Genetic analysis of xenocoumacin antibiotic production in the mutualistic bacterium Xenorhabdus nematophila

Authors


*E-mail sforst@uwm.edu; Tel. (+1) 414 229 6373; Fax (+1) 414 229 3926.

Summary

Xenocoumacin 1 (Xcn1) and xenocoumacin 2 (Xcn2) are the major antimicrobial compounds produced by Xenorhabdus nematophila. To study the role of Xcn1 and Xcn2 in the life cycle of X. nematophila the 14 gene cluster (xcnA–N) required for their synthesis was identified. Overlap RT-PCR analysis identified six major xcn transcripts. Individual inactivation of the non-ribosomal peptide synthetase genes, xcnA and xcnK, and polyketide synthetase genes, xcnF, xcnH and xcnL, eliminated Xcn1 production. Xcn1 levels and expression of xcnA–L were increased in an ompR strain while Xcn2 levels and xcnMN expression were reduced. Xcn1 production was also increased in a strain lacking acetyl-phosphate that can donate phosphate groups to OmpR. Together these findings suggest that OmpR-phosphate negatively regulates xcnA–L gene expression while positively regulating xcnMN expression. HPLC-MS analysis revealed that Xcn1 was produced first and was subsequently converted to Xcn2. Inactivation of xcnM and xcnN eliminated conversion of Xcn1 to Xcn2 resulting in elevated Xcn1 production. The viability of the xcnM strain was reduced 20-fold relative to the wild-type strain supporting the idea that conversion of Xcn1 to Xcn2 provides a mechanism to avoid self-toxicity. Interestingly, inactivation of ompR enhanced cell viability during prolonged culturing.

Introduction

Xenorhabdus nematophila, a member of the family Enterobacteriaceae, engages in a mutualistic association with the entomopathogenic nematode Steinernema carpocapsae and is also pathogenic towards different insect hosts (Poinar, 1979; Forst and Clarke, 2002; Herbert and Goodrich-Blair, 2007; Snyder et al., 2007). X. nematophila is transmitted by the infective juvenile stage (IJ) of the nematode into an insect host through natural openings such as the mouth or anus. The bacterial–nematode complex invades the insect's body cavity (haemocoel) and kills the host. The nematode subsequently reproduces in the haemocoel feeding on both X. nematophila and the nutrients derived from insect sources. With the depletion of nutrient supplies the nematodes develop into the IJ stage, which is colonized by its Xenorhabdus partner. The IJs emerge from the insect cadaver into the soil where they search for a new insect host.

As the IJ penetrates the intestinal wall to gain access to the nutrient-rich haemolymph microbial competitors derived from the insect gut enter the haemocoel (Isaacson and Webster, 2002; Walsh and Webster, 2003; Gouge and Snyder, 2006; van der Hoeven et al., 2008). Saprophytic microbes from the soil as well as bacteria adhering to the cuticle of the nematode represent other potential sources of competitors that can grow within the insect cadaver. It was recently shown that bacteria present in the insect gut appeared in the haemolymph soon after nematode invasion (Gouge and Snyder, 2006). These competitors were eliminated as X. nematophila became the dominant microbial species in the haemolymph.

To successfully compete for nutrient resources of the insect X. nematophila produces several water-soluble and non-polar antimicrobial compounds (Paul et al., 1981; Akhurst, 1982; McInerney et al., 1991; Sundar and Chang, 1993; Li et al., 1997; Webster et al., 2002), phage-derived bacteriocins (Thaler et al., 1995) and colicin E3-type killer proteins (Singh and Banerjee, 2008). The water-soluble peptide antimicrobial compounds, xenocoumacin 1 (Xcn1) and xenocoumacin 2 (Xcn2), are the major antibiotics produced in broth culture by X. nematophila strain All (McInerney et al., 1991). Both Xcn1 and Xcn2 were also shown to be produced in the haemocoel of Xenorhabdus-infected insect cadavers (Maxwell et al., 1994). Xcn1 is active against Gram-positive and Gram-negative bacteria and several fungal species while Xcn2 is less active against the bacteria and inactive towards the fungal species examined (McInerney et al., 1991).

Xcn1 and Xcn2 are composed of an arginine residue, a leucine residue and four acetate units creating a benzopyran-1-one (isocoumarin) ring structure (McInerney et al., 1991; Fig. 1A). As the xenocoumacins are hybrids of amino and carboxylic acid moieties, non-ribosomal peptide synthetases (NRPSs) and polyketide synthtases (PKSs) (Crosa and Walsh, 2002; Finking and Marahiel, 2004) are most likely involved in their biosynthesis. These large enzyme complexes were originally found in Bacillus and Streptomyces species that are known for the production of a diverse array of secondary metabolites. Several other genera including Gram-negative bacteria such as Burkholderia, pseudomonads and myxobacteria have also been shown to be rich in these enzymes as can be deduced from various genome sequencing projects. In contrast, NRPS and PKS enzymes in the Enterobacteriaceae are more often involved in the production of iron chelating siderophores and not in the biosynthesis of antibiotic compounds (Crosa and Walsh, 2002). Recently, a vibriobactin-like siderophore (photobactin) and a cytotoxic NRPS-derived peptide have been characterized in Photorhabdus species, the sister taxa to Xenorhabdus (Ciche et al., 2003; Waterfield et al., 2008).

Figure 1.

