Efflux pump-deficient mutants as a platform to search for microbes that produce antibiotics

Pseudomonas putida DOT-T1E-18 is a strain deficient in the major antibiotic efflux pump (TtgABC) that exhibits an overall increased susceptibility to a wide range of drugs when compared with the wild-type strain. We used this strain as a platform to search for microbes able to produce antibiotics that inhibit growth. A collection of 2400 isolates from soil, sediments and water was generated and a drop assay developed to identify, via growth inhibition halos, strains that prevent the growth of DOT-T1E-18 on solid Luria–Bertani plates. In this study, 35 different isolates that produced known and unknown antibiotics were identified. The most potent inhibitor of DOT-T1E-18 growth was an isolate named 250J that, through multi-locus sequence analysis, was identified as a Pseudomonas sp. strain. Culture supernatants of 250J contain four different xantholysins that prevent growth of Gram-positive bacteria, Gram-negative and fungi. Two of the xantholysins were produced in higher concentrations and purified. Xantholysin A was effective against Bacillus, Lysinibacillus and Rhodococcus strains, and the effect against these microbes was enhanced when used in combination with other antibiotics such as ampicillin, gentamicin and kanamycin. Xantholysin C was also efficient against Gram-positive bacteria and showed an interesting antimicrobial effect against Pseudomonas strains, and a synergistic inhibitory effect with ampicillin, chloramphenicol and gentamicin.


Introduction
The increasing antibiotic resistance level of microorganisms and the emergence of multi-drug-resistant bacteria due to the accumulation of several resistance genes have evoked a critical situation for the treatment of immunosuppressed patients and elderly people who are infected by pathogenic bacteria (Teran et al., 2006;Donadio et al., 2009). Different modes of antibiotic resistance have been described in bacteria, for instance; acquired resistance, linked to mutations in chromosomal genes or the acquisition of new genes by horizontal transfer (D'Costa et al., 2011;Bernal et al., 2013); adaptive resistance, which involves a temporary increase in the ability of bacterial cells to survive the action of an antibiotic, mainly as the result of alterations in gene and/or protein expression triggered by environmental conditions (Poole, 2012); and intrinsic resistance, which is not related to antibiotic selection but to the specific characteristics of the bacteria, due to the presence of outer membrane lipopolysaccharides that prevent antibiotic entry and the presence of efflux pumps that extrude antibiotics out of the cells Bernal et al., 2013;Li et al., 2015).
These variety of resistance mechanisms and the fast evolution of multi-drug pathogen bacteria have made that one of the current challenges for antibiotic therapy is the identification and development of novel and more effective antibiotics to combat them, particularly those resistant to last-line antibiotic agents (Gwynn et al., 2010). The main classes of antibiotic drugs commonly used today were discovered in the second half of last century, and their targets are proteins involved in cell-wall biosynthesis (beta-lactams, glycopeptides), cell membrane structures (daptomycin, colistin), type II topoisomerases (fluoroquinolones), ribosome functions (macrolides, aminoglycosides, tetracyclines), transcriptional machinery (rifamycins) and folate biosynthesis (sulfonamides and trimethoprim) (Fernandes, 2006;Lange et al., 2007). Therefore, new antimicrobial agents with a greater potential to address the deficiencies of existing classes of antibiotics are needed to combat bacterial resistance (Gwynn et al., 2010;Laverty et al., 2010). In addition to newly discovered antimicrobials, other strategy could be the use of two or more antibiotics alone or in combination therapies to prevent or delay the emergence of resistant strains and to take advantage of antibiotic synergism as a potential approach to treat infectious diseases (Zhao et al., 2002;Mitosch and Bollenbach, 2014). Among the strategies to discover new antimicrobial compounds, chemical modification of existing drugs either with the aim to increase their spectrum of activity or to enhance their activity, and target-directed screening methods to identify new natural compounds are often used (Moellering, 2011). Studies performed by other research groups (Hsieh et al., 1998) pointed out the possibility to use hypersensitive strains in the discovery of new antimicrobial compounds.
Pseudomonas putida strain DOT-T1E is able to thrive in the presence of high concentrations of organic solvents, such as toluene, (Ramos et al., 1995) and toxic compounds such as dyes, heavy metals and a broad array of antimicrobial compounds of different families (Ramos et al., 1998;Teran et al., 2003;Fernandez et al., 2012). The extrusion of antimicrobial compounds is mainly achieved through the action of resistance nodulation cell division (RND) efflux pumps, being TtgABC the most important one in antibiotics extrusion (Mosqueda and Ramos, 2000;Rojas et al., 2001;Fernandez et al., 2012;Molina-Santiago et al., 2014). The lack of efflux pump has been shown to lead to enhanced antibiotic susceptibility as in other microbes such as Escherichia coli, i.e. mutants deficient in AcrAB (Okusu et al., 1996) or Pseudomonas aeruginosa, i.e. mutants deficient in MexAB (Li et al., 1998;Poole and Srikumar, 2001).
This study describes a non-directed strategy for the discovery of antimicrobial compounds using a P. putida DOT-T1E strain deficient in the TtgABC efflux pump as reporter for the identification of antimicrobial producer bacteria. This methodology allowed us to identify highly active antimicrobials, including compounds of the xantholysin family produced by Pseudomonas sp. strain 250J which showed antimicrobial activity against Grampositive bacteria but also against Gram-negative, which is rarely described for lipodepsipeptides produced by Gramnegative strains.

