Xenorhabdus khoisanae SB10 produces Lys-rich PAX lipopeptides and a Xenocoumacin in its antimicrobial complex

Background Xenorhabdus spp. live in close symbiosis with nematodes of the Steinernema genus. Steinernema nematodes infect an insect larva and release their symbionts into the haemocoel of the insect. Once released into the haemocoel, the bacteria produce bioactive compounds to create a semi-exclusive environment by inhibiting the growth of bacteria, yeasts and molds. The antimicrobial compounds thus far identified are xenocoumacins, xenortides, xenorhabdins, indole derivatives, xenoamicins, bicornutin and a number of antimicrobial peptides. The latter may be linear peptides such as the bacteriocins xenocin and xenorhabdicin, rhabdopeptides and cabanillasin, or cyclic, such as PAX lipopeptides, taxlllaids, xenobactin and szentiamide. Thus far, production of antimicrobial compounds have been reported for Xenorhabdus nematophila, Xenorhabdus budapestensis, Xenorhabdus cabanillasii, Xenorhabdus kozodoii, Xenorhabdus szentirmaii, Xenorhabdus doucetiae, Xenorhabdus mauleonii, Xenorhabdus indica and Xenorhabdus bovienii. Here we describe, for the first time, PAX lipopeptides and xenocoumacin 2 produced by Xenorhabdus khoisanae. These compounds were identified using ultraperformance liquid chromatography, linked to high resolution electrospray ionisation mass spectrometry and tandem mass spectrometry. Results Cell-free supernatants of X. khoisanae SB10 were heat stable and active against Bacillus subtilis subsp. subtilis, Escherichia coli and Candida albicans. Five lysine-rich lipopeptides from the PAX group were identified in HPLC fractions, with PAX1’ and PAX7 present in the highest concentrations. Three novel PAX7 peptides with putative enoyl modifications and two linear analogues of PAX1’ were also detected. A small antibiotic compound, yellow in colour and λmax of 314 nm, was recovered from the HPLC fractions and identified as xenocoumacin 2. The PAX lipopeptides and xenocoumacin 2 correlated with the genes and gene clusters in the genome of X. khoisanae SB10. Conclusion With UPLC-MS and MSe analyses of compounds in the antimicrobial complex of X. khoisanae SB10, a number of PAX peptides and a xenocoumacin were identified. The combination of pure PAX1’ peptide with xenocoumacin 2 resulted in high antimicrobial activity. Many of the fractions did, however, contain labile compounds and some fractions were difficult to resolve. It is thus possible that strain SB10 may produce more antimicrobial compounds than reported here, as suggested by the APE Ec biosynthetic complex. Further research is required to develop these broad-spectrum antimicrobial compounds into drugs that may be used in the fight against microbial infections. Electronic supplementary material The online version of this article (10.1186/s12866-019-1503-x) contains supplementary material, which is available to authorized users.


Background
Xenorhabdus bacteria are in a species-specific association with Steinernema nematodes, i.e. a specific Steinernema sp. is associated with a specific Xenorhabdus sp. [1]. At the beginning of the Xenorhabdus-Steinernema life cycle, nematodes in the infective juvenile phase infect the insect host by entering the mouth, anus or respiratory spiracles [2]. Once inside the insect, the nematodes release the symbiotic bacteria by defecation. Steinernema nematodes produce proteins that suppress the insect's immune response, which allows Xenorhabdus to multiply [3]. The release of exoenzymes and toxins by both mutualists leads to septicaemia and the insect dies within 24 to 48 h after infection [4][5][6]. Nematodes reproduce sexually by going through phases J1 to J4 until resources are depleted, after which they return to the infective juvenile state, acquire symbionts and leave the cadaver in search of a new host.