A. Structures of Xcn1 and Xcn2 (from McInerney et al., 1991). Xcn1 and Xcn2 are benzopyran structures derived from acetate units, a leucine residue and an arginine residue. Xcn1 contains a guanidinium group of arginine while this group is absent in Xcn2.
B. Characterization of xenocoumacin cluster. Operon organization of the 14 xcn genes (xcnA–N) involved in Xcn1 and Xcn2 synthesis. NRPS and PKS genes are shown in black (see Table S1). Other annotated xcn genes are shown as grey boxes and unannotated genes are shown as white boxes. Six major transcriptional units identified by overlap RT-PCR are indicated by open arrows underneath the genes.

While antimicrobial compounds have been extensively used as therapeutic agents their role in the natural biology of the microbe producing them is not well understood. Although some of these compounds are clearly made to kill competitors, there is increasing evidence that some compounds might also serve as signalling molecules (Yim et al., 2007). The major antimicrobial compound, isopropylstilbene, produced by Photorhabdus luminescens, was shown to inhibit insect immune responses and also serve as a developmental signal for the infective juvenile nematode partner (Eleftherianos et al., 2006; Joyce et al., 2008). In this respect, insect–nematode–bacterial tripartite associations such as those involving Xenorhabdus species provide attractive systems for both discovery of new natural products and identification of novel compounds involved in interkingdom signalling.

During the investigation of the role of the master flagella regulator, FlhDC, in the co-ordinate regulation of motility, exoenzyme and antibiotic reproduction in X. nematophila we found that antibiotic production and expression of a NRPS gene (xcnA) located adjacent to a cluster of flagellar genes were elevated by inactivation of either the response regulator gene, ompR, or the cognate histidine kinase gene, envZ (Park and Forst, 2006). Since xenocoumacin is the major antibiotic produced by X. nematophila these findings suggested that the NRPS encoded by xcnA may be involved in xenocoumacin synthesis. OmpR is a global response regulator involved in the regulation of outer membrane porin genes, various transporters, virulence genes, flagella and curli fibres (Mattison et al., 2002; Feng et al., 2003; Goh et al., 2004; Jubelin et al., 2005; Park and Forst, 2006; van der Hoeven and Forst, 2009).

Recent results from detailed feeding experiments revealed that the pyrrolidine ring of Xcn2 is formed from the guanidinium moiety of Xcn1 (Reimer et al., 2009). Initial genetic analysis also identified two NRPS genes (xcnA and xcnK) required for production of Xcn1 and two genes (xcnM and xcnN) involved in the conversion of Xcn1 to Xcn2 (Reimer et al., 2009). In the present study we show that individual inactivation of three PKS genes (xcnF, xcnH and xcnL) as well as the two NRPS genes xcnA and xcnK eliminated xenocoumacin synthesis and dramatically reduced total antibiotic activity. In contrast, inactivation of xcnM markedly increased Xcn1 production, eliminated Xcn2 production and reduced cell viability 20-fold suggesting that conversion of Xcn1 to Xcn2 provides a mechanism to avoid self-toxicity. The xcn genes were expressed on several separate monocistronic and polycistronic mRNA species. OmpR-phosphate was shown to negatively regulate xcnA–L and Xcn1 synthesis during exponential growth and positively regulate the conversion of Xcn1 to Xcn2 later in the growth phase. These findings represent the first example of regulation of secondary metabolism genes by OmpR.

Results

Identification of genes involved in xenocoumacin production

Xcn1 and Xcn2 (Fig. 1A) are major antibiotics produced by X. nematophila in broth culture (McInerney et al., 1991) and in insects (Maxwell et al., 1994). The NRPS gene, xcnA, previously designated nrps1, was shown to be highly expressed in an ompR strain that displayed a high level of antibiotic activity (Park and Forst, 2006). The genome of X. nematophila has been recently sequenced (http://www.genoscope.cns.fr/agc/mage/) allowing us to locate xcnA in a 39 kb region containing 14 genes (Fig. 1B and Table S1). Based on activities of characterized enzymes the 39 kb region contains two NRPS genes (xcnA and xcnK), three PKS genes (xcnF, xcnH and xcnL), three genes encoding enzymes predicted to be involved in hydroxymalonyl CoA synthesis (xcnBCE), a gene encoding a type II thioesterase (xcnI) that may be involved in clearing intermediates from misprimed NRPS and PKS enzymes and genes annotated as β-lactamase (xcnG), saccharopine dehydrogenase (xcnM) and fatty acid desaturase (xcnN). The xcnD and xcnJ genes are predicted to encode small proteins, 85 and 107 residues, respectively, of unknown function.

Distance tree analysis revealed that the closest orthologue of xcnA was found in a γ-proteobacterium (Table S1), the same taxonomic group as X. nematophila. The other xcn genes were orthologous to genes from different phylogenetic lineages; xcnB–L (Firmicutes), xcnM (cyanobacteria) and xcnN (β-proteobacteria). The GC content of xcnA is 46%, the same as the X. nematophila genome, while the GC content for the xcnB–N genes ranges between 31% and 42% (Table S1). These findings suggest that the xcnB–N cluster is a mosaic of genes that X. nematophila acquired by lateral transfer.

Analysis of xcn gene expression

RT-PCR analysis was performed to assess whether all of the xcn genes were expressed in cells grown under LB broth conditions. Figure 2 shows that all xcn analysed were expressed under these conditions. To address the question of whether the xcn genes were encoded on a single polycistronic mRNA or several independently transcribed mRNAs, primers designed to span the region between the 3′ end of the upstream gene and the 5′ end of the adjacent downstream gene were used for RT-PCR analysis (Fig. 3). Prominent RT-PCR products were obtained for the xcnB–C and xcnC–E primer pairs indicating that a major polycistronic transcript was present for xcnBCDE. Prominent RT-PCR products were also obtained for the xcnG–H and xcnH–I primer pairs, the xcnK–L primer pair and xcnM–N primer pair. In contrast, RT-PCR products were barely detectable for the xcnE–F and xcnF–G primer pairs suggesting that xcnF is expressed as a monocistronic mRNA. Similarly, the RT-PCR product generated with the xcnA–B primer pair was present at very low levels suggesting that xcnA was predominantly expressed as a monocistronic mRNA. A transcript spanning xcnL–M was also detectable at low levels. Thus, it appears that the xcn cluster is expressed on several separate monocistronic and polycistronic mRNA species that encode different regions of the xcn cluster (Fig. 1B).

