Flexible exportation mechanisms of arthrofactin in Pseudomonas sp. MIS38


Masaaki Morikawa, Division of Biosphere Science, Graduate School of Environmental Science, Hokkaido University, Sapporo 060-0810, Japan. E-mail: morikawa@ees.hokudai.ac.jp


Aims:  To obtain further insights into transportation mechanisms of a most effective biosurfactant, arthrofactin in Pseudomonas sp. MIS38.

Methods and Results:  A cluster genes arfA/B/C encodes an arthrofactin synthetase complex (ArfA/B/C). Downstream of the arfA/B/C lie genes encoding a putative periplasmic protein (ArfD, 362 aa) and a putative ATP-binding cassette transporter (ArfE, 651 aa), namely arfD and arfE, respectively. The arfA/B/C, arfD, and arfE form an operon suggesting their functional connection. Gene knockout mutants ArfD:Km, ArfE:Km, ArfD:Tc/ArfE:Km, and gene overexpression strains MIS38(pME6032_arfD/E) and ArfE:Km(pME6032_arfD/E) were prepared and analysed for arthrofactin production profiles. It was found that the production levels of arthrofactin were temporally reduced in the mutants or increased in the gene overexpression strains, but they eventually became similar level to that of MIS38. Addition of ABC transporter inhibitors, glibenclamide and sodium ortho-vanadate dramatically reduced the production levels of arthrofactin. This excludes a possibility that arthrofactin is exported by diffusion with the aid of its own high surfactant activity.

Conclusions:  ArfD/E is not an exclusive but a primary exporter of arthrofactin during early growth stage. Reduction in the arthrofactin productivity of arfD and arfE knockout mutants was eventually rescued by another ABC transporter system. Effects of arfD and arfE overexpression were evident only for 1-day cultivation. Multiple ATP dependent active transporter systems are responsible for the production of arthrofactin.

Significance and Impact of the Study: Pseudomonas bacteria are characterized to be endued with multiple exporter and efflux systems for secondary metabolites including antibiotics, plant toxins, and biosurfactants. The present work demonstrates exceptionally flexible and highly controlled transportation mechanisms of a most effective lipopeptide biosurfactant, arthrofactin in Pseudomonas sp. MIS38. Because lipopeptide biosurfactants are known to enhance efficacy of bioactive compounds and arfA/B/C/D/E orthologous genes are also found in plant pathogenic P. fluorescens and P. syringae strains, the knowledge would also contribute to develop a technology controlling plant diseases.


Exportation of many nonribosomal peptides and polyketides requires ABC (ATP-binding cassette) transporter system that couples the transport with ATP hydrolysis. The ABC transporter proteins are generally composed of two hydrophobic transmembrane domains (TMD) and hydrophilic nucleotide-binding domains (NBD) bound or fused to the cytosolic face of the TMD (Biemans-Oldehinkel et al. 2006). TMD are hydrophobic parts that create a channel through which the substrate passes during translocation. They are composed of bundles of α-helices that transverse the cytoplasmic membrane several times in a zig-zag fashion. NBD are the engines of ABC transporters that power substrate translocation by ATP hydrolysis and commonly have a set of ATP binding motifs, Walker A and B.

An ABC transporter system usually requires two accessory envelope proteins for its full function. One is a membrane fusion protein (or periplasmic protein) which consists of a short N-terminal hydrophobic domain anchoring it to the inner membrane, a large hydrophilic domain located at the periplasm, and a C-terminal domain with a possible β-sheet structure which could interact with the outer membrane protein (Dinh et al. 1994). Another accessory component is an outer membrane TolC family protein, which transports the efflux substrates to the culture medium (Wandersman and Delepelaire 2004). These whole components are also required for the type I protein secretion system in most of the gram-negative bacteria (Binet et al. 1997).

