Burkholderia ambifaria displays a remarkable potential as a PGPR and biocontrol agent, but like the other Bcc members, its commercial use is placed under a moratorium (Chiarini et al. 2006). The production of commercially interesting molecules in vitro through biotechnological processes requires the understanding of mechanisms that direct and regulate their biosynthesis, such as QS.
Phenotypes of the HSJ1 cepI mutant
Several phenotypes classically associated with QS, such as siderophores production and proteolytic activity, are in fact disparately present and QS-controlled among Bcc members (Gotschlich et al. 2001; Huber et al. 2001; Aguilar et al. 2003). We report here that the cepI mutant of B. ambifaria strain HSJ1 overproduces siderophores, similarly to the cepI mutant of B. cenocepacia K56-2 (Lewenza et al. 1999). The HSJ1 cepI mutant has a decreased protease activity, as most of Bcc strains in absence of C8-HSL (Wopperer et al. 2006). This phenotype has been linked in B. cenocepacia to two QS-regulated metalloproteases, ZmpA and ZmpB (Gingues et al. 2005; Kooi et al. 2006). These proteases were also identified in B. ambifaria strain HSJ1 (Vial et al. 2010); we found here that zmpA and zmpB are strongly downregulated in the HSJ1 cepI mutant (Fig. S2 and data not shown), which probably explains the decreased proteolytic activity.
We have also looked at some phenotypes previously described in B. ambifaria HSJ1 WT strain (Vial et al. 2010), such as the colonial morphology on Congo Red agar plates. The colonial wrinkling is a QS-regulated character in HSJ1 strains; a such colony wrinkling has been linked to the QS-regulation of the exopolysaccharide Pel in Pseudomonas aeruginosa (Gupta and Schuster 2012). The secretion of a FAD-dependent cholesterol oxidase, identified as Bamb_6465 and correlated with a cholesterol-degrading activity, was also previously reported in B. ambifaria HSJ1 (Vial et al. 2010). We found here that this phenotype is positively controlled by QS (Fig. 2C), and confirmed that it is due to Bamb_6465, which is strongly QS-activated in B. ambifaria HSJ1 (Figs. 5B, 6C, and S2).
Another noteworthy phenotype is the beta-hemolytic activity, which was highlighted when B. ambifaria was initially described as a new Bcc species (Coenye et al. 2001). Factors implicated in such effects in Bcc members are poorly identified; however, a hemolytic compound named cepalycins, displaying also antifungal properties, has been previously isolated from the supernatant of B. cepacia JN106 (Abe and Nakazawa 1994). Similar hemolytic properties were recently reported for occidiofungins in B. vietnamiensis DBO1, compounds initially described for their antifungal activities in B. contaminans MS14 (Lu et al. 2009; Thomson and Dennis 2012). These dual activities likely result from the interaction of these extracellular molecules with cholesterol in the membrane; indeed, cepalycins were more inhibited by ergosterol than by cholesterol (Abe and Nakazawa 1994). Moreover, environmental strains of Bcc apparently display more hemolytic activity than clinical strains (Bevivino et al. 2002), which is coherent if hemolytic molecules are in fact antifungal molecules; this hypothesis is also compliant with the natural ecology of B. ambifaria. In our screening, we have identified three genes implicated in occidiofungins biosynthesis (Table 1). The mutant trBamb_6472/trBamb_6476 is impaired in hemolytic function (Fig. 6A), while trBamb_6469 behave as the cepI mutant (data not shown), which is in agreement with recent published data (Thomson and Dennis 2012). On the other hand, trBamb_6477 behaved also like the cepI mutant (data not shown); this discrepancy with the study of Thomson and Dennis (2012) could well be explained by the use of different Bcc species or by the compensation of the disrupted gene by another one with similar function.
Additionally, the antifungal activities of B. ambifaria strains against several fungi have been already described (Cain et al. 2000; Zhou et al. 2003), and implication of the QS-regulation for such antifungal activities has also been reported (Zhou et al. 2003; Schmidt et al. 2009). Accordingly, our phenotypic assay against P. ultimum, R. solani, and C. albicans demonstrated that B. ambifaria HSJ1, although being from clinical origin, exhibits antifungal activity, while its cepI mutant displays reduced (or even abolished) antifungal activities.