Figure 2.

Analysis of xcn gene expression in the wild-type strain. The transcript level of each xcn gene was determined by RT-PCR.

Figure 3.

Analysis of transcriptional units in the xcn gene cluster by overlap RT-PCR. Primers were designed to span the intergenic region between two adjacent genes. The top gel shows RT-PCR product for each of the overlap primer sets. Control PCR products using X. nematophila genomic DNA as a template are shown below each RT-PCR reaction.

Insertional inactivation of xcn genes

To determine whether the xcn genes encoding NRPSs (xcnA and xcnK) and PKSs (xcnF, xcnH and xcnL) were involved in xenocoumacin production each of the five genes was disrupted by insertional inactivation. HPLC-MS analysis of cell-free culture supernatants revealed that Xcn1 and Xcn2 were not produced in any of the mutant strains (Fig. S1). Antibiotic activity was assessed using an overlay plate assay with Micrococcus luteus as the indicator strain (Fig. 4). All five xcn strains displayed a dramatic decrease in antibiotic activity relative to the parent strain. Residual antibiotic activity was presumably due to the production of non-xenocoumacin antibiotics. Furthermore, inactivation of xcnB, the first gene in the operon involved in the synthesis of hydroxymalonyl CoA, dramatically reduced antibiotic production (Fig. 4). Inactivation of xcnI minimally affected antibiotic production indicating that the thioesterase encoded by xcnI was not required for cleavage of the final product from the enzyme complex (Fig. 4).

Figure 4.

Antibiotic activities of wild-type and xcn::Cm strains. Xcn activity was analysed in the respective xcn::Cm mutant strains by an overlay assay using Micrococcus luteus as the indicator. Activity is visible as zones of inhibition surrounding the colonies of the various xcn mutant strains.

The finding that xcnA and xcnF were primarily expressed as monocistronic mRNAs predicted that inactivation of these genes would not be polar on adjacent downstream genes. As expected, xcnB and xcnG were expressed in the xcnA::Cm and xcnF::Cm strains respectively (Fig. 5). In contrast, inactivation of xcnH in an operon of xcnGHIJ was polar on xcnI. Since antibiotic production was minimally affected in the xcnI strain absence of Xcn production in the xcnH mutant was due to inactivation of xcnH rather than a polar effect on xcnI. Inactivation of xcnI was not polar on xcnK as predicted by the overlap RT-PCR results. Finally, xcnL transcripts were detected in the xcnK strain suggesting that an internal promoter may be located in the 104 bp xcnK–L intergenic region and that the absence of Xcn1 production in the xcnK strain was likely due to the loss of the NRPS encode by xcnK rather than a polar effect on xcnL. Together, these findings indicate that NRPS and PKS genes of the xcn cluster as well as xcnBCDE are necessary for Xcn1 production and that xenocoumacin was the major antibiotic activity detected in the overlay assay.

Figure 5.

RT-PCR analysis of polar effects on downstream xcn genes. The mutant strain from which the RNA was derived is shown on the top of each panel. The downstream gene that was analysed is identified on the right side of the panel.

Analysis of Xcn1 and Xcn2 production

While the levels of Xcn1 and Xcn2 were determined previously in 48 h broth cultures (McInerney et al., 1991) their production during earlier stages of growth had not been studied. Furthermore, whether Xcn1 and Xcn2 were produced sequentially or simultaneously remained unknown. An HPLC-MS approach was taken to assess the temporal production of Xcn1 and Xcn2. In the wild-type strain, Xcn1 levels were detectable during exponential growth (6 h), increased twofold during early stationary phase (12 h) and increased slightly (1.3-fold) during late stationary (24 h) phase (Table 1). In contrast, Xcn2 levels were barely detectable during exponential growth (6 h), increased ∼10-fold during early stationary phase (12 h) and continued to increase (sixfold) during late stationary phase (24 h). These findings were consistent with the hypothesis that Xcn1 was produced initially and subsequently converted to Xcn2 later in the growth phase.

Table 1.  Xcn1 and Xcn2 production in wild-type, xcnM and ompR strains.
StrainTime (h)ODXcn1
Arbitrary units OD−1
Xcn2
Arbitrary units OD−1
  1. Experiments were performed in triplicate. Values represent mean and standard deviation in parentheses.