Pseudomonas sp. MIS38 produces a cyclic lipopeptide, named as arthrofactin, which is one of the most effective biosurfactants (Morikawa et al. 1993). Arthrofactin synthetase genes (arfA/B/C) encode a multimodular nonribosomal peptide synthetase (NRPS; ArfA/B/C) whose unique architecture has been reported recently (Roongsawang et al. 2003, 2005). ArfA, ArfB, and ArfC proteins consist of two, four, and five modules, where each module contains a set of condensation (C), adenylation (A), and thiolation (T) domains. Seven of the eleven modules incorporate D-form amino acids, where in this case, dual condensation/epimerization (C/E) domains are present in the place of authentic C domains (Balibar et al. 2005). At the C-terminal end of the last module of ArfC, two thioesterase (TE) domains are present that are responsible for cyclization and release of the product peptide from the enzyme (Roongsawang et al. 2007). The present work focuses on the function of further downstream two genes, encoding a putative periplasmic protein (ArfD) and a putative ABC transporter protein (ArfE).

There are reports that ABC transporter genes are clustered along with synthetase genes of secondary metabolites to be exported out from the cells (Méndez and Salas 2001). The knowledge prompted us to examine how these two genes, embedded in the arthrofactin synthetase gene cluster, play a role in the exportation of arthrofactin by MIS38 cells. A series of experimental results suggested that ArfD/E is not an exclusive transporter system for arthrofactin and that exportation of arthrofactin is exceptionally flexible enough to regain normal production levels in the transporter gene knockout mutants and the gene overexpression strains.

Materials and methods

Bacterial strains, plasmids and culture conditions

The bacterial strains and plasmids used in this study are listed in Table 1. Pseudomonas sp. MIS38 and its derivatives were grown in Luria-Bertani (LB) broth (per litre: 10 g tryptone, 5 g yeast extract and 5 g NaCl, pH 7·3) at 30°C for arthrofactin production. Escherichia coli (E. coli) DH5α was grown at 37°C while E. coli SM10λpir was grown at 30°C in LB broth. SOC medium was used for cultivation after electroporation (Sambrook and Russell 2001). Antibiotics were used at the following concentrations (mg l−1): kanamycin (Km) 35, chloramphenicol (Cm) 34 and tetracycline (Tc) 40 for Pseudomonas sp. MIS38; ampicillin (Ap) 50, Tc 25, and Km 35 for E. coli.

Table 1.   Strains and plasmids used in this study
Strains and plasmids Genotype/relevant characteristicsReference
 Escherichia coli
 DH5αsupE44 ΔlacU169(ξlacZ ΔM15) hsdR17 recA1 endA1 gyrA96 thi-1 relA1Sambrook and Russell (2001)
 SM10λpirthi thr leu tonA lacY supE recA::RP4-2-Tc::Mu KmDonnenberg and Kaper (1991)
 Pseudomonas sp.
 MIS38 (wild type)Apr Cmr, isolated from oil spillMorikawa et al. (1993)
 ArfD:KmApr Cmr KmrThis study
 ArfE:KmApr Cmr KmrThis study
 ArfD:Tc/ArfE:KmApr Cmr Kmr TcrThis study
 MIS38 (pME6032_arfD/E)Apr Cmr TcrThis study
 ArfE:Km (pME6032_arfD/E)Apr Cmr Kmr TcrThis study
 pGEMTCloning vectorPromega
 pUC18Cloning vectorSambrook and Russell (2001)
 pSMC32Source of kan resistant gene cassetteBartolome et al. (1991)
 pCVD442R6KoriDonnenberg and Kaper (1991)
 pKm/KpnIpGEMT (KpnI site) containing kanThis study
 pKm/SmaIpGEMT (SmaI site) containing kanThis study
 pME6032Shuttle vector between Pseudomonas and E. coli containing lacIq-Ptac fragment for gene expression; source of tetA gene cassette, TcrHeeb et al. (2002)
 pTetA32pUC18 (SmaI) containing 1·4 kb tetA, TcrThis study
 pArfDΔN-β4pGEMT (XbaI site) containing arfDΔN-β4This study
 pArfETvpGEMT (XbaI site) containing arfEThis study
 pArfD:KmpGEMT (XbaI site) containing arfDΔβ4::kan, KmrThis study
 pArfE:KmpGEMT (XbaI site) containing arfE::kan, KmrThis study
 pArfD:Km442pCVD442 (XbaI site) containing arfDΔβ4::kan, KmrThis study
 pArfE:Km442pCVD442 (XbaI site) containing arfE::kan, KmrThis study
 pArfD:TcpGEMT (XbaI site) containing arfDΔβ4::tetA, TcrThis study
 pArfD:Tc442pCVD442 (XbaI site) containing arfDΔβ4::tetA, TcrThis study
 pME6032_arfD/EpME6032 containing arfD and arfEThis study