Results and discussion
Method for the identification of antimicrobial-producing bacteria using P. putida DOT-T1E-18 as a reporter We reasoned that P. putida DOT-T1E efflux pumpdeficient strains would be more sensitive to antibiotics than the wild-type strain and that this hypothesis, if confirmed, would add a relevant strain for screening for new drugs due to the availability of a sensitive strain. To this end, we carried out minimum inhibitory concentration (MIC) assays and inhibition zone experiments for nine different antibiotics using the wild-type DOT-T1E strain and, ∇TtgABC (DOT-T1E-18) and ∇TtgGHI (DOT-T1E-PS28) mutant strains. We found that for the complete set of antibiotics used, the minimal concentration necessary to inhibit DOT-T1E-18 growth was significantly lower than for the wild-type strain and the DOT-T1E-PS28 strain (Table 1). Similar results were found when inhibition zone assays were performed, i.e. the halos in response to a given concentration of antibiotics of different families were largest in the case of the DOT-T1E-18 mutant strain (Fig. S1). Therefore, the strain DOT-T1E-18 inactivated in TtgABC efflux pump showed higher sensitivity to different antibiotic classes. We therefore reasoned that the use of this hypersensitive strain could facilitate the detection of compounds produced by the isolates present in our collection of 2400 environmental bacteria (described in Materials and methods).

Identification of environmental isolates that prevent growth of P. putida DOT-T1E-18
We performed inhibition zone assays to determine indicative antimicrobial activity of the environmental isolates against DOT-T1E-18. In these assays, we spread the indicator strain on Luria-Bertani (LB) plates, and once dried, we dropped 3 μl of each isolated bacteria on the dried surface of the LB plates. After 24 h incubation at 30°C, plates were inspected for the appearance of inhibition halos. We found that 150 bacteria from the collection produced an inhibition halo, and they were considered positive candidates for antimicrobial production (Fig. S2).
To further test the production of antimicrobials, culture supernatants were harvested and tested to confirm growth inhibition in 96 well-plates. For these assays, overnight cultures were centrifuged and filtered, and 65 μl of the filtrate was placed in wells with 25 μl of LB and 10 μl of diluted DOT-T1E-18 reporter strain. After 24 h, growth was measured to determine the antimicrobial effect of the supernatants. We found that 86 of the 150 culture supernatants tested inhibited growth in this test.
To learn about the potential phylogenetic adscription of the above 86 isolates, fragments of the 16S rRNA gene were polymerase chain reaction (PCR)-amplified and sequenced. BLAST analysis revealed that some of the isolates could be siblings since they shared 100% of the sequences. This screening reduced the number of potential different microbes to 35 different strains. Of these strains, Bacillus and Pseudomonas species were the most represented with 12 and eight different members respectively; bacteria of the genus Enterobacter, Raoultella, Alishewanella, Rhodococcus, Aeromonas, Vibrio, Shewanella, Acinetobacter, Alcaligenes, Cupravidus and Lysinibacillus were also identified (Table S1). In addition, we also found eight strains whose 16S rRNAs matched sequences of strains deposited in the databases as 'uncultured' microbes.