High-throughput screening to identify new QS-regulated genes
Global approaches using high-throughput screenings have been developed to identify more rapidly and efficiently a wide range of QS-regulated genes, often with the aim to identify those coding for potential virulence factors. We have identified 20 QS-controlled genes employing a procedure derived from Chambers et al. (2006) that had permitted the identification of seven QS-controlled genes in B. cenocepacia K56-2. This approach does not allow to discriminate genes directly or indirectly controlled by QS (Wei et al. 2011). To partially circumvent this shortcoming, we have performed a bioinformatics search for the presence of putative cep boxes upstream of transcriptional units identified in the screening. cep boxes are short sequences upstream the promoter of target genes that allow CepR to recognize its chromosomal binding site. Although the cep box upstream of cepI is well conserved among Bcc members (Fig. S1), the conservation is less obvious upstream of other QS-controlled genes. We have used the consensus sequence described in B. cenocepacia by Chambers et al. (2006) to predict cep boxes in B. ambifaria HSJ1. We have notably identified a putative cep box upstream of prnA (Bamb_4726), which has been identified six times in our screening and for which the direct regulation by CepR has been experimentally demonstrated in Burkholderia lata 383 (Schmidt et al. 2009). A recent study has reported a different consensus sequence in B. cenocepacia K56-2, as well as the experimental demonstration of the direct regulation for two genes, BCAL0510 and BCAM1869 (Wei et al. 2011). BCAM1869 is an ortholog of Bamb_4117, which is located between cepR and cepI, and for which we have predicted a putative cep box (Fig. S1). On the other hand, BCAL0510 is an ortholog of Bamb_3128, which has been identified in the screening but not in the cep box prediction. We have also used this consensus sequence in our bioinformatics study (Fig. S1). Our predictive method, based on the Chambers’ study, allowed thus to cross data with two others predictive and experimental reports, reinforcing confidence in our results.
Genes identified in the screening
Our screening allowed us to identify 20 genes corresponding to 17 loci (Table 1). According to the LacZ reporter activity challenge, three genes were strongly downregulated after C8-HSL addition (Fig. 5A). These genes are implicated in metabolic functions, such as Bamb_2520, which is the ortholog of B. cenocepacia cysN, part of an operon implicated in sulfur metabolism (Iwanicka-Nowicka et al. 2007). This operon is regulated by two LysR-type regulators, CysB and SsuR, but additional QS-regulation has not been reported (Iwanicka-Nowicka et al. 2007). Bamb_3350 (trpA) is the last of a six-genes operon implicated in the biosynthesis of tryptophan, which can be then catabolized via the tricarboxylic acids (TCA) cycle, or used as the precursor of many metabolites, such as pyrrolnitrin, 4-hydroxy-2-alkylquinolines (HAQ) produced by P. aeruginosa, or their methylated counterparts (HMAQ) discovered in Burkholderia (Déziel et al. 2004; Vial et al. 2008; Schmidt et al. 2009). Some of these metabolites are implicated in, or regulated by QS, but the QS-regulation of tryptophan biosynthesis is not established. However, in a quorum-quenching study realized in Azospirillum lipoferum, TrpA was identified among the QS-repressed proteins (Boyer et al. 2008), which agrees with our observations. The third mutant included in panel A is trBamb_2378, which is interrupted in the gene coding a spermidine synthase (speE)-like protein. Spermidine is a polyamine, implicated in several biological processes, both in eukaryotic and prokaryotic cells (Igarashi and Kashiwagi 2010). As for the other QS-repressed genes, the link with QS is not clear; in B. pseudomallei, inhibition of intracellular spermidine synthesis lead to reduced export of AHL via efflux pumps, which suggested that spermidine had an effect on AHL, but the reciprocal was not suggested (Chan and Chua 2010).