General molecular biological methods

DNA manipulations were according to standard protocols (Sambrook and Russell 2001) unless described in details. Chromosomal DNA of MIS38 and mutant strains were prepared by Marmur method (Marmur 1961). DNA fragments were recovered from agarose gel by QIAquick Gel Extraction Kit (Qiagen) and plasmids were extracted using QIAprep Spin Miniprep Kit (Qiagen). DNA from phage λ-S9 and λ-S12 were extracted using Lambda Maxi Kit (Qiagen). Cohesive ends of gene fragments were occasionally filled using Takara DNA Blunting Kit (Takara Bio) while ligation was performed using Takara Ligation Kit ver. 2.1 (Takara Bio). DNA sequencing was performed by ABI Prism 3100 Genetic Analyzer with BigDye Terminator ver. 3.1 Cycle Sequencing kit (Applied Biosystems).

Standard PCR was performed for 30 cycles using PTC-100 Programmable Thermal Controller (MJ Research) and Ex-Taq DNA polymerase (Takara Bio) or KOD Plus DNA polymerase (Toyobo). Oligonucleotides for PCR primers were synthesized by Hokkaido System Science. Primers used in this study were listed in Table 2. The nucleotide sequences were analysed by GENETYX software (GENETYX) and Blast programs (http://www.ncbi.nlm.nih.gov). Amino acid sequences were analysed by ClustalW program (http://clustalw.genome.jp).

Table 2.   Primers used in this study
Primer nameSequences

Reverse transcription polymerase chain reaction

Reverse transcription polymerase chain reaction (RT-PCR) experiment was performed as follows. Total RNA was extracted from the culture of MIS38 according to manufacturer’s protocols using RNAeasy Mini Kit (Qiagen). One microgram of DNase-treated RNA was used to synthesize first-strand cDNA in 20 μl volume with random primers and Reverse Transcriptase System (Promega). After completion of reverse transcription, 1 μl of products was used for PCR amplification of DNA fragments using each specific primer and KOD Plus DNA polymerase (Toyobo).

Prediction of transmembrane segments of ArfE

Transmembrane segments (TMS) of ArfE were predicted using the Membrane Protein Explorer (MPex) programme by Jayasinghe et al. (2001) at http://blanco.biomol.uci.edu/mpex. This sliding-window hydrophobicity analysis of amino acid sequences of membrane proteins is useful to identify putative transmembrane parts.

Gene cloning of a putative periplasmic protein, ArfD and insertion of kanamycin resistant gene cassette for gene knockout experiment

Nucleotide sequences of the gene encoding putative periplasmic protein (ArfD) in MIS38 were obtained by Roongsawang et al. (2003) (AB107223). Km resistant gene cassette (kan) was inserted in the structurally important N-terminal β4 region to construct ArfD:Km. Primers MFP/XbaI(f) and 708r(MFP) were used to obtain 700 bp fragment at the N-terminal part of ArfD while primers MFP/XbaI(r) and 736f(MFP) were used to obtain 1·3 kb fragment at the C-terminal part of ArfD. Both 708r(MFP) and 736f(MFP) primers had an additional KpnI site, for ligation purposes between the 700 bp and 1·3 kb fragments. The ligation product yielded arfDΔN-β4. DNA sequencing was performed in order to confirm introduction of the deletion without unexpected mutation. On the other hand, 1·2 kb fragment of kan was obtained by PCR using pSMC32 and primers pSMC32/KpnI(f) and pSMC32/KpnI(r). It was then cloned into KpnI gap of arfDΔN-β4 in pGEMT, creating pArfD:Km. After that pArfD:Km was digested with XbaI and subcloned into suicide vector pCVD442, producing pArfD:Km442 and transferred into E. coli SM10λpir by electroporation. After that, conjugation between MIS38 and SM10λpir(pArfD:Km442) was carried out (Roongsawang et al. 2007). Transconjugants were selected on a plate containing both Km (35 mg l−1) and Cm (34 mg l−1). MIS38 was originally resistant to Cm, and Cm inhibited the growth of donor E. coli cells. In order to verify successful construction of the mutant gene in MIS38 chromosome, chromosomal DNA from mutants were extracted using InstaGene Matrix (Bio-Rad) and used for PCR amplification with a primer set Fl(MFP)ORF5-7521f and Fl(MFP)ABC-493r. A mutant strain ArfD:Km was thus obtained and used for further studies.