Antimicrobial compound detection
Extracts of fermentation broths of the 35 selected strains were analysed by low-resolution mass spectrometry (LR-MS) to detect potential new antimicrobial compounds. The strains were grouped according to their LR-MS profiles, and we found that seven strains were able to produce already described antimicrobials, like Alishewanella sp. strain 255W, which produced compounds of the valin-surfactins and plipastatin families ( Fig. S3 and Table S2). This information could be used as a proof-ofconcept of the screening method. Furthermore, we also identified three strains producing unknown antimicrobial compounds or compounds described in the literature as rarely produced by bacteria. For instance, Lysinibacillus fusiformis strain 249MT produced erythrolic acid D and molecules with a known molecular formula which have been described to be produced by fungi, plants and even animals, but not bacteria ( Fig. S4 and Table S3). We found that Vibrio sp. strain 225TR produced serratomolide C and 4-deoxy thiomarinol H ( Fig. S5 and Table S4). Among the isolated strains which produced compounds no identified to the date of the start of our project (December 2012), we found that the largest inhibition halo was produced by the supernatant of Pseudomonas sp. strain 250J. In this supernatant, a family of four unknown compounds were detected (Molina-Santiago et al., 2015; Table S5). We then decided to study and to analyse the compounds produced by Pseudomonas sp. 250J. Highresolution mass-spectrometry (HR-MS) analyses led us to determine that the mass of the compounds were of 1775.08 Da (A), 1761.07 Da (B), 1802.0 Da (C) and 1775.09 Da (D), being compounds A and C the two majoritarian ones (Fig. S6). Therefore, we decided to isolate and to elucidate the conformational structure of these compounds using Nuclear Magnetic Resonance (NMR) and tandem mass spectrometry analyses (Figs S7 and S8). During the elaboration of our analysis, the same compounds were described by Li et al., (2013) and named as xantholysins. We not only confirmed the mass of the compounds but also elucidated the conformational structure of the amino acids of xantholysin A (Appendix S1; Fig. 1; Figs S9-S14) using Marfey methodology and high performance liquid chromatography-UV-mass spectrometry (HPLC-UV-MS).

Antimicrobial activity of xantholysins and synergistic effects with other antibiotics
To assess the antimicrobial activity of the supernatant of Pseudomonas sp. strain 250J containing xantholysin, we evaluated its potential against Gram-negative and Gram-positive bacteria, such as different Pseudomonas species, Bacillus subtilis, Rhodococcus erythrophyla, E. coli K12 and Lysinibacillus fusiformis. For this, culture supernatant of 250J strain was used in growth inhibition assays (Table 2).
Purified xantholysin A inhibited growth of Gram-positive bacteria Bacillus sp., L. fusiformis and R. erythrophyla (MIC of 15-50 μg ml −1 ). Xantholysin A also inhibited growth of Gram-negative strains but higher concentrations were needed, i.e. Pseudomonas strains (MIC of 250-500 μg ml −1 ), E. coli and P. aeruginosa (> 1000 μg/ml) ( Table 3). In contrast, when we used purified xantholysin C, we observed growth inhibition in all Pseudomonas tested (Table 2) in ranges between 62.5 μg/ml and 125 μg/ml, a fourfold stronger effect than xantholysin A. Furthermore, different to the observed with xantholysin A, xantholysin C is able to inhibit growth of P. aeruginosa PAO1, a pathogen strain. However, it is interesting to remark that, although the detection and isolation of a new xantholysin producer was performed using DOT-T1E-18, xantholysin compounds showed similar antimicrobial activity against wildtype and mutant strains. The different effect of molecules of the same family could be indicating that 250J presents an armory of compounds to combat against other microorganisms present in the same ecological niche, or its capacity to colonize other environmental places.
To analyse a hypothetical role of xantholysins as adjuvant of antibiotics against different strains, checkerboard assays were performed to study synergism of xantholysins A and C with classical antimicrobial compounds like ampicillin, gentamicin, chloramphenicol and kanamycin (Table 3). According to the results shown in Table 3, xantholysin A presented a synergistic effect (FIC ≤ 0.5) when it was used along with ampicillin  Table 3, additive effects were found with xantholysin A in combination with some antibiotics, while in other cases, synergistic effects were not found. In the case of xantholysin C, we only observed synergistic effects when it was supplied with chloramphenicol against DOT-T1E, with ampicillin and gentamicin against Bacillus sp., and with gentamicin against E. coli K12 (Table 3).

