The proteolytic fragments of OprFs of Pseudomonas aeruginosa and Pseudomonas fluorescens were identified, respectively, as the first 175 and 177 amino acids from the N-terminal domain. They induced ion channels after reincorporation into planar lipid bilayers (85 and 75 pS, respectively, in 1 M NaCl). A similar conductance value (72 pS) was found for the eight β-strand OmpA N-terminal domain (OmpA171) of Escherichia coli. We conclude that the N-terminal domain of OprFs is sufficient to induce ion channels and the comparison with OmpA171, provides strong evidence of the existence of an eight-stranded β-barrel in the N-terminal domain of OprFs.
Pseudomonas aeruginosa is a Gram-negative pathogenic bacterium that is more involved in severe hospital infection due to its resistance to a large set of common antibiotics. Although the production of periplasmic β-lactamases and the induction of efflux pump can be involved in this resistance, the main cause remains the small permeability of its outer membrane [1,2]. In fact, the permeability of the P. aeruginosa outer membrane was shown to be 10–100-fold lower than that Escherichia coli[3,4]. Proteins, called porins, are embedded in this membrane and form water-filled channels responsible for the permeability barrier. Among these porins, the major constitutive porin OprF  is monomeric and non-specific, and hydrophilic compounds pass through the pore by passive diffusion. The ion channel experiments carried out with OprF from P. aeruginosa and P. fluorescens inserted into planar lipid bilayers gave similar 360-pS channels in 1 M KCl  and 240–250-pS channels in 1 M NaCl [7–9].
OprF of P. aeruginosa seems to be multifunctional since it is non-covalently linked to the peptidoglycan and involved in the maintenance of cell shape . These characteristics are common to the OmpA protein from E. coli. A putative structure of OprF from P. aeruginosa would comprise two main domains: a N-terminal part of about 160 residues containing eight antiparallel β-sheets able to form a β-barrel and a C-terminal domain strongly associated with peptidoglycan which is highly conserved with the C-terminal domain of E. coli OmpA [10,11]. The two domains would be separated by a repeating Pro–Ala region and two disulfide bonds between the 160th and 210th residues. Without crystallographic data to date, the functional tertiary structure of OprF is not well established. Nikaido's group agrees with a periplasmic C-terminal part , whilst other authors hypothesise that OprF is constituted of 16 β-strands, suggesting that the structure of the C-terminal part is not globular [11,13]. Nevertheless, there is no consensus to propose the number of β-strands necessary for building the ion channel in native OprF. Moreover, the exact role of N- and C-terminal domains in the pore-forming activity remains to be elucidated.
In order to clarify these points, we tested the pore-forming properties of the N-terminal fragments OprF175 and OprF177 obtained after enzymatic digestion of native OprFs from P. aeruginosa and P. fluorescens. A similar study was carried out with the N-terminal domain from OmpA of E. coli (OmpA171). This OmpA171 was crystallised by Pauscht and Schulz to a resolution of 2.5 A, and revealed an extended eight-stranded β-barrel . The comparison of conductance values obtained in this study from the different N-terminal domains with previously reported native OprFs and OmpA data [7,8,15] allowed us to discuss the involvement of the C-terminal domain of these proteins in the channel architecture in the native porins.
2Materials and methods
2.1Bacterial strains and growth conditions
Pseudomonas aeruginosa PAO1 and Pseudomonas fluorescens MFO were grown, respectively, at 37°C and 28°C in Nutrient Broth (Difco). After vigorous shaking overnight, cells were harvested in late exponential phase by centrifugation at 8000×g for 10 min at 4°C (Sorvall RC5B, rotor GSA).
2.2Isolation of OprF proteins and proteolytic fragments
Outer membranes (OMs) from P. aeruginosa and P. fluorescens were isolated by the method of Mizuno and Kageyama  with the modifications described by Dé et al. . OprF purification was performed by preparative electrophoresis followed by electroelution in the presence of SDS. This method was also used for the purification of proteolytic fragments obtained by pronase digestion. 500 μg of OprF was digested with pronase from Streptomyces griseus (Fluka) at 200 μg ml−1. The homogeneity of the samples was confirmed by analytical SDS–PAGE (7/15%) with silver staining. The concentration of the electroeluted proteins was determined by the Micro BCA kit (Pierce) or estimated from SDS–PAGE stained with Coomassie brilliant blue G250 using increasing BSA concentrations as a standard. Both electroeluted fragments were dialysed against 0.3% octyl-polyoxyethylene (octyl-POE from Bachem) buffered with 10 mM N-2-hydroxyethylpiperazine-N′-2 ethanesulfonic acid (HEPES), pH 7.4.