Gualtieri and co-workers [16] identified five lysine-rich PAX peptides from X. nematophila F1/1 with strong antifungal activity against Fusarium oxysporum and several plant pathogenic fungi. Fuchs et al. [25] identified another 13 lysine-rich PAX peptides produced by X. nematophila HGB081. The PAX peptides have an identical backbone structure, with cyclization occurring between the ε-amino group of Lys 3 and the C-terminus. According to NMR studies, the fatty acids in PAX peptides from X. nematophila F1/1 are iso-branched [26], whilst those of X. nematophila HGB081 are straight-chain fatty acids [25]. Three NRPS genes (paxA, paxB and paxC) encode biosynthesis of the PAX peptides [25]. PAX peptides have antifungal and antibacterial activity, but they are not cytotoxic towards insects [25,26].

Results and discussion
Xenorhabdus bacteria are known to produce various antimicrobial compounds, but it is a highly neglected antimicrobial source that has not been exploited to its full potential. Although many genes relevant to antimicrobial compound biosynthesis have been identified in Xenorhabdus spp., the isolation, purification, identification and characterisation of antimicrobial compounds have not been done for all species belonging to this genus. Our antiSMASH [38] analysis of the genome of X. khoisanae SB10 revealed the presence of the xenocoumacin biosynthetic gene cluster and an APE Ec gene cluster (Fig. 1). The APE Ec gene cluster is widely distributed amongst prokaryotes and is related to secondary metabolites such as aryl polyenes [39]. Identification of the four modules coding for the PAX synthetase complex in the genome of X. khoisanae SB10 was accomplished by using tblastx (Table 1). This study, therefore, focussed on the first isolation and confirmation that specific antimicrobial compounds are indeed produced by X. khoisanae SB10. The following section describes the purification and characterisation of a selection of antimicrobial compounds from X. khoisanae SB10 cultures. A summary of the UPLC-MS data and identification of the compounds in the antimicrobial fractions are presented in Table 2. Detailed analysis utilising ESMS, UPLC-MS and MS e data on these fractions can be found in the Additional file 1 section.
The first chromatography of the SPC active fractions on 15 RPC resin by FPLC yielded fraction A1 with broad-spectrum antimicrobial activity (Fig. 2a). The SPC A1 fraction was shown to have broad-spectrum activity towards the four target organisms (B. subtilis, E. coli and C. albicans), as well as retaining activity after heating at 121°C for 20 min (results not shown). The latter result indicated marked heat stability of the antimicrobial compounds and eliminated activity related to labile, or volatile compounds and proteins such as proteases. The A1 fraction was further separated with C 18 -HPLC into seven peaks (Fig. 2b). Fractions B, D and G showed strong UV absorption ( Fig. 2b), suggesting that they did not only contain PAX peptides, but also xenocoumacin and breakdown products. Antimicrobial activity was observed in fractions A to G (Fig. 2b) and is shown in Fig. 3. We first focused on the three major UVabsorbing fractions, B, D and G, as well as C and F (Fig.  2b), for further analysis using high resolution UPLC-MS and UPLC-MS e (or MS/MS) analyses (refer to Additional file 1 and Table 2).
The major UPLC-MS peak observed in fraction B contained a small compound with a yellow colour (absorption maximum at 314 nm) and a monoisotopic M r of 406.2104 ( Fig. 4b and c, insert shows UV-spectrum). The UPLC-MS e analysis of the peak, containing its molecular ion with m/z 407.2176 at 4.27 min, yielded six mayor ions (Fig. 5a). The compound was identified as xenocoumacin 2 (expected m/z = 407.2182) based on the characteristic fragmentation pattern. The ion with m/z 250.1428 (Fig. 5a) represents the benzopyran-1-one fragment (expected m/z = 250.1443), while the ion with m/z 158.0788 (Fig. 5a) correlates to the remaining fragment with pyrrolidine as R group (expected m/z = 158.0817) [37]. Ions with m/z 176.0699, 190.0849, 215.1058 and 232.1324 (Fig. 5b) are the hydration and dehydration products of the two main fragments with m/z 158.0788 and 250.1428 (Fig. 5b). This identification of a xenocoumacin produced by X. khoisanae SB10, was supported by antiSMASH [39] results of 100% similarity between the genome of strain SB10 and the xenocoumacin biosynthetic gene cluster (refer to Fig. 1).