wt61.093.76 (0.58)0.38 (0.12)
125.817.66 (0.61)3.11 (0.16)
249.6410.33 (1.82)19.47 (3.12)
488.588.82 (0.30)15.47 (0.83)
xcnM61.501.79 (0.68)0.01 (0.01)
126.6310.0 (0.30)0.03 (0.03)
247.5013.5 (1.52)0.05 (0.01)
485.9716.7 (1.33)0.26 (0.21)
ompR61.007.36 (1.15)0.19 (0)
126.1117.70 (1.45)3.14 (0.31)
248.5923.00 (4.40)5.11 (0.99)
488.2628.41 (4.45)4.92 (0.76)

xcnMN is required for conversion of Xcn1 to Xcn2

During analysis of xcn genes we found that inactivation of xcnM and xcnN resulted in a marked increase in antibiotic activity (Fig. 6). In addition, quantitative RT-PCR (qRT-PCR) analysis of the expression of xcnN as a function of growth showed elevated expression during transition-phase and stationary-phase growth (D. Park, unpubl. data). To elucidate the role of xcnMN in Xcn1 and Xcn2 production culture supernatants of the xcnM strain were analysed by HPLC-MS. Xcn1 was produced at significantly higher levels than the wild-type strain during stationary phase and Xcn2 was not produced in the xcnM strain (Table 1). These findings support the idea that the xcnA–L genes are responsible for Xcn1 production early in growth and the xcnMN genes are required for the subsequent conversion of Xcn1 to Xcn2. These findings also suggest that Xcn1 is the predominant antibiotic activity measured in overlay assay.

Figure 6.

Antibiotic activities of the xcnM, xcnN and ompR strains. Xcn activity was analysed by an overlay assay using Micrococcus luteus as the indicator. Activity is visible as zones of inhibition surrounding the colonies of the various xcn::Cm mutant strains.

Increased production of Xcn1 in the ompR strain

The increased antibiotic activity of the ompR strain (Fig. 6; Park and Forst, 2006) suggested that Xcn1 was produced at elevated levels in this strain. To address this possibility Xcn1 and Xcn2 production in the ompR strain was measured by HPLC-MS. Xcn1 production was twofold higher during exponential growth (6 h) and 3.3-fold higher at 48 h in the ompR strain relative to the parent strain (Table 1). Xcn2 production was barely detectable during exponential growth and remained at significantly reduced levels relative to the parent strain over the 48 h growth period suggesting a lower rate of Xcn1 to Xcn2 conversion in the ompR strain.

We had previously shown that xcnA expression was elevated in the ompR strain grown for 6 h on 0.8% LB agar (Park and Forst, 2006). These conditions were found to be optimal for detection of negative regulation of xcnA by OmpR. To further analyse the role of OmpR in xenocoumacin production xcn transcript levels were compared in the parent and ompR strains grown on agar. RT-PCR analysis revealed that the level of mRNA for the xcnA–xcnL genes was significantly higher in the ompR strain (Fig. 7A). To assess the regulation of xcnM and xcnN by OmpR total RNA was extracted from cells grown to late exponential phase under standard LB broth conditions. RT-PCR analysis revealed that both xcnM and xcnN were expressed at lower levels in the ompR strain (Fig. 7B) suggesting these genes are either directly or indirectly positively regulated by OmpR. These finding were supported by qRT-PCR analysis showing that xcnA expression increased 64% while xcnM and xcnN expression decreased 1.9- and 2.5-fold, respectively, in the ompR strain (Table 2). Thus, the 3.3-fold increase in Xcn1 production in the ompR strain was likely due to the combined effect of elevated expression of the xcnA–L genes and reduced xcnMN expression.

Figure 7.

A. RT-PCR analysis of xcn genes in wild-type and ompR strains. Total RNA was obtained from the wild-type and ompR strains grown for 6 h on 0.8% LB agar plates. RT-PCR products from the wild-type and ompR strains for each xcn gene are shown in the first and second lane respectively.
B. RT-PCR analysis of xcnM and xcnN in the ompR strain. Total RNA was obtained from the wild-type and ompR strains during exponential growth in LB broth. RT-PCR products from the wild-type and ompR strains for each xcn gene are shown in the first and second lane respectively.

Table 2. xcnA, xcnM and xcnN expression in wild-type and ompR strains.
GeneCtFold change
(ompR/wt)
wtompR
  • a. 

    Values represent mean and standard error in parentheses.

xcnA3.99 (0.22)a3.28 (0.19)1.64 (0.06)
xcnM3.46 (0.25)4.39 (0.09)0.54 (0.08)
xcnN2.62 (0.14)3.92 (0.24)0.39 (0.09)

Role of OmpR-phosphate in xcn gene regulation

Antibiotic production was previously shown to be elevated in an envZ strain of X. nematophila (Park and Forst, 2006). Since EnvZ is the cognate histidine kinase that phosphorylates OmpR these findings suggested that OmpR-phosphate negatively regulates the xcnA–L genes. OmpR can also be phosphorylated by the small-molecular-weight phosphodonor, acetyl-phosphate (Shin and Park, 1995). Acetyl-phosphate is synthesized from acetyl-CoA and inorganic phosphate by phosphotransacetylase encoded by the pta gene and is converted to acetate and ATP by acetyl kinase encoded by ackA. Acetyl-phosphate levels can increase up to fivefold in an ackA strain and are eliminated in an ackA-pta strain (Klein et al., 2007; Keating et al., 2008). To assess the effect of acetyl-phosphate on the regulation of antibiotic production ackA and ackA-pta deletion strains were constructed. The zone of inhibition in the overlay assay was reduced in the ackA strain (20 mm) relative to the wild-type strain (28 mm). Conversely, the zone of inhibition was elevated in the ackA-pta strain (32 mm) which lacks acetyl-phosphate. These findings are consistent with the hypothesis that acetyl-phosphate can donate phosphate to OmpR (Wolfe, 2005) and supports the idea that OmpR-phosphate negatively regulates Xcn1 production.