Gene cloning of ABC transporter, ArfE and insertion of kan for gene knockout experiment

The first 120 bp gene encoding N-terminal part of a putative ABC transporter gene had been previously obtained (AB107223). Rest of the complete gene, arfD/E was obtained in this study from a genomic library of phage λ-S12 (Roongsawang et al. 2003). In order to construct a gene knockout mutant strain ArfE:Km, kan cassette was inserted in the SmaI site located between linker peptide and Walker B region of ArfE. The 2 kb fragment including arfE was amplified by a set of primers ABC/XbaI-563f and ABC/XbaI-2583r. Then, it was cloned into XbaI site of pGEMT-vector, producing pArfETv. The 1·2 kb kan was prepared by PCR using pSMC32/SmaI(f) and pSMC32/SmaI(r). This fragment was then cloned into pArfETv at a newly introduced SmaI site in arfE, followed by transfer of the resulting 3·2 kb XbaI fragment into pCVD442. The following procedures were according to above described. Selection of positive mutants was confirmed using a primer set, ABC/XbaI-518f and ABC/XbaI-2628r. A mutant strain ArfE:Km was thus obtained and used for further studies.

Construction of a double mutant strain ArfD:Tc/ArfE:Km

The 1·4 kb tetracycline resistant gene cassette (tetA) was obtained by PCR with a primer set TetA-2(f) and TetA-2(r), and pME6032 as a template. This DNA fragment was cloned into pUC18 at SmaI site to construct pTetA32. Then, XbaI fragment containing tetA was blunt ended and inserted into the blunt ended KpnI site in arfDΔN-β4, creating pArfD:Tc. Then, pArfD:Tc442 was obtained by cloning arfDΔN-β4::tetA into XbaI site of pCVD442. After that, conjugation between strain ArfE:Km and E. coli SM10λpir harboring pArfD:Tc442 was carried out. Selection of mutants was performed on the LB agar plate containing both Tc (40 mg l−1) and Cm (34 mg l−1). Then, genomic DNA from the candidate strain was extracted for PCR template, and a set of primers Fl(MFP)ORF57521f and Fl(MFP)ABC493r was used to verify introduction of tetA in arfD. Another primer set, MfpABC1503f and MfpABC5661r was used to detect the presence of both the tetA and kan resistant gene cassette in the double mutant, strain ArfD:Tc/ArfE:Km. Double crossover event occurred during the conjugation process without sucrose treatment.

Construction of a gene expression plasmid for arfD and arfE

The 3222 bp gene encoding both arfD and arfE was amplified using primer set SacIPPP-ABC79f and KpnIPPP-ABC3300r. It was then cloned into the SacI and KpnI site of an expression vector pME6032 to construct pME6032_arfD/E, and transformed E. coli DH5α. After confirming the correct sequence of the gene fragment, this plasmid or pME6032 vector only was then electrotransferred into wild type MIS38 and strain ArfE:Km. Electroporation was carried out using Electro Gene Transfer Equipment (GTE-10, Shimadzu) with 0·1 cm electrode distance cuvette at a pulse condition of 12·5 kV cm−1, 35 μF for 3 ms. After applying the electric pulse, the cells were cultured in SOC medium at 30°C for 2 h, 150 rev min−1 shaking before spreading on LB agar plate supplemented with 40 mg l−1 Tc. The plates were incubated at 30°C overnight. The obtained strains MIS38(pME6032), MIS38(pME6032_arfD/E), ArfE:Km(pME6032) and ArfE:Km(pME6032_arfD/E) were used for further analysis.