Fraction B also contained an earlier eluting peak at 2.90 min (refer to Fig. 4a) of a larger compound with m/ z 1052.7948 (M r = 1051.7870). The fragmentation pattern of the ion with m/z 1052.7948 showed a neutral loss of 128.09 from the major fragments, which is indicative of the loss of multiple Lys-residues from a peptide chain. The majority of ions with a neutral fragment loss also had resultant dehydration products. This dehydration is the consequence of a fragmentation reaction at the Cterminal of a Lys residue that leads to cyclisation in which the amino group of the Lys side chain participates, similar to fragmentation reactions found for ornithine-containing peptides [40]. Ions with m/z 129.1015 and 84.0799 also maps to Lys and its immonium ion. The ion spectrum of m/z 1052.7948 and proposed fragmentation pattern is presented in Fig. 6. From this spectrum we were able to map the sequence to a Lys-rich lipopeptide from the PAX peptide group. This particular peptide was identified as PAX1' with the R group as (3R)-3-hydroxytetradecanoyl coupled to Gly-Lys 2 -cyclo(Lys 4 ) with an expected M r of 1051.7845 (refer to Fig. 6 for structure). Although fraction B did not show  high activity against E. coli Xen 14, high activity was recorded against all other target strains when purified PAX1' and xenocoumacin were combined in a 1:1 ratio (Fig. 3). This result warrants further investigation to determine if the activity between PAX peptides and small water soluble xenocoumacins are synergistic.
With the identification of the PAX1' peptide there was a high probability of other PAX peptides in the HPLC fractions. In fraction C two more PAX peptides were found, namely a PAX peptide (m/z 1070.8091) which correlated to a linear analogue of PAX1' (denoted PAX1'L) and a novel PAX peptide from the A group at m/z 1078.8119, that correlated to a PAX7 with an enoyl group in the lipid chain, denoted PAX7E1. This putative identification was done from the accurate M r , fragmentation pattern of the peptide moiety (refer to Additional file 1) and the fact that it eluted just before PAX7. This elution pattern correlated very well with that of PAX1' and its 7-enoyl analogue PAX5 (see discussion below and Table 2). As there is already this identified 7-enoyl analogue in the PAX group, it is possible PAX7E1 is also a 7-enoyl analogue. We were, however, not able to confirm the structure of the R group with our in-analysis MS e methodology, as the CID energy only released the lipid moiety and limited fragmentation was achieved. Fraction D contained PAX3' (m/z 1066.8097) and fraction F contained PAX7 (m/z 1080.8286). This fraction also contained UV absorbing compounds (refer to Fig. 2) that are possibly the result of xenocoumacin breakdown (detected compounds with m/z 250.2162, 268.2263, 270.2425, 286.2369). The high PAX7 concentration in fraction F led to UPLC peak broadening, possibly due to the aggregation of this lipopeptide. Although PAX7 was the major compound in fraction F, this fraction also contained other PAX-like peptides. Two low abundance peptides with different elution times, but with the same m/z as PAX7E1 with a putative double bond (enoyl) in lipid chain, were observed at 3.71 and 3.88 min (m/z 1078.8096 and m/z 1078.8096, denoted PAX7E2 and PAX7E3). PAX7E2 displayed similar peptide chain fragments to that of PAX7 and PAX7E1, suggesting that the difference in elution time may or due to the position of the putative double bound in the lipid chain (refer to Additional file 1). PAX7E3 co-eluted with various compounds, so the fragmentation pattern was inconclusive although many similar Lys derived fragments were observed. Alternatively, it could be that PAX7E2 and PAX7E3 elutes later than PAX7E1 due to aggregation with PAX7 and other compounds, rather than structural differences. A peptide with m/z 1054.8090 co-eluted with the PAX7 peptide. From the accurate M r it was derived that this peptide could be a linear PAX1' without a hydroxyl-group (denoted PAX1'L-DH), but the structure remains unconfirmed due to the co-elution. Fraction G contained a number of PAX peptides, namely PAX5 (m/ z 1050.7723), PAX1' (m/z 1052.7936), PAX7 (m/z 1080.8280) and PAX8 (m/z 1106.8412). The fact that the same PAX peptides eluted in more than