Concluding remarks
The emergence of multi-resistant bacteria and the lack of new antibiotics to combat them has become a serious risk for humanity. One class of compounds that, during the recent decades, have attracted an increasing amount of attention due to their promising role as therapeutics or drug leads are short antimicrobial peptides that usually exhibit rapid and efficient antimicrobial toxicity against a wide range of pathogens, mainly Gram-positive strains (Ganz and Lehrer, 1999;Hadley and Hancock, 2010;Hughes and Fenical, 2010;Laverty et al., 2010). In this study, we provide a platform for the detection of antimicrobial compounds based on a P. putida DOT-T1E strain deficient in the TtgABC efflux pump, the main element required for antibiotic extrusion. The mutant strain is more sensitive to antibiotics, making it a more responsive method for the detection of antimicrobials.
Out of the 2400 isolates previously obtained from different environmental niches, we identified 35 antimicrobial producing strains that belonged to different genera; with a predominance of Bacillus and Pseudomonas strains. High-performance liquid chromatography-mass spectrometry analyses of the culture supernatants from the strains led us to identify different known and unknown antimicrobial compounds, confirming that the platform fulfils the proof-of-concept for detection of microbes that produce antibiotics.
Pseudomonas sp. strain 250J was the most promising one, due to its ability to produce a family of, in the discovery moment, four unknown compounds further described as xantholysins. Purified xantholysin A and C were used to elucidate their structures and their L-and D-amino acid configurations. Analysis of their antimicrobial activities showed that both compounds had similar antimicrobial patterns against Gram-positive bacteria. However, xantholysin C also exhibited an interesting activity against Gram-negative bacteria, including P. aeruginosa. Further analysis also showed a synergistic effect of both compounds in combination with classical antibiotics against Gram-positive and Gram-negative strains. Taken together, these results support an interesting method for the detection of antimicrobial compounds with higher sensitivity. Further studies are needed to determine the potential use of these compounds in clinical assays.

Strains and general growth conditions
The bacterial strains used in this study are shown in Table 4. Bacterial cells were grown in liquid LB medium at 30°C in an orbital shaker at 200 r.p.m. (Kühner). When necessary, the appropriate antibiotics were added to reach the following final concentrations: 50 μg ml −1 kanamycin and 20 μg ml −1 rifampicin. The concentration of other antibiotics used in this study is indicated in the text. For production of xantholysins, KIDO medium (glycerol 2% [v/v]; sodium L-glutamate monohydrate 1% [w/v]; yeast extract 0.25% [w/v] adjusted to pH 7) was used and cultures were grown at 18°C for 3 days (Pascual et al., 2014). . The estuaries of the Odiel River (Huelva, Spain) are characterized by tidal pulls and a high salt concentration due to their proximity to the Atlantic Ocean. Three hundred millilitres water samples were taken from the river channel and kept at 4°C until used. Sediments were sampled when the tidal influence in the river was maximal and minimal. In each case, the samples were prepared by collecting 100 g sediments from each location at sample points that were least 10 m apart.
Inoculated media were incubated at 30°C for 24 h, 48 h, 72 h, 1 week and 2 weeks with agitation (200 rpm). At the indicated times aliquots were collected and plated in the same isolation medium that were used for enrichment, but were supplemented with agar (2% w/v). Different colonies were selected according to their apparent phenotype, such as colour, roughness, size and morphology in order to keep as many different bacteria as possible and enabled creation of a collection of around 2400 microorganisms.

Antimicrobial producer identification
Pseudomonas putida DOT-T1E-18 (TtgABC mutant) reporter strain was grown aerobically at 30°C. Antimicrobial activity of the isolated bacteria against the reporter strain was assessed with a colony overlay assay. The indicator bacteria was spread on LB plates and allowed to dry; subsequently, isolated bacterial colonies from an overnight pre-culture were loaded. Plates were incubated for 24 h, and inhibition zones around the spots were scored as positive. In the case of a positive result, liquid antimicrobial activity screening assays were carried out in 96-well plates. These assays were carried out in microdilution plates in which 65 μl of filtrated supernatant was mixed with 25 μl LB and 10 μl of the reporter bacterium that allowed to reach an OD 660nm of 0.05. These plates were incubated at 30°C with agitation (200 rpm) for 24 h and growth was assayed. Inhibition was scored as positive if the turbidity was 10-fold lower than in the control without filtrate supernatant. Experiments were repeated in triplicate.

Checkerboard analysis
Standard powder forms of ampicillin, chloramphenicol, gentamicin and kanamycin were stored at 2°C to 8°C until use. Immediately prior to testing, the stock solutions and serial twofold dilutions of each drug were prepared at concentrations that were at least double the MIC, according to National Committee for Clinical Laboratory Standards (NCCLS) recommendations. A total of 100 μl of LB broth inoculated with the bacterial inoculum at a 0.5 McFarland turbidity was distributed into each well of the microdilution plates. The first drug of the combination was serially diluted along the ordinate, while the second drug was diluted along the abscissa. Plates were incubated at 30°C or 37°C for 24 h at 200 r.p.m.
In accordance with the NCCLS guidelines for broth microdilution, the MIC was defined as the lowest concentration of antibiotic that completely inhibited the growth of the organism as detected by reading turbidity at 600 nm. Synergy is more likely to be expressed according to ΣFICs that were calculated as follows: ΣFIC = FIC A + FIC B, where FIC A is the MIC of drug A in the combination/MIC of drug A alone, and FIC B is the MIC of drug B in the combination/MIC of drug B alone. The combination is considered synergistic when the ΣFIC is ≤ 0.5, additive when the ΣFIC is > 0.5 to ≤ 1, indifferent (no interaction) when ΣFIC is > 1 to ≤ 4 and antagonistic when the ΣFIC is ≥ 4 (Sopirala et al., 2010).