The xcnM strain displays reduced viability during prolonged growth

The above findings raised the question of the biological relevance of the conversion of Xcn1 to Xcn2. It was shown previously that Xcn1 was a more potent antibiotic than Xcn2 (McInerney et al., 1991). We considered the possibility that high concentrations of Xcn1 may negatively affect viability and induce self-toxicity. To assess this possibility we compared the growth rate, final cell density and viability during prolonged growth of the wild-type and xcnM strains. The growth rate and final cell density of the xcnM strain after 18 h were indistinguishable from that of the wild-type strain (data not shown). To assess whether prolonged incubation affected cell viability cultures were monitored over a 72 h time-course. Cell viability of the xcnM strain incubated for 24 and 48 h was not significantly different from the wild-type strain (Fig. 8A). However, after 60 h viability of the xcnM strain was 30% that of the wild-type strain. At 72 h viability of the xcnM strain was reduced more than 20-fold relative to the wild-type strain. In contrast, the cell viability of the xcnA strain during prolonged growth was not significantly different from the wild-type strain (D. Park, S. Singh and S. Forst, unpubl. data). These findings suggested that X. nematophila was sensitive to the higher levels of Xcn1 produced in the xcnM strain and that the conversion of Xcn1 to Xcn2 provides a mechanism to avoid self-toxicity.

Figure 8.

A. Comparison of cell viability of the wild-type and xcnM strains during prolonged growth. The wild-type strain (black bar) and xcnM strain (white bar) were grown in LB broth and cell viability as measured by colony-forming units (cfu) was determined in triplicate at the indicated time points. Regression analysis of the slopes derived from the 48, 60 and 72 h data indicated a significant difference (P-value < 0.001) between the wild-type and xcnM strains.
B. Comparison of cell viability of the wild-type, xcnM, ompR and xcnMompR strains during prolonged growth. The wild-type (black bar), xcnM (white bar), ompR (light grey bar) and xcnMompR (dark grey bar) strains were grown in LB broth and cell viability as measured by colony-forming units (cfu) was determined in triplicate at the indicated time points.

The ompR strain also produced high levels of Xcn1 but unlike the xcnM strain, still produced some Xcn2. To further address the question of whether elevated Xcn1 production affected viability, the ompR strain was cultured for 72 h as above. To assess the viability of a strain that produced elevated levels of Xcn1 but no Xcn2 an xcnM-ompR strain was constructed. During the first 48 h of culturing the viability of the ompR and xcnM-ompR strains was closely similar to the wild-type strain (data not shown). By 60 h the viability of the xcnM strain had decreased 75% while the viability of both the ompR and xcnM-ompR strains was similar to the wild-type strain (Fig. 8B). These differences were more dramatic at 72 h at which time the viability of the xcnM strain was 7% of the wild-type strain while the viability of the ompR and xcnM-ompR strains was approximately 14-fold and 10-fold greater than the wild-type strain respectively. Thus, inactivation of ompR rescued the reduced viability of the xcnM strain and generally enhanced the viability of X. nematophila during prolonged culturing.

Discussion

To better understand the role of xenocoumacin in the life cycle of X. nematophila we identified and genetically analysed the 14 gene xcn cluster required for Xcn1 and Xcn2 synthesis. Inactivation of xcnA, xcnK, xcnF and xcnL eliminated production of Xcn1 and was not polar on downstream genes. Xcn1 production was also eliminated by disruption of xcnH that was polar on the downstream gene, xcnI. However, xcnI encodes a type II thioesterase that was not required for Xcn1 production. In addition, Xcn1 production was eliminated by inactivation of xcnB involved in the synthesis of hydroxymalonyl CoA. In contrast, Xcn1 levels increased and Xcn2 production was eliminated in the xcnM strain. Similarly, Xcn1 production was increased in the xcnN strain. Together, these findings indicate that the xcnA, B, F, H, K and L genes are involved in the synthesis of Xcn1 while xcnMN are involved in the conversion of Xcn1 to Xcn2.

Overlap RT-PCR analysis revealed that six separate xcn transcripts and other minor transcripts were produced during late exponential growth. This transcriptional organization is distinct from NRPS–PKS biosynthetic clusters found in other γ-proteobacteria. For example, the pigment–antibiotic compound prodigiosin of Serratia marcescens, synthesized by the NRPS–PKS pig operon, is encoded on a major polycistronic transcript controlled by a single σ70 promoter (Williamson et al., 2006).

Xcn1 was shown to be produced at low levels during early exponential growth and increased as cells transitioned to late exponential phase. xcnA expression has also been found to increase later in the growth phase (D. Park, unpubl. data). In addition, xcnA–L expression was elevated in the ompR strain and Xcn1 production was increased in the ompR, envZ and ackA-pta strains while production was decreased in the ackA strain. Together, these findings support a model in which OmpR-phosphate either directly or indirectly negatively regulates xcnA–L during exponential growth. To further define the transcriptional organization of the xcn cluster 5′RACE analysis was carried out. A transcriptional start site and consensus σ70 promoter were identified 168 bp upstream of the AUG start codon of xcnA while start sites for other transcriptional units could not be established using this approach (D. Park, unpubl. data).

OmpR was previously shown to co-ordinately repress flagella synthesis and exoenzyme production in X. nematophila (Park and Forst, 2006) by negatively regulating the flhDC operon. FlhDC activates the class II flagella genes that includes fliAZ required for both flagellin and exoenzyme gene expression (Givaudan and Lanois, 2000; Park and Forst, 2006; Lanois et al., 2008). Unlike these functions, antibiotic production and xcn gene expression were not reduced in a flhC mutant strain (D. Park, unpubl. data) suggesting that OmpR may directly regulate the xcn genes. At present purified X. nematophila OmpR is not available to test this possibility. The repression of xcn genes in X. nematophila would represent the first example of regulation of secondary metabolite production by OmpR. During early stages of infection X. nematophila colonizes the connective tissue and musculature surrounding the insect midgut, proliferates and produces virulence factors (Morgan et al., 2001; Sicard et al., 2005). Co-ordinate repression of energy expensive processes such as flagella synthesis, exoenzyme and antibiotic production may confer an adaptive advantage during early stages of infection.