one fraction is Early elution of broad peak, fronting and tailing due to aggregation at high concentration *Putative identification as PAX peptides from peptide moiety fragments and accurate mass determination, E denotes an enoyl group, L denotes linear, DH denotes dehydroxylated, structure of R-group was not elucidated Refer to Additional file 1 for UPLC-MS, ESMS and MS/MS data on all the compounds possibly due to the formation of hetero-oligomers by the different lipopeptides, leading to elution at different acetonitrile concentrations during reverse phase chromatography. As in fraction D, fraction G also contained some UV absorbing compounds (refer to Fig. 2) that are possibly the result of xenocoumacin breakdown. Examples of the UPLC-MS chromatograms and spectra of the five most abundant PAX peptides are presented in Fig. 7. The primary structures of the known PAX peptides that were found in this study are given in Fig. 8. We were able to confirm the peptide sequence of most of the identified PAX peptides with our UPLC-MS e procedure, except those that were found in very low concentrations. Similar fragment patterns to that depicted in Fig. 5 for PAX1' were observed for the PAX peptides discussed above (refer to Additional file 1). This discovery and identification of the PAX lipopeptides were supported by the identification of the four modules coding for the PAX synthetase complex in the genome of X. khoisanae by our tblastx study (refer to Table 1).

Conclusions
It was not surprising to discover the production of various antimicrobial compounds by X. khoisanae due to the Xenorhabdus-Steinernema-insect host tripartite interaction [1]. The known PAX peptides have only been isolated from X. nematophila, however, this is the first report of PAX peptides produced by X. khoisanae. PAX lipopeptides were first characterised by Gaulteri et al. [16] and to date 13 unique PAX peptides have been reported [34]. Our UPLC-MS and MS e analyses and identification of PAX peptides and compounds in the antimicrobial complex of X. khoisanae SB10 have not been exhaustive, because of the complexity of many fractions containing labile compounds or unresolved compounds. There are certainly more antimicrobial compounds to discover in this strain's natural antimicrobial complex in future studies, such as other xenocoumacins, xenorhabdins and compounds from the aryl polyene group, as indicated by the detection of the APE Ec biosynthetic complex (refer to Fig. 1). Because we consistently found the PAX lipopeptides and xenocoumacin in all the X. khoisanae SB10 culture extracts, we focused in this study on these compounds. Possible metabolites and degradation products related to the xenocoumacins were observed in some fractions (D and G), but after in-depth analysis of all the chromatographic fractions using UPLC-MS only one intact xenocoumacin, namely xenocoumacin 2, was identified in fraction B. It is possible that other xenocoumacins were also produced, but that the more labile compounds were lost during purification in a highly acidic environment. We were, however, able to identify five known Lys-rich lipopeptides from the PAX A-group with PAX1' and PAX7 being the most abundant. We also discovered two putative linear PAX analogues, which could possibly be metabolic precursors of PAX1'. PAX7 could be related to the three unknown PAX peptides with the same M r of 1077.80 but different retention on a reverse phase matrix. From their M r , fragmentation patterns and UPLC elution behaviour we putatively classified the three peptides as enoyl-derivatives of PAX7. We were, however, unable to confirm the R-group because of low abundance and co-elution, as well as limited fragmentation of the R-group in MS e mode. This study is the first to identify both the PAX peptides and a xenocoumacin in the antimicrobial complex of a Xenorhabdus species, as well as the first study on the antimicrobial complex of the Southern African X. khoisanae SB10. This report also highlights the natural tendency of Xenorhabdus species to produce antimicrobial complexes consisting of small antibiotics and AMPs. With the rising antibiotic resistance it may be wise to consider combining AMPs and small antibiotics, mimicking Xenorhabdus-type antimicrobial complexes.

