite genomic information currently available on the genus Azospirillum, little is known about the molecular mechanisms related to bacterium-bacterium and bacterium-plant communication. Interestingly, some reports about mechanisms based on QS in some strains of the genus Azospirillum agree with the in silico analysis presented in this paper. Vial et al. (2006) used two biosensor strains to test AHL production in 40 strains belonging to 6 species of Azospirillum, obtained or isolated from different geographic locations. They found that only three strains of A. lipoferum (TVV3, B52, B518) and a related strain (B510) were able to produce this signal molecule. We also found that the genome of Azospirillum sp. RU38E presents two luxI genes that are cognate to their respective luxRs. In the case of A. brasilense, other authors recently investigated QS mechanisms in Ab-V5 and Ab-V6, the strains most commonly used for inoculant formulation in Brazil (Fukami et al. 2017). They found no genes associated with an AHL synthase but multiple LuxR solos in the genome, although their publication does not include a detailed analysis. Similarly, in the case of A. brasilense Az39, there is no luxI gene associated with the production of AHLs, something which was subsequently confirmed in silico and in vitro by both the use of reporter strains C. violaceum CV026 and A. tumefaciens NTL4/pZLR4, and the LC-MS/MS analysis. Several genes encoding putative proteins related to QS systems were identified in this paper, but the absence of LuxI in all A. brasilense strains suggests that AHL production may not be related to this bacteria species.
On the other hand, A. brasilense Az39 contains a LuxR orphan or solo. An analysis of multiple sequence alignment of this LuxR compared with LuxR cognates and LuxR solos already described in the literature allowed to show that some amino acid residues characteristic of the N-terminal domain of binding to the autoinductor remain conserved, which classified them outside the family of typical LuxR regulators (data not shown). The conservation of amino acid residues present in the LuxR of Az39 is a fact that could be associated with LuxRs that respond to exogenous AHLs (by "eavesdropping") from bacteria with which they share niche and/or other molecules chemically similar from their host plants (Patel et al. 2013;Venturi et al. 2018).
Signalling mediated by QS in bacteria can be interrupted by a wide variety of phenomena collectively known as QQ. The coding sequence for a N-acyl-homoserine lactone acylase (EC: 3.5.1.97) was found in A. brasilense Az39, A. brasilense Sp7; A. brasilense Sp245, Azospirillum sp. B510, A. lipoferum 4B and A. thiophilum. These findings suggest that mechanisms of quorum signal interception prevail in different species of the genus, regardless of whether they produce such molecules or not. In addition, the appearance of such mechanisms in these strains, and especially in A. brasilense Az39, points towards the important role this kind of regulation fulfils, not only in selecting the ecological niche and exchanging signals with the host plant but also in adapting to a lifestyle in the rhizospheric environment. We also demonstrated, through the use of reporter strains, that the inactivation of synthetic AHLs by Az39 was related to an enzyme activity. In this sense, the capacity of this strain to degrade AHL was confirmed in vitro and justified by the presence of two coding sequences for two putative AHL-acylases. Considering the results we obtained in this paper using reporter strains, the tentative location of the putative AHL-acylase activity would be a cellular compartment, likely the plasmatic membrane or the periplasmic space.
The ability of A. brasilense Az39 to degrade AHLs of different lengths (4-14 carbon atoms) was confirmed by the use of LC-MS/MS. According to the treatments proposed, the AHL levels in pure Az39 cultures incubated with unsubstituted AHLs and substituted at C3 were lower than in noninoculated LB medium. These results unequivocally indicate that although A. brasilense Az39 does not produce AHLs, it is capable of degrading them in liquid culture conditions. We compared the penicillin acylase (AHL-acylase) coding sequence in the genome of Az39 with the in silico and in vitro characterization by Mukherji et al. (2014)  Mean peak area (n=3) Mean peak area ( a Penicillin-G-acylase from Kluyvera citrophila, an enzyme that also has the ability to cleave AHLs, and found them to have a high similarity. This is an important biotechnological approach that represents a new positioning in the large-scale production of biofunctional enzymes that govern the flow of chemical information in the rhizosphere, where complex bacterial communication networks take place. In this sense, several experiments have shown how plants respond to QS signals such as the AHLs produced by Gram-negative bacteria (Bauer and Mathesius 2004;Von Rad et al. 2008). It is currently known that plants, in addition to responding to AHLs, produce molecules that can mimic such QS signals by somehow manipulating behavioural mechanisms associated with bacteria in the rhizosphere (Teplitski et al. 2000;Corral-Lugo et al. 2016). On the other hand, Palmer et al. (2014) showed that plants can produce AHL-acylase enzymes using Lhomoserine for their own benefit. The accumulation of Lhomoserine has several effects on plant growth: it increases transpiration which favours nutrient uptake by the roots, promotes defence responses mediated by Ca 2+ , stimulates the production of ethylene and promotes the synthesis of auxins. This last effect is correlated in the rhizosphere with the capacity of A. brasilense Az39 to produce several phytohormones, including auxins (Cass an and Diaz-Zorita 2016). This, coupled with its AHL QQ capacity, enhances the synergy of the interaction between Az39 and the plant. The results obtained in this paper suggest that under the prevailing conditions in the rhizosphere, Az39 is mute in the sense that it cannot speak the language mediated by AHLs, but it can interrupt conversations between other bacteria and plants by a QQ mechanism. This mechanism could regulate the capacity of the microbial populations interacting with plants and this should be investigated in further experiments.

Supporting Information
Additional Supporting Information may be found in the online version of this article: Figure S1. Evaluation of violacein production and b-galactosidase activity induced by the presence of AHLs in cultures of Az39. Right: Bioassay using Chromobacterium violaceum CV026. C (control): 10 lmol l À1 C6-AHL. Treatments 2 and 3: filtered supernatants obtained from different stages of Az39 growth curve at DO 595 0Á823 and 1Á654 respectively. Treatment 1: noninoculated LB culture medium modified by the addition of C6-AHL. Left: Bioassay using Agrobacterium tumefaciens NTL4/ pZLR4. C (control): 10 lmol l À1 C10-AHL. Treatments 2, 3, 4 and 5: filtered supernatants obtained from Az39 growth curve at OD 595 0Á621, 1Á054 and 1Á872 respectively. Treatment 1: noninoculated LB culture medium modified by the addition of C10-AHL. Figure S2. Identification and relative quantification of AHLs by liquid chromatography coupled with mass-mass spectrometry (LC-MS/MS). In the experiments, AHLs of 4-14 carbon atoms substituted at C3 with a hydroxyl group (-OH) were used at a final concentration of 500 nmol l À1 . The bars represent a mean peak area calculated from three biological replicates of the following treatments: Az39 + AHLs, Az39-AHLs and noninoculated LB + AHLs after 1, 3 and 6 h of incubation time. Columns marked with a different letter of the same group of treatments differ significantly by Tukey post hoc test at P < 0Á05. Figure S3. Identification and relative quantification of AHLs by liquid chromatography coupled with mass-mass spectrometry (LC-MS/MS). In the experiments, AHLs of 4-14 carbon atoms substituted at C3 with an oxo group (-oxo) were used in a final concentration of 500 nmol l À1 . The bars represent a mean peak area calculated from three biological replicates of the following treatments: Az39 + AHLs, Az39-AHLs and noninoculated LB + AHLs after 1, 3 and 6 h of incubation time. Columns marked with a different letter of the same group of treatments differ significantly by Tukey post hoc test at P < 0Á05. Figure S4. Structural organization of the Az39 genome at the level of the putative Penicillin acylase.