Southern hybridization

Southern hybridization was carried out according to Lim et al. (2007) in order to verify successful gene disruption at a single locus of MIS38 chromosome.

Analysis of arthrofactin production

Analysis of arthrofactin production was carried out according to Roongsawang et al. (2007). Briefly, arthrofactin was precipitated from culture supernatant by addition of diluted HCl to pH 2 followed by centrifuge (10 000 g for 20 min). Hydrophobic fraction of the precipitates containing arthrofactin was extracted by methanol. Samples were then separated by reverse-phase HPLC using Cosmosil 5C18 AR column (4·6 × 150 mm, Nacalai). Detection of compounds was performed either by UV detector (HP1100, Agilent Technologies) or ESI-mass spectrometer (LCQ, Thermo Scientific). The amount of arthrofactin was calculated from the area of peaks recorded by UV detector.

ABC transporter inhibitors and their effect on arthrofactin production

It was shown that sodium ortho-vanadate (Ovd) effectively inhibit the activity of an ABC transporter MacA/B in E. coli (Tikhonova et al. 2007). It has been reported that the Walker A motif in NBD is involved in the Ovd binding for inhibition of ATPase activity (Pezza et al. 2002). Glibenclamide (Glb), a sulphonylurea also has been shown to inhibit the activities of various ABC transporters (Serrano-Martin et al. 2006). Strain ArfE:Km was grown in LB broth for 14 h until the early stationary phase (OD600 ∼ 2·2), after which 0·1 mmol l−1 or 0·25 mmol l−1 of Glb (Sigma), or 2 mmol l−1 sodium Ovd (Wako Pure Chemicals) was added to the culture medium, and further cultivated for 1 and 2 days. Then, the production of arthrofactin was analysed.

Nucleotide and amino acid sequence accession numbers

Nucleotide sequences for the gene encoding putative periplasmic protein (ArfD), putative ABC transporter (ArfE) were registered in the GenBank under accession number AB286215. Amino acid sequences of each protein were submitted under accession number BAC67537 (ArfD) and BAF40423 (ArfE).


Analysis of a putative periplasmic protein gene, arfD and an ABC transporter gene, arfE located downstream of arthrofactin synthetase genes

A putative periplasmic protein gene, arfD, was previously found 127 bp downstream of arfC (orf5 in AB107223). BLASTP analysis of ArfD showed high homology with members of the periplasmic protein component of membrane transporters such as HlyD from P. fluorescens PfO-1 (YP_347946), macrolide ABC efflux type carrier MacA from P. fluorescens Pf-5 (YP_259255), P. syringae pv. tomato DC3000 (AAO56330), and E. coli K12 (P75830) with 91%, 87%, 75%, and 46% identities, respectively. Each protein represents a three-β-strand hammerhead-shaped structure plus an N-terminal fourth strand, β4 (Athappilly and Hendrickson 1995). It was previously shown that lipoyl/biotinyl proteins and these periplasmic protein components share a common fold known as a flattened β-barrel (Johnson and Church 1999).

A putative ABC transporter gene, arfE, was found just 3 bp downstream of arfD. It was consisted of 1953 bp nucleotide sequences encoding a 651 aa protein, ArfE. N-terminal half of ArfE contained a NBD fold which is characterized by two short conserved sequence motifs, named as Walker A (GASGSGKS) and Walker B (VILAD). Another conserved sequence motif called linker peptide (C-loop), LSGGQQQRVS, was also found before Walker B. This linker peptide is the signature of ABC transporter family proteins (Schneider and Hunke 1998).

Located at the C-terminal half of this protein contains five putative TMS (Fig. 1), which probably form a TMD. This structural feature suggested that ArfE belongs to an ABC transporter protein group that possesses NBD and TMD in a single polypeptide (Biemans-Oldehinkel et al. 2006). BLASTP search revealed homology of ArfE with orthologous proteins from various Gram-negative bacteria, such as hypothetical protein PflO1_2215 in P. fluorescens PfO-1 (YP_347947), macrolide ABC efflux proteins MacB in P. fluorescens Pf-5 (YP_259256), P. syringae pv. tomato DC3000 (Q881Q1) and E. coli K12 (P75831) at 92%, 84%, 80% and 53% identities, respectively.