2.3Mass spectrometry matrix-assisted-induced desorption and ionisation time of flight (MALDI-TOF)
Mass spectra were obtained with a time of flight mass spectrometer (Voyager Elite XL, Perseptive Biosystems, Framingham, MA, USA). All spectra were acquired in the positive-ion mode and the acceleration voltage was set to 20 000 V. Aliquots of 0.5 μl of the protein solution and 0.5 μl of 2.5 dihydroxybenzoic acid dissolved in a 50% (v/v) of acetonitrile/aqueous 0.1% TFA solution were mixed on the stainless plate and dried prior to analysis. External calibration was performed with bovine serum albumin (m/z 66431).
The N-terminal sequence analysis of OprF177 and OprF175 was determined by automated Edman degradation (477A Protein Sequencer, Applied Biosystems) after electroblotting on polyvinylidene difluoride membrane (Millipore).
2.4Reconstitution into planar lipid bilayers
Virtually solvent-free planar lipid bilayers were formed by the method of Montal and Mueller , as modified by Saint et al. . Briefly, from a diphytanoylphosphocholine (DPhPC, Avanti, Birmingham, USA) solution in hexane, lipid bilayers were formed by the apposition of two monolayers on a 100-μm-diameter hole in a thin Teflon film (10 μm) sandwiched between two half glass cells pre-treated with hexadecane/hexane solution (1:40, v/v). Bilayer formation was monitored by the capacitance response and the voltage and current sign conventions are the usual ones. The electrolyte solution was 1 M NaCl, 10 mM HEPES, (pH 7.4). The proteins (2 ng in 0.3% octyl-POE) were added to the measurement compartments in a symmetric manner.
3Results and discussion
We have shown in previous studies that OprFs from P. aeruginosa and P. fluorescens as well as OmpA of E. coli could form ion channels in planar lipid bilayers [7,8,15]. In order to test the eventual pore-forming properties of the N-terminal domain of OprFs of P. aeruginosa and P. fluorescens, enzymatic cleavage with pronase was performed on native proteins. The OprFs, isolated from OM of P. aeruginosa and P. fluorescens, were treated with pronase (200 μg ml−1) for 6 h and in both cases yielded proteolytic fragments of apparent molecular mass ranging from 21 to 24 kDa. The protease-resistant fragments were purified by SDS–PAGE, followed by an electroelution (Fig. 1). Their exact molecular masses, determined by MALDI-TOF mass spectrometry, were 18951 and 18667 Da, respectively, for the P. aeruginosa and P. fluorescens fragments. In order to determine the nature of these remaining proteolytic fragments, their N-termini were sequenced, and the QGQGAVEGELFYKKQ and QGQNSVEIEAFGKRY sequences were obtained for the N-terminus of the native OprFs from P. aeruginosa and P. fluorescens, respectively. From the known amino acid sequences of these OprFs [8,18] and the molecular mass of both fragments, it was possible to calculate the position of the proteolytic cleavage. The cleavages occurred at valine-175, just before the disulfide bridges, for P. aeruginosa (OprF175) and at threonine-177 in the proline-rich region for P. fluorescens (OprF177). Besides these fragments, no other proteolytic fragments >10 kDa were detected, showing that the C-terminal part of the molecule was not recovered and was probably digested.
These results show that the N-terminal domain of OprFs is not sensitive to pronase while the C-terminal part is partially or totally digested. A similar result was already described for OmpA digestion by pronase, where a unique 19-kDa protease-resistant fragment was found, corresponding to the N-terminal part . The resistance of the N-terminal part of OmpA to protease treatments is likely to be due to a very tight structural domain, as suggested by the high proportion of antiparallel β-strands in this protein [20,21]. This hypothesis was recently confirmed by the crystal structure of the N-terminal fragment 1–171 of E. coli OmpA (OmpA171) revealing that this domain consists unambiguously of an extended eight-stranded β-barrel . The N-terminal transmembrane domain of OmpA has been reported to be exceptionally stable since it remains undenatured even in SDS at up to 80°C . The N-terminal fragments of OprFs from P. aeruginosa and P. fluorescens we purified by SDS–PAGE followed by electroelution, are likely undenatured domains. Thus, the N-terminal end of OprFs are predicted to contain eight β-strands like OmpA suggesting a strong stability of this domain in presence of SDS.