Bacterial strains, growth media and growth conditions
Xenorhabdus khoisanae SB10 was maintained on NBTA [41], consisting of nutrient agar supplemented with bromothymol blue (0.025%, w/v) and TTC (0.004%, w/v). Incubation was at 30°C. Bacillus subtilis subsp. subtilis BD170, Escherichia coli Xen 14 and Candida albicans CAB 392 were used as targets in the testing for antimicrobial activity. The bacteria and yeast were incubated at 37°C. Bacteria were cultured on Brain Heart Infusion agar (Biolab Diagnostics, Gauteng, South Africa) and C. albicans on Potato Dextrose Agar (PDA, Biolab Diagnostics).

Isolation of antimicrobial compounds
XAD-16 beads were activated by treating with 80% isopropanol containing 0.1% (v/v) TFA and added to TSB. After 30 min at 4°C on an orbital shaker (100 rpm), the XAD-16 beads were removed and the medium autoclaved. X. khoisanae SB10 was inoculated into 5 ml untreated TSB and incubated at 26°C for 24 h on a rotating wheel. The culture was added to 5 g activated XAD-16 beads, spread-plated onto XAD-16-treated TSB agar in petri dishes with a diameter of 135 mm and incubated at 26°C for 96 h. Beads were collected from the plates and washed with sterile deionised water to remove the cells. Water was removed from beads by vacuum suction. The beads were washed with 150 mL 30% (v/v) ethanol for 15 min at 4°C on an orbital shaker (100 rpm) . Ethanol was removed by vacuum suction and the beads were washed with sterile deionised water. Amphipathic compounds were liberated from the beads, using 70% (v/v) isopropanol containing 0.1% (v/v) TFA (isopropanol-TFA). The eluent was filtered through a 0.45 μM cellulose nitrate filter and the isopropanol removed by using a rotary evaporator (RotaVapor® R-114, Büchi).

Purification of antimicrobial compounds
The concentrated eluent was subjected to reverse phase chromatography on a 10 ml Sep-Pak C18 column (Waters, Milford, USA) on Perista Pump SJ-1211 (Chromatograph ATTO corporations, Tokyo, Japan). The column was washed with deionised water and compounds eluted by using a stepwise gradient ranging from 10 to 70% (v/v) isopropanol in analytical quality water with constant 0.1% TFA (v/v) in solvent system. The gradient was created with 10% increments per 10 min at 2 mL/minute flowrate. The fractions (denoted SPC fractions) were dried by rotary evaporation and tested for antimicrobial activity using the agar-well diffusion assay as described elsewhere.
Active SPC fractions were loaded onto a HiScale column (100 × 16 mm) packed with 15 RPC resin (GE Healthcare, South Africa) fitted to fast protein liquid chromatography (FPLC, ÄKTA purifier, GE Healthcare, South Africa). Fractions were eluted by a linear gradient of 10 to 55% B over 30 min, at a flow rate of 2.5 ml/min (A: analytical quality water containing 0.1%, v/v, TFA; B: HPLC grade acetonitrile containing 0.1%, v/v, TFA). Readings were recorded at 254 nm. Fractions were tested for antibacterial activity against B. subtilis subsp. subtilis BD170, using the agar-well diffusion assay as described elsewhere.