It is interesting to note that genes identified in the screening that are QS-repressed or moderately affected are mainly found in the chromosome 1, genes located on chromosome 2 are diversely QS-regulated and genes located on the chromosome 3 are exclusively QS-induced (Table 1). In B. cenocepacia, chromosome 1 carries most of essential (“housekeeping”) genes, while the remaining two chromosomes contain much accessory genes implicated in adaptation to niches; the third chromosome has even been described as a virulence plasmid (Holden et al. 2009; Agnoli et al. 2012; Juhas et al. 2012). As discussed above, the link between QS and metabolism is difficult to decipher, as the genes are also regulated by other factors such as the availability of nutriments, whereas QS-regulation of secreted virulence factors is more obvious. The genes moderately affected by C8-HSL addition, for which a function is predicted, are implicated in metabolic functions and stress adaptation (Fig. 5C and D; Table 1). Four of these six genes are predicted to be preceded by a putative cep box (Table 1; Fig. S1). These genes were at least partially affected by experimental procedures, as two mutants displayed increased β-galactosidase activity if grown in solid rather in liquid medium (Fig. 5D). Another interesting example is trBamb_1141, interrupted in a gene coding a heat shock protein HSP20, which displayed opposite β-galactosidase activities in response to C8-HSL addition, according to the temperature of growth (Fig. 6D). In B. cenocepacia K56-2, the ortholog of this gene is positively regulated by CepR2, an orphan LuxR transcriptional regulator (Malott et al. 2009). We can thus suggest that the genes that appear moderately affected by C8-HSL addition are, additionally to the regulation exerted by C8-HSL, controlled by supplementary factors, such as environmental stresses or other regulation circuitry.
Finally, panel B of Figure 5 contains 11 genes strongly induced by C8-HSL, such as Bamb_6465, responsible for the cholesterol oxidase activity as mentioned above, or Bamb_5109, located upstream of a large nine-genes operon implicated in the biosynthesis and transport of polysaccharide. The most reactive mutant is trBamb_4726, which is interrupted in prnA, the first gene of the operon directing pyrrolnitrin biosynthesis from tryptophan. Pyrrolnitrin is active against a broad spectrum of bacteria and fungi (el-Banna and Winkelmann 1998), and the regulation of its biosynthesis by QS has been demonstrated (Schmidt et al. 2009). The genes Bamb_5911 and Bamb_5925 are implicated in the biosynthesis of enacyloxins, which are antimicrobial compounds produced by B. ambifaria and especially active against B. multivorans (Mahenthiralingam et al. 2011). Bamb_5925 is important in the biosynthesis, whereas Bamb_5911 is involved in the regulation of the cluster, as it is included in a two-gene operon coding LuxR-type transcriptional regulators (Mahenthiralingam et al. 2011). Indeed, the mutant trBamb_5925 is totally impaired in B. multivorans inhibition, while trBamb_5911 is slightly restored when C8-HSL is added, revealing that the second LuxR-type transcriptional regulator could partially activate the biosynthesis of enacyloxins (Fig. 6B). Although several elements indicate that the biosynthesis of enacyloxins is QS-regulated (Mahenthiralingam et al. 2011), our data support this assertion. The remaining genes of panel B are almost all orthologs of those implicated in occidiofungins biosynthesis. As discussed above, these antifungal compounds initially identified in B. contaminans MS14 have recently been described as hemolytic molecules in B. vietnamiensis DBO1 (Lu et al. 2009; Gu et al. 2011; Thomson and Dennis 2012). Another team has identified antifungal molecules named burkholdines in B. ambifaria 2.2N, which have structures similar to occidiofungins, demonstrating that the occidiofungin cluster of B. ambifaria is expressed (Tawfik et al. 2010). In B. contaminans, while the cluster contains two LuxR-type regulators that have C-terminal domains able to bind DNA, neither the autoinducer-binding domain nor the response regulatory domain in N-terminal have been identified, suggesting that the signal molecule is from another nature (Gu et al. 2009). Yet, the results of our screening lead us to conclude that the production of occidiofungins is QS-controlled, at least in B. ambifaria.
In conclusion, we have confirmed in the clinical B. ambifaria HSJ1 strain some genes and phenotypes already known to be QS-regulated in Bcc species, and we have furthermore identified new QS-regulated genes. Predominantly, the production of antifungal/antimicrobial compounds is a very important trait controlled by QS in the HSJ1 WT strain, as our study revealed genes implicated in the biosynthesis of pyrrolnitrin, enacyloxins, and occidiofungins. This arsenal could appear redundant, but each molecule is effective against a different spectrum of microorganisms. Interestingly, a recent study reported that B. cepacia strains produced HMAQs displaying antifungal properties (Kilani-Feki et al. 2011). We have previously reported that at least three species of Burkholderia, including the B. ambifaria strain used in this study, are able to produce HMAQs, and that mutant deficient in the biosynthesis of HMAQs produces increased concentrations of C8-HSL (Vial et al. 2008). Burkholderia ambifaria HSJ1 expresses thus at least four molecules with antifungal/antimicrobial properties, three of them being QS-regulated; studies are currently underway to determine if biosynthesis of the fourth family of molecules, namely HMAQ, is also regulated by QS.