Figure 1.

 Hydropathy plot of ArfE. Abbreviations are NBD, nucleotide binding domain; TMD, transmembrane domain; TMS, transmembrane segment; A, Walker A; B, Walker B; C, C-loop.

Polycistronic transcription of the genes

RT-PCR experiment showed that each spacer region between arfA, arfB, arfC, arfD, and arfE was normally amplified (Fig. 2). This result demonstrates that arthrofactin synthetase genes (arfA/B/C), together with exporter genes arfD and arfE were co-transcribed in a single mRNA, sharing the same promoter for gene expression. This operon structure indicates functionally close connection of each gene. ArfD probably constitutes an ABC-transporter system with ArfE. It should be noted that we could not find a gene encoding an outer membrane protein component such as OprM or TolC homologue in the downstream region of arfE.

Figure 2.

 Reverse transcription polymerase chain reaction of gene gaps using respective primers sets. Total RNA was prepared from 12 h culture in this experiment. Four sets of primers were designed for amplifying approximately 300 bp DNA fragments at each inter-gene locus, gapArfA/B(f) and gapArfA/B(r) (gap between arfA and arfB); gapArfB/C(f) and gapArfB/C(r) (gap between arfB and arfC); and gapArfC/D(f) and gapArfC/D(r) (gap between arfC and arfD); gapArfD/E(f) and gapArfD/E(r) (gap between arfD and arfE). DNA bands were all confirmed to be single from a side view of the gel.

Production of arthrofactin by mutant strains

Successful construction of the mutants at a single locus of the chromosome was verified by both PCR and Southern hybridization experiments. Arthrofactin production was analysed at 6, 9, 12 and 18 h of cultivation in LB broth, where the growth curves (OD600) of the mutants and MIS38 were almost completely fitted. There was a reduction in arthrofactin production by mutant strains ArfD:Km, ArfE:Km, and ArfD:Tc/ArfE:Km, which was similarly reduced to 50% and 70% of strain MIS38 at 6 and 9 h, respectively (Table 3). However, after 12 h, there was no significant difference between the production levels of these mutants and strain MIS38. These results suggested that arthrofactin was dominantly exported by ArfD/E transporter but in the mutant strains eventually capable of being exported by another compatible transport system. Furthermore, RT-PCR experiment indicated that expression level of arthrofactin synthetase gene was similar for strains MIS38 and ArfD:Tc/ArfE:Km (Fig. 3).

Table 3.   Relative percentage (%) of arthrofactin production between wild type MIS38 and its mutant strains
Time (h)MIS38ArfD:KmArfE:KmArfD:Tc/ArfE:Km
  1. Each score is an average of independent duplicate experiments.

Figure 3.

 Reverse transcription polymerase chain reaction showing expression of arfA in wild type strain MIS38 (lanes 1, 3, 5) and a double mutant strain ArfD:Tc/ArfE:Km (lanes 2, 4, 6). A set of primers for amplifying a part of arfA, arfA(594f) and arfA(863r), were used in this experiment.

Overexpression of arfD/E in strains MIS38 and ArfE:Km

Strains MIS38 and ArfE:Km were transformed by either pME6032_arfD/E or pME6032 vector only. The amount of extracellular arthrofactin after 1-day cultivation was increased by 52% in MIS38(pME6032_arfD/E) when compared with MIS38(pME6032) (Table 4). Similar effect was observed for ArfE:Km (pME6032_arfD/E) where an increase level was 54%. These results indicate that arfD and arfE contribute to effective exportation of arthrofactin. However, after 2 days, production levels of arthrofactin were indistinguishable between MIS38(pME6032_arfD/E) (94%) or ArfE:Km(pME6032_arfD/E) (98%) and vector controls (100%). This result again demonstrates that ArfD/E is not essential exporter system for arthrofactin production in the late growth stage.