While OmpR differentially regulates more than 100 genes in Escherichia coli (Oshima et al., 2002), and controls virulence genes and stationary-phase acid tolerance genes in Salmonella typhimurium (Lee et al., 2000; Bang et al., 2002) it has not been shown to both positively and negatively regulate separate genes within a single biosynthetic cluster. In the present study, OmpR in X. nematophila was found to negatively regulate xcnA–L and positively regulate xcnMN expression. OmpR usually functions as a repressor when it binds to sequences near the start of transcription and functions as an activator by binding to regions further upstream of the promoter (Feng et al., 2003; Goh et al., 2004; Rhee et al., 2008). One possible mechanism for the differential regulation of the xcn gene cluster is that OmpR binds near the start of transcription for the monocistronic and polycistonic mRNAs of the xcnA–L cluster and upstream of the promoter for xcnMN. Alternatively, OmpR may function indirectly by controlling a repressor that regulates the xcnA–L cluster and an activator that controls the expression of xcnMN. Purification of active X. nematophila OmpR and the mapping of the start of transcription and promoters for the xcn gene cluster will allow us to address these questions.

Inactivation of xcnM concomitantly elevated production of Xcn1, eliminated Xcn2 production and increased the zone of inhibition in the overlay assay. Together these findings support the idea that the increased antibiotic activity observed in the overlay assay was due to increased Xcn1 production. The contribution of Xcn2 to the antibiotic activity measured in the overlay assay is difficult to assess since it cannot be produced in the absence of Xcn1. Purified Xcn2 is less active than Xcn1 against the bacterial and fungal species tested to date (McInerney et al., 1991). The primary structural difference between Xcn1 and Xcn2 is the absence of the guanidinium group and presence of the pyrollidine ring in the latter structure which could effectively decrease its solubility relative to Xcn1. The higher solubility may in part account for higher antibiotic and antifungal activity of Xcn1 and its greater diffusibility in the overlay assay. In addition, Xcn1 may possess a higher intrinsic antimicrobial activity than Xcn2. The temporal regulation of Xcn1, which is produced earlier than Xcn2, together with the differences in the structural properties and biological activities of Xcn1 and Xcn2 suggests these antimicrobial compounds play distinct roles in the life cycle of X. nematophila.

Here we show that increased production of Xcn1 in the xcnM strain was correlated with a 20-fold loss in viability raising the possibility that elevated Xcn1 levels exceeded a threshold for resistance. This loss of viability of the xcnM strain during prolonged growth suggests that the conversion of Xcn1 to Xcn2 provides a mechanism to avoid self-toxicity. During early stages of infection bacteria derived from the insect gut predominate in the haemolymph. As X. nematophila proliferate to high levels the competitor population declines (Gouge and Snyder, 2006) presumably as a result of increased Xcn1 production. Thus, Xcn1 and Xcn2 production is balanced between the requirement to produce sufficient concentrations of Xcn1 for suppression of competitors and the need to maintain Xcn1 levels below a threshold of self-toxicity. The differential expression of the xcnA–L and the xcnMN genes may therefore be part of the adaptive response to optimize fitness during growth in the competitive haemocoelic environment. In addition, we found that inactivation of ompR rescued the reduced viability of the xcnM strain and enhanced viability of X. nematophila during prolonged culturing. Since OmpR positively regulates the outer membrane porins OpnP and OpnS inactivation of ompR could enhance viability during prolonged incubation by reducing outer membrane permeability to Xcn1. Alternatively, since OmpR functions as a global regulator, inactivation of ompR may alter adaptive responses that affect the sensitivity to Xcn1. Finally, how modulation of OmpR-phosphate levels controls xcn gene expression remains to be determined. The question also arises whether Xcn2 possesses biological functions other than the antimicrobial activity already identified. The insect–nematode–X. nematophila model system will allow us to address these questions and help to further our understanding of the role of antimicrobial compounds in interspecies competition and the natural biology of microorganisms.

Experimental procedures

Bacterial strains, media and growth conditions

Strains and plasmids used in this study are listed in Table 3. Cells were routinely grown on Luria–Bertani (LB) broth (1.0% Bacto trypton, 0.5% yeast extract, 0.5% NaCl) supplemented with 0.01 mM MgSO4, by shaking at 150 r.p.m. or on corresponding solid agar media (15 g l−1 agar). When required, ampicillin, chloramphenicol and kanamycin were added to a final concentration of 50, 25 and 30 μg ml−1 respectively. X. nematophila was cultured at 30°C, and E. coli was grown at 37°C. Unless otherwise noted, strains were initially grown overnight in 2 ml of LB selective broth, inoculated into 5–10 ml of fresh LB selective media in a 250 ml flask and incubated for desired period. Final bacterial cultures were normalized based on the optical density at 600 nm (OD600) and used for further analysis. Graces insect culture medium (Volgyi et al., 1998) was prepared as described by the manufacturer (Gibco).

Table 3.  Bacterial strains and plasmids used in this study.
Strain or plasmidRelevant genotype, phenotype or characteristic(s)Reference or source
  1. Amp, ampicilin resistance; Km, kanamycin resistance; Cm, chloramphenicol resistance; Tet, tetracycline resistance; Gm, gentamicin resistance.