To determine the contribution of the N-terminal domain in the pore-forming activity of native OprFs, OprF175 and OprF177 were reconstituted into planar lipid bilayers. A few minutes after addition of OprF175 into the electrolytic compartment, current fluctuations were observed when a voltage is applied. The average single-channel conductance value was (85±3) pS in 1 M NaCl, 10 mM HEPES, pH 7.4 (Fig. 2A). The incorporation of the OprF177 showed similar fluctuations of current as those observed for OprF175, corresponding to a (75±3) pS conductance value in 1 M NaCl, buffered with 10 mM HEPES, pH 7.4 (Fig. 2B). For both OprF175 and OprF177, single-channel currents were a linear function of voltage and no voltage-dependent closures of channels were observed between −170 mV and +170 mV. In the same way, we tested a recombinant protein of E. coli obtained from the Schulz's group, which corresponds to the N-terminal fragment 171 of OmpA (OmpA171). This protein was reincorporated into planar lipid bilayers in similar conditions and induced current fluctuations corresponding to a conductance value of (72±3) pS in 1M NaCl (Fig. 2C).
It can be noticed that ion channels formed by both OprF175 and OprF177 fragments in planar lipid bilayers are characterised by short lifetimes of a few hundred milliseconds. In comparison, OmpA171 showed slower transitions of current when reconstituted in the same conditions. Higher conductance values, much rarer than the current fluctuations described above, ranging from 150 to 600 pS were also observed for the three N-terminal domains. They could correspond to large aggregates formed by these N-terminal fragments in the lipid bilayer.
Thus, the N-terminal domains of both OprFs as well as of OmpA are able to induce similar ion channels in planar lipid bilayers. Our findings that OmpA171 can form ion channels is consistent with recent studies showing ion channel activities with the N-terminal part of a refolded tryptophan mutant of OmpA (residues 1–176) . The conductance value they observed with this Trp mutant was 78 pS in 1 M KCl. Experiments done in our laboratory with OmpA171 in 1 M KCl gave a conductance value of 110 pS (Molle, personal communication). The slight difference between our conductance value in 1 M KCl and Arora's value could be due to the fact that we did not investigate exactly the same fragment. OmpA171 is a recombinant mutant, used for crystallographic experiments, with three substitutions (F23L/Q34K/K107Y) and one additional methionine at the N-terminal end.
The crystal structure of OmpA171 shows that the β-barrel interior would contain no channel. In these conditions, passage of ions through this structure is difficult to understand. However, formation of ion channels could be explained by the presence of a more dynamic structure of the barrel in lipid bilayers than in protein crystals as suggested by Arora et al. . This hypothesis is supported by the OmpA crystallographic study  which has revealed the high mobility of three extracellular loops of the transmembrane domain, suggesting a potential flexibility of this part of the molecule.
The similar ion channel activity of the three N-terminal domains observed in the present study suggests that the pores of OprF175 and OprF177 are very likely constituted of eight β-strands as OmpA171. This would confirm the previous structural predictions made for the N-terminal domain of OprF [11,13,20]. Thus, ion channel activity data and structural models would suggest similar function for the amino-terminal ends of OprFs and OmpA. This conclusion is quite surprising, knowing that the N-terminal region of these proteins shows only 18.5% similarity of amino acid sequences . However, there is growing evidence that porins can adopt a similar β-barrel structure despite having no sequence similarity with one another .
If we look at the conductance values previously found for the native OprFs of P. aeruginosa, P. fluorescens and the OmpA of E. coli, ranging from 230 to 250 pS in 1 M NaCl [7,8,15], it appears that the native proteins would form ion channels with a conductance value approximately three times larger than the respective N-terminal transmembrane domain. This would indicate that the C-terminal domain of OprFs and OmpA is largely involved in the pore formation of native proteins. This conclusion agrees with a recent study, which showed that an OmpA mutant of E. coli would induce channels with two conductance states in planar lipid bilayers . The authors claimed the smaller channels would be associated with the N-terminal transmembrane domain, whereas both domains would be required to form the larger channels.
In conclusion, we have demonstrated in this study that the N-terminal domains of OprFs from Pseudomonas strains are able to form ion channels in planar lipid bilayers similar to those induced by OmpA N-terminal transmembrane domain, suggesting that the eight-stranded β-barrel structure of OmpA N-terminus is common to OprFs.
The authors wish to thank Dr. Schulz for the generous gift of OmpA171 and Alexandre Stella for determining the N-terminal sequences of the proteins. This work was supported by the GDR 790 CNRS.