Table 4.   Relative percentage (%) of arthrofactin productivity between wild type MIS38 and gene overexpression mutants
Time (days)MIS38 (pME6032)MIS38 (pME6032_arfD/E)ArfE:Km (pME6032)ArfE:Km (pME6032_ arfD/E)
  1. Each score is an average of independent duplicate experiments.


Effect of ABC transporter inhibitors on arthrofactin production

It is known that arthrofactin carries strong surfactant activities, therefore there is a possibility that arthrofactin was passively exported by diffusion through cellular membranes in the ArfD and ArfE mutant strains. Another possibility is that resistance-nodulation-cell division (RND) antiporter efflux systems are involved in the arthrofactin exportation. In order to examine these possibilities, several inhibitors for general ABC transporters were tested for arthrofactin production.

It was first confirmed that the colony forming units (CFU) were not seriously affected by treatment of Glb and Ovd at tested concentrations, suggesting that cellular metabolisms functioned normally over the time (Fig. 4). There was significant reduction of arthrofactin production by ArfE:Km in the presence of 0·1 mmol l−1 Glb (73%), 0·25 mmol l−1 Glb (85%), and 0·25 mmol l−1 Glb + 2 mmol l−1 Ovd (87%) in 1 day (Fig. 4). Inhibitory effect of 2 mmol l−1 Ovd was small in 1 days (26%) but it became obvious after 2 days (94%). These results allowed us to conclude that exportation of arthrofactin is highly dependent on ABC transporters that require the energy derived from ATP hydrolysis.

Figure 4.

 Arthrofactin production of ArfE:Km in the absence and presence of their respective ABC inhibitors in 1 and 2 days. Glb, Ovd, and cfu are glibenclamide, sodium ortho-vanadate, and colony forming units, respectively. Standard deviations were calculated from two independent experiments.


Genes encoding functionally connected enzymes often form a cluster or an operon structure in the bacterial chromosome. This seems to fit the case for a NRPS and the product transporter genes. P. syringae pv. syringae strain B301D-R produces two lipodepsipeptide phytotoxins, syringomycin (Syr) and syringopeptin (Syp), and whose synthetase gene clusters are adjacently located with opposite direction. Secretion of these phytotoxins requires two transporter systems, known as SyrD, a protein homologous to membrane proteins of the ABC transporter family, and PseABC, a tripartite transporter system homologous to RND efflux system. syrD and pseABC are located just downstream of Syr and Syp synthetase genes, respectively. A mutation in syrD has been shown to lead almost completely loss of both Syr and Syp production (Quigley et al. 1993; Grgurina et al. 1996). Moreover, transcription level of syrB, the synthetase gene, was reduced to 60% (in 2 days) and 35% (in 4 days) by syrD mutation. Then, it was concluded that SyrD is required for full expression of syrB. Interestingly, there is no periplasmic and outer membrane protein counterpart gene in the syr cluster. On the other hand, pseC mutant strain showed mild reduction in Syr production by 41% at 72 h and Syp production by 67% at 48 h compared to wild type strain B301D-R. There is no report that production levels of Syr and Syp were restored later by another transporter in the syrD or pseABC mutants.

Pyoluteorin is a chlorinated polyketide antibiotic secreted by P. fluorescens Pf-5. Brodhagen et al. (2005) showed that pltI (encoding a periplasmic protein) and pltJ (encoding an ABC transporter) mutant strains displayed low pyoluteorin production (23–30% of wild type strain) at 48 h and did not accumulate proportionately more of the pyoluteorin intracellularly. Interestingly, transcription of pltI and pltJ was enhanced by exogenous pyoluteorin. These are known as reciprocal regulation mechanisms to prevent intracellular accumulation of the product. In the case of pyoluteorin production by Pseudomonas sp. M18, the gene disruption of corresponding periplasmic protein and ABC transporter protein both led to a nondetectable production of this antibiotic (Huang et al. 2006). We also observed that disruption of the gene encoding ABC transporter for pyoverdine (pvdE38-ABC) dramatically reduced production level of pyoverdine, 9% of MIS38 at 72 h (unpublished data). In another study of Gram positive Bacillus subtilis 168 by Tsuge et al. (2001), disruption of yerP, an RND efflux protein gene homologue resulted in 6-fold reduction of a lipopeptide biosurfactant surfactin production.