Strains
 X. nematophila
  AN6/1Wild-type, phase I variant; AmpLaboratory stock
  ABR2AN6/1 ompR::KmForst and Boylan (2002)
  ackAAN6/1 ΔackA::KmR. van der Hoeven
  ackA-ptaAN6/1 ΔackA-pta::KmR. van der Hoeven
  xcnA::CmAN6/1 xcnA::CmThis study
  xcnB::CmAN6/1 xcnB::CmThis study
  xcnF::CmAN6/1 xcnF::CmThis study
  xcnH::CmAN6/1 xcnH::CmThis study
  xcnI::CmAN6/1 xcnI::CmThis study
  xcnK::CmAN6/1 xcnK::CmThis study
  xcnL::CmAN6/1 xcnL::CmThis study
  xcnM::CmAN6/1 xcnM::CmThis study
  xcnN::CmAN6/1 xcnN::CmThis study
  xcnM-ompRAN6/1 xcnM::Cm ompR::KmThis study
 E. coli
  XL-1 Blue MRF'recA1 endA1 gyrA96 thi-1 hsdR17 supE44 relA1 lac[F'proAB lacIqZΔM15 Tn10 (Tet)]Stratagene
  S17-λpirrecA, thi, pro, hsdR-M+. RP4-2Tc::Mu Km::Tn7 in the chromosomeLaboratory stock
Plasmids
 pSTBlue-1Cloning vector; Amp KmNovagen
 pUC19Cloning vector; AmpLaboratory stock
 pKnock-CmBroad-host-range suicide vector; Cm RP4 oriT oriR6KD. Saffarini
 pKnock-xcnAInternal fragment of xcnA cloned into pKnock-CmThis study
 pKnock-xcnBInternal fragment of xcnB cloned into pKnock-CmThis study
 pKnock-xcnFInternal fragment of xcnF cloned into pKnock-CmThis study
 pKnock-xcnHInternal fragment of xcnH cloned into pKnock-CmThis study
 pKnock-xcnIInternal fragment of xcnI cloned into pKnock-CmThis study
 pKnock-xcnKInternal fragment of xcnK cloned into pKnock-CmThis study
 pKnock-xcnLInternal fragment of xcnL cloned into pKnock-CmThis study
 pKnock-xcnMInternal fragment of xcnM cloned into pKnock-CmThis study
 pKnock-xcnNInternal fragment of xcnN cloned into pKnock-CmThis study
 pER2p15A oriR6K sacB Mob+ GmD. Saffarini
 pER2-ΔackA:KmackA flaking regions with Km cloned into pER2This study
 pER2-ΔackA-pta:KmackA-pta flaking regions with Km cloned into pER2This study

Construction of the xcn mutant strains

To construct mutant strains by insertional inactivation, primers for each gene were designed to amplify a 300–600 bp internal fragment located near the 5′ end of the gene. The amplified products were purified with PCR Clean Kit (Roche) and subsequently blunt end-ligated into the EcoRV site of pSTBlue-1 (Novagen). Ten recombinant colonies were selected and subsequently analysed by colony PCR using a T7 and SP6 primer pair from flanking region of EcoRV site of the vector to confirm the size of the cloned fragment. A colony having the desired plasmid was grown overnight and the recombinant plasmids were purified using spin column (Qiagen). A PstI–XbaI fragment containing either xcnA, xcnB, xcnF, xcnH, xcnI, xcnK, xcnL, xcnM or xcnN was gel-purified and ligated into the conjugal suicide vector pKnock-Cm (Alexeyev, 1999). The resultant plasmids were transformed into E. coli S17-λpir and conjugally transferred into the wild-type strain of X. nematophila. In addition, the xcnM pKNOCK plasmid was conjugally transferred into the ompR deletion strain, ABR2, to create the xcnMompR mutant strain. Selection on ampicillin and chloramphenicol identified mutant cells in which the recombinant pKnock-Cm had integrated into the chromosome by single-cross-over homologous recombination. Gene disruption was confirmed by PCR. In addition, the absence of mRNA for the disrupted gene was verified by RT-PCR analysis using primers designed from the coding regions flanking the cloned region of each gene.

Construction of ackA and ackA-pta strains

Primers engineered with restriction sites (SacI and SphI) were used to PCR-amplify chromosomal fragments upstream and downstream of the ackA and ackA-pta genes. Fragments were cloned along with the kanamycin cassette into the SacI and SphI sites of cloning vector, pUC19. The resulting ΔackA:Km and ΔackA-pta:Km inserts were screened by PCR and cloned into the suicide vector pER2. These plasmids were transformed into E. coli S17-λpir and conjugally transferred into the wild-type strain of X. nematophila. Allelic replacements creating ΔackA:Km and ΔackA-pta:Km were confirmed by PCR and RT-PCR using ackA and pta internal primers.

Overlay assay for antibiotic activity

For measurement of antibiotic activity, Xenorhabdus strains were initially grown overnight in 2 ml of LB selective broth, inoculated into 5–10 ml of fresh LB media in a 250 ml flask and incubated for 18–20 h. Final bacterial cultures were normalized based on the OD600. Six-microlitre aliquots of Xenorhabdus culture were spotted onto LB agar plate and subsequently incubated for 15–24 h. The bacteria were then exposed to chloroform fumes for 30 min and air dried for 30 min. Five hundred microlitres of an overnight culture of either M. luteus or other indicator strains was added to 12 ml of top agar (0.7% agar) which was poured over the bacterial colonies (Akhurst, 1982; Volgyi et al., 1998). After incubation overnight, zones of growth inhibition surrounding bacterial colonies were observed. Experiments were performed in triplicates.