MIC
Minimum inhibitory concentration assays of DOT-T1E, T1E-18 and T1E-PS-28 were performed in liquid LB medium using the twofold serial dilution test according to the Clinical and Laboratory Standards Institute guidelines (2003). The highest concentration of the antibiotics used were: tetracycline (10 000 μg ml −1 ); chloramphenicol (3000 μg ml −1 ); norfloxacin (200 μg ml −1 ); erythromycin (3000 μg ml −1 ); gentamicin (1000 μg ml −1 ), kanamycin (2500 μg ml −1 ) and ampicillin (10 000 μg ml −1 ). In the cases of xantholysin A and C, a concentration of 2000 μg ml −1 was used. At least three independent experiments were carried out for each determination, and each experiment was run in triplicate. The minimum inhibitory concentration (MIC) was determined as the lowest concentration of antibiotic that inhibited the growth of the strain by > 90%.

Extraction and purification of xantholysin A and xantholysin C
A culture of Pseudomonas sp. strain 250J (1 L) grown in 2 L flasks for 3 days at 18°C was extracted by acidifying the fermentation broth with HCl to a pH of 2-3. Subsequently, methyl ethyl ketone (EtCOMe) was added and three 10 min centrifugation steps were carried out. After each centrifugation, the organic phase was transferred to glass tubes. The sample was evaporated on an orbital shaker to concentrate the sample and re-suspended in dimethylsulphoxide. The solution was loaded onto a semipreparative HPLC column [Zorbax SB-C8, 21.2_250 mm, gradient H 2O 0.1% (v/v) CF3COOH (TFA)/Acetonitrile 0.1% (v/v) TFA from 5% to 100% (v/v) of Acetonitrile/TFA in 45 min, 3.6 ml min −1 , UV detection). The peak eluting at 31 min was re-purified on the same column with a 68% isocratic Acetonitrile/TFA yielding 16 mg of xantholysin A and 5 mg of xantholysin C.
Retention times (min) of the observed peaks in the HPLC trace of the FDVA-derived hydrolysis products of xantholysin A were as follows: D-Ser (9.7), L-Glu (9.4), D-Glu (11.9), D-Val (25.1), L-Leu (20.8), D-Leu (29.1), L-Ile (20.8) and D-Ile (29.1). Due to the close retention times observed for L-Ile, L-Leu, D-Ile and D-Leu, the presence of L-Ile or D-Ile was confirmed using a second HPLC method with isocratic elution (33% B during 60 min). Under these conditions, retention times of 19.1 and 19.9 min were obtained for L-Ile and L-Leu respectively.  Fig. S3. High-performance liquid chromatography-mass spectrometry profile of 255W culture extracts. Fig. S4. High-performance liquid chromatography-mass spectrometry profile of 249MT culture extracts. Fig. S5. High-performance liquid chromatography-mass spectrometry profile of 225TR culture extracts. Fig. S6. High-performance liquid chromatography-mass spectrometry profile of 250J culture extracts. Fig. S7. Nuclear Magnetic Resonance (NMR) of xantholysin A in MeOD. Fig. S8. Tandem mass spectrometry analysis of xantholysin C. The analyses were performed to confirm the difference in the lipid tail between xantholysin A and xantholysin C, proposed by (Bernal et al., 2013).  Table S1. Isolation of species capable of inhibiting DOT-T1E-18 growth. Table S2. Antimicrobial compounds found in the 255W culture extracts. Positions correspond with the numbers that appear in Fig. S2. Table S3. Antimicrobial compounds found in the 249MT extract. Positions correspond with the numbers that appear in Fig. S3. NPB, not produced by bacteria. Table S4. Antimicrobial compounds found in the 225TR extract. Positions correspond with the numbers that appear in Fig. S4. DNP, not described in database of natural products. Table S5. Antimicrobial compounds found in the 250J extract. Positions correspond with the numbers that appear in Fig. 1. Appendix S1. Marfey analysis of xantholysin A.