In contrast to above information, observation results were very different in each mutant of exporter genes for arthrofactin. Although polar effect of arfD mutation on the expression of arfE cannot be ruled out, our experimental results suggest that effective exportation only in the early stage of production requires active ArfD/E, whose genes constitute an operon with arfA/B/C synthetase genes. The distance between arfC and arfD was relatively large, 127 bp, for an operon structure. However, there is an example that the distance between sypB and sypC is 423 bp in syringopeptin synthetase gene operon (AF286216). Gene disruption of arfD/E still allowed normal transcription level of arfA/B/C (Fig. 3) and did not lead to dramatic loss of arthrofactin production, supporting assumption that there is another transport system that is functionally flexible enough to export arthrofactin. On the other hand, overexpression of arfD and arfE led to an increase of 50% arthrofactin production in 1 day, indicating that these genes are indeed important for exportation, but the production level was not different after 2 days between the overproducers and MIS38. Exportation may not be a rate limiting step for the production of arthrofactin at this time. In summary, the production level of arthrofactin was highly controlled at a proper level. Although we tried to detect intracellular arthrofactin by sonication and methanol extraction, little accumulation of arthrofactin was observed in mutants and overproducing strains as well as MIS38. This result suggests that production of arthrofactin is coupled with translocation. Because MIS38 was isolated from a highly hydrophobic oil field, production of the strong biosurfactant may be essential to survive and keep preferred habitats.

Escherichia coli have two notable systems involved in the resistance of macrolide antibiotics, i.e. the AcrAB-TolC and MacAB-TolC system. AcrAB belongs to the RND efflux pumps whereas MacAB belongs to the ABC transporters superfamily (Zgurskaya and Nikaido 2000; Kobayashi et al. 2001). Both these protein constitute a complete transporter system with one of multifunctional outer membrane proteins, TolC, for the translocation of substrate antibiotics from periplasm to the extracellular space. Overexpression of macAB in E. coli KAM3 (ΔmacAB) similarly enhanced resistance to erythromycin, clarithromycin, and oleandomycin by 8-fold, demonstrating functionality of the gene products with multidrug efflux (Kobayashi et al. 2001). In P. aeruginosa, four RND multidrug efflux systems including MexAB-OprM, MexCD-OprJ, MexEF-OrpN, and MexXY-OrpM are known (Poole 2001). It is not clear whether ArfD/E recruits an outer membrane protein component like OprM/J/N or TolC to form an active exporter apparatus for arthrofactin. There are significantly high similarities between ArfD/E and putative MacA/B homologues from Pseudomonas strains such as P. fluorescens PfO-1 (YP_347946/YP_347947), P. fluorescens Pf-5 (YP_259255/ YP_259256), P. putida (ABW17378/ABW17379), and P. syringae pv tomato (Q881Q2/ Q881Q1). All these sets of ABC transporter system lack outer membrane protein components in the gene cluster. This is the first report that characterizes this group exporter for nonribosomal peptides in Pseudomonas cells. When we look into the genome sequence of P. fluorescens PfO-1, there are five TolC like outer membrane protein components of putative ABC transporters (YP_348413, YP_347194, YP_346224, YP_345867, YP_345894) and thirteen putative outer membrane protein components that would constitute tripartite RND efflux systems including NodT and OprN. Further study on these candidate proteins is necessary to clarify the whole view of exceptionally flexible arthrofactin-family nonribosomal peptide exporter system in Pseudomonas cells.


Escherichia coli SM10λpir, pSMC32, pCVD442 and pME6032 were kindly donated by Dr D. A. Hogan (Darthmouth Medical School, NH) and Dr Dieter Haas (Université de Lausanne, Switzerland).

S.P.L acknowledges her PhD fellowship from MEXT (no. 040318). This work was supported by a Grant-in-Aid for Scientific Research for Exploratory Research from Ministry of Education, Culture, Sports, Science and Technology (MEXT) (nos. 17510171, 19380189), Institute for Fermentation, Osaka (IFO), and New Energy and Industrial Technology Development Organization (NEDO).