Quantification of Xcn1 and Xcn2

For quantification of Xcn1 and Xcn2 the following protocol was used: overnight cultures of all strains grown in LB with the required antibiotics were diluted to an OD600 of 0.05 in fresh LB (with antibiotics) in triplicates (10 ml in 50 ml flasks). After 6, 12, 24 and 48 h of cultivation at 30°C and 200 r.p.m., 1 ml of these cultures were taken, centrifuged to remove cells and the supernatant was diluted with MeOH to a final rate of 1:1 or 1:10 (supernatant/MeOH, v/v) following another centrifugation step in order to remove insoluble material. For the determination of Xcn1 and Xcn2 the ions m/z[M+H]+ 466.3 and 407.3 were quantified, respectively, using a UPLC system (Themo Scientific) connected to a Nanomate ESI-source (Advion) followed by a LTQ Orbitrap (Thermo Scientific). HPLC separation was performed using an AQUITY RP18 column from Waters (1.7 μm) and a water/acetonitrile gradient (+ 0.1% formic acid) (gradient: 0–14 min, 5–95% acetonitrile, MS: positive mode between 200 and 2000 m/z, injection volume: 5 μl).

RNA purification

Standard conditions for isolating total RNA from cells grown in LB medium was as follows: Xenorhabdus strains were initially grown overnight in 2 ml of LB selective broth, inoculated into 5–10 ml of fresh LB selective media in a 250 ml flask and incubated to desired cell density. For RT-PCR analysis, exponentially growing cells were used to extract total RNA. To assess the regulation of xcn genes by OmpR growth on 0.8% LB agar for 6 h after inoculation was found to be optimal (Park and Forst, 2006). To prepare total RNA from cells grown on 0.8% agar plates, 6 μl of 18 h bacterial broth culture was spotted (∼30 spots per plate) on the agar plate and grown for 6 h at 30°C. Bacterial colonies were collected using 2 ml of LB, pelleted and stored at −20°C. Total RNA was extracted with Trizol reagent (Sigma), following the standard protocol. Final RNA pellet was re-suspended with 100 μl of Nuclease-free distilled H2O. RNA concentration was determined by optical density at 260 nm. Triple measurements were performed for each RNA sample and an average value was obtained as a final concentration. Total RNA was digested with the RNase-free DNase reagent (Promega). This RNA (100 ng μl−1 after the DNase digestion) was used in 5′RACE-PCR and RT-PCR. For every RNA preparation, DNA contamination was assessed by performing a control PCR reaction prior to RT-PCR analysis.

RT-PCR and qRT-PCR analysis

RT-PCR was performed using AccessQuick RT-PCR system (Promega). The RT-PCR reaction (25 μl) contained the following components: 300 μg of total RNA, 20 pmol each of forward and reverse primer and 1 unit of reverse transcriptase. cDNA synthesis was conducted at 52°C for 45 min. The following cycle condition was used for PCR reaction: 30 s at 94°C, 30 s at 55°C and 60 s at 72°C for extension. Annealing temperature varied depending on which primer was used. For RT-PCR analysis, 21–23 cycles of PCR reaction was used for all xcn genes. For overlap PCR analysis, 25 cycles was used. 16S rDNA was used as the internal control gene to confirm that equal amounts of total RNA was used in each reaction. Primers specific to a single 16S rDNA gene was used in this study to increase the sensitivity and 18 cycles of PCR reaction were performed. For the analysis of polarity effects, RNA was obtained from strains in which xcn genes were inactivated and primers internal to the downstream gene were used to assess expression by RT-PCR.

Quantitative RT-PCR was performed using SuperScriptTM III Platinum Two-Step qRT-PCR Kit with SYBR® Green (Invitrogen). Equal amounts of DNase-treated total RNA (800 ng) were used to generate cDNA with random hexamer primers according to the manufacturer's protocol. The resultant cDNA was diluted eight times with Nuclease-free water. The diluted cDNA (2.5 μl) was subsequently used in 25 μl of qRT-PCR reaction, which was carried out in triplicate on cDNA with the SYBR Green master mix and DNA Engine Opticon® 2 thermal cycler. Sequences of primers used in qRT-PCR are shown in Table S2. As a negative control and a control to detect DNA contamination, water and DNase-treated RNA were used in place of cDNA template respectively. Cycle threshold (Ct) results and melting curve for each sample were generated by Opticon MonitorTM software, Version 1.0. The fold change in the amount of xcnA, xcnM and xcnN transcripts (target gene) relative to the recA transcripts (control gene) were determined by the following equation:

image

Final mean values of Δ(ΔCt) and fold change were obtained from three independent RNA samples.

Cell viability assays

Strains were grown overnight in 2 ml of LB broth with selection.

After 18 h of growth, cultures were diluted 1:5 in Graces media and normalized to equal absorbance (OD600). Five millilitres of LB broth was inoculated with 100 μl of normalized culture and growth was monitored by turbidity. Dilutional plating was performed in triplicate at 24, 48, 60 and 72 h for each of the cultures and colony-forming units (cfu) ml−1 was calculated. The cell viability experiments were repeated three times yielding highly similar results.

Acknowledgements

We would like to thank Ransome van der Hoeven for construction of the ackA and ackA-pta deletion strains, Erika Buell for assistance in making the xcnA::Cm mutant strain, Dr Jane Witten for providing Manduca sexta larvae and Dr John Berges for assistance with statistical analyses. We are grateful to Dr A. Wolfe for his critically reading of and valuable suggestions on this manuscript, and to the members of the Forst laboratory for valuable discussions. This work was supported by a Research Growth Initiative (RGI) grant from the University of Wisconsin-Milwaukee.

Ancillary