Fractions with antimicrobial activity collected from the ÄKTA purifier (FPLC active fractions) were lyophilized, resuspended in 50% (v/v) acetonitrile, loaded onto a Discovery BIO Wide Pore C 18 HPLC column (10 μm, 250 × 10 mm; Sigma-Aldrich) and eluted by using a linear gradient (25 to 45%) of eluent B over 28 min at a flow rate of 2.0 ml/min. Further separation was on a Surveyor plus HPLC (Thermo Fisher Scientific, Waltham, Massachusetts, USA). Readings were recorded at 254 nm. Peak fractions were collected, dried and the antimicrobial activity tested as described elsewhere.  Analysis of fractions with ultra-performance liquid chromatography and electrospray ionization mass spectrometry Fractions with antimicrobial activity collected from the HPLC were subjected to UPLC, using an Aquity UPLC™ linked to a Waters Synapt G2 Mass Spectrometer (Waters Corporation, Miliford, USA). This was denoted UPLC-MS. Samples were prepared in 50% acetonitrile in water (1:1, v/v) at a concentration 200-1000 μg/mL. Samples were injected at 1-5 μl via Waters Acquity UPLC™ and chromatography was monitored in positive ESI mode and via PDA (230-400 nm). Separation of the components in each HPLC fraction were done on an Acquity UPLC® HSS T3 C 18 column (1.8 μm particle size, 2.1 × 150 mm, Waters Corporation, Dublin, Ireland) . Chromatography was done with analytical quality water modified with 0.1% (v/v) formic acid as eluent A and acetonitrile modified with 0.1% (v/v) formic acid as eluent B. The gradient developed at flow rate of 300 μl/ min was as follows: 0 to 0.5 min at 40% B, linear gradient from 40 to 95% B from 0.5 to 11 min and 11 to 14 min at 95% B. The rest of the instrument settings for the UPLC-MS mode were as follows: cone voltage set at 15 V, a capillary voltage of 2.5 kV, cone voltage of 15 V, extraction cone voltage 4 V, source temperature of 120°C, desolvation gas of 650 l/h and desolvation temperature of 275°C. Data were collected in positive mode by scanning through m/z = 100 to 2000 in centroid mode at a rate of 0.2 scans/sec.
High resolution collisionally induced dissociation (CID) analyses were done in the MS e mode (tandem MS or MS/MS) during the UPLC-MS and monitored on a second MS channel. CID were done at a collision energy gradient of 20 to 60 eV at 1 s MS/MS scan time. Data were collected in the second mass analyser (MS2) through m/z = 40 to 1500 in centroid mode. The rest of the instrument settings were as described above. To ensure reliable high-resolution MS data, the MS instruments Antimicrobial activity of fractions from purification SPC active fractions were suspended in analytical quality water, containing 0.1% (v/v) TFA to 350 mg/ml. The antimicrobial activity of fractions was tested using an agar-well diffusion assay in a micro well titre plate. In short, the appropriate growth media containing 1.0% (w/ v) agar was seeded with a dense 12-h-old culture of B. subtilis subsp. subtilis BD170, E. coli Xen 14 (1.0%, v/v) or C. albicans CAB 392 (1.0%, v/v). Wells were made into the agar and 15 μL of each fraction dispensed into a well. Plates were incubated for 24 to 48 h at 37°C. A clear zone surrounding the well indicated activity. Analytical quality water, containing 0.1% (v/v) TFA, was used as negative control. Ciprofloxacin was used as positive control for B. subtilis subsp. subtilis BD170 and E. coli Xen 14 and amphotericin B for C. albicans CAB 392.

Temperature stability
SPC active fractions of 350 mg/mL were prepared in MilliQ water, containing 0.1% (v/v) TFA. The suspension was autoclaved for 20 min and tested for antimicrobial activity against B. subtilis subsp. subtilis BD170, using the agar-well diffusion assay as described elsewhere. Plates were incubated at 37°C for 24 h. The diameter of growth inhibition zones was recorded and compared to controls. This was done by using the software program ImageJ (v. 1.48).