Influence of the passenger domain of a model autotransporter on the properties of its translocator domain

Abstract

Autotransporters are a superfamily of proteins secreted by Gram-negative bacteria including many virulence factors. They are modular proteins composed of an N-terminal signal peptide, a surface-exposed ‘passenger’ domain carrying the activity of the protein, and a C-terminal ‘translocator’ domain composed of an α-helical linker region and a transmembrane β-barrel. The translocator domain plays an essential role for the secretion of the passenger domain across the outer membrane; however, the mechanism of autotransport remains poorly understood. The whooping cough agent Bordetella pertussis produces an autotransporter serine-protease, SphB1, which is involved in the maturation of an adhesin at the bacterial surface. SphB1 also mediates the proteolytic maturation of its own precursor. We used SphB1 as a model autotransporter and performed the first comparisons of the biochemical and biophysical properties of an isolated translocator domain with those of the same domain preceded by the C-terminal moiety of its natural passenger. By using cross-linking and dynamic light scattering, we provide evidence that the passenger domain promotes the auto-association of SphB1, although these interactions appear rather labile. Electrophysiological studies revealed that the passenger domain of the autotransporter appears to maintain the translocator channel in a low-conductance conformation, most likely by stabilizing the α-helix inside the pore. That the passenger may significantly influence AT physicochemical properties is likely to be relevant for the in vivo maturation and stability of AT proteins.

• Autotransporter
• SphB1 protein
• Bordetella pertussis
• passenger domain
• protein secretion

Introduction

The transport of proteins across membranes to their final destination in the cell is an essential function of all living organisms. Secretion to the surface or the extracellular milieu of enzymes, adhesins and toxins is frequently an important strategy of bacterial pathogenesis. Gram-negative bacteria have a complex cell envelope composed of two distinct membranes, the cytoplasmic and the outer membranes, and they have developed specific mechanisms to translocate proteins across the outer membrane. Among these secretion pathways, the ‘type V’ pathway is widespread in Gram-negative bacteria and dedicated to the secretion of so-called autotransporter (AT) proteins [1], [2]. The proteins of the AT superfamily are characterized by a modular structure. They are composed of a signal-peptide for Sec-dependent export across the cytoplasmic membrane, a ‘passenger’ domain that carries the activity of the protein, and a C-terminal ‘translocator’ domain involved in the secretion process across the outer membrane. While the passenger domains of AT are quite varied and relate to the particular function of the proteins – adhesins, proteases, etc. [3] – translocator domains are much more conserved and belong to one of only two families, i.e., the conventional ATs [4], [5] or the trimeric AT family [6–8]. The X-ray crystal structures of the translocator domains of two AT proteins have been reported recently. NalP is a protease that belongs to the conventional ATs [9]. Its translocator domain is monomeric and forms a 12-stranded β-barrel with an α-helix in the center of the pore [10]. The NalP structure probably represents the translocator domain in the outer membrane after secretion of the passenger. The second structure, that of the translocator domain of the trimeric AT adhesin Hia, is also a 12-stranded β-barrel, with each of the 3 monomers contributing 4 transmembrane β-strands and an α-helical linker [11]. Thus, three long α-helices are found in the centre of the β-barrel. The similarities between the Hia and NalP translocator structures argue in favor of related mechanisms of secretion for the two families of AT proteins.

The mechanism of AT translocation is currently the focus of intense debate. These proteins were initially proposed to be autonomous for secretion across the outer membrane, which led to coin the name autotransporter [12], [13]. According to that initial model, the β-barrel domain of the AT inserts into the outer membrane and mediates the translocation of the passenger domain covalently linked to it. An alternative mechanism was suggested by the identification of an oligomeric complex in the outer membrane formed by the assembly of at least 6 translocator domains of the prototypic AT IgA protease [14]. In that case, the corresponding passenger domains were proposed to be secreted through a common central channel in the middle of the ring-like structure. Several studies have been carried out with artificial passenger domains, and therefore their results might not necessarily be extrapolated to the natural AT proteins. Recently, it has been proposed that the outer membrane Omp85 (or YaeT) complex might mediate translocator insertion into, and possibly also passenger secretion across, the outer membrane [10], [15], [16].

The whooping cough agent Bordetella pertussis secretes a number of virulence proteins that participate to its infectious cycle [17]. We have recently identified a serine-protease AT called SphB1, which is involved in the maturation of an important adhesin of B. pertussis, the filamentous haemagglutinin (FHA) [18]. We have shown that FHA maturation by SphB1 plays a role in the pathogenicity of B. pertussis in an animal model of infection [19]. SphB1 is a conventional AT, with a translocator domain of approximately 35 kDa. In addition to processing FHA, SphB1 was shown to mediate the proteolytic maturation of its own precursor at the bacterial surface, essentially in trans [20]. Sequence alignments of SphB1 with several homologues have suggested that their passenger domains are composed of two domains, a subtilisin-like domain with a typical Asp-His-Ser catalytic triad, followed by a 200-residue-long domain of unknown function, hereafter called the ‘X’ domain [18].

In this work, we used SphB1 as a model AT protein and performed the first comparison of the biochemical and biophysical properties of an isolated translocator domain with those of the same domain preceded by the C-terminal portion of its natural passenger. We show that the passenger domain of SphB1 influences both the self-association of the protein and the conductance properties of the SphB1 channels.

Materials and methods

Chemicals

The homobifunctional crosslinker suberic acid bis (N-hydroxysuccinimide ester) (DSS) was purchased from Sigma (France). The detergents n-dodecyl-ß-D-maltoside (DDM) and octyl-ß-D-glucopyranoside (OG) were obtained from Sigma and Fluka (France), respectively.

Plasmid construction

All expression constructs except for pFJD158 include the sequence coding for an affinity Strep tag located 5 residues after the N-terminus of the mature protein. Preliminary experiments with SphB1-α-β that was tagged either with a Strep-tag or a 6-His tag in identical positions indicated that affinity chromatography using the Strep-tag (see below) resulted in a better purification, albeit with a slightly lower yield.

pT7SB1-X-β-S is a vector for the production in E. coli of the second portion of the SphB1 passenger (‘X’ domain, called domain IV in [18]) followed by the C-terminal translocator domain. It was obtained as follows, using an ampicillin-resistant version of pFJD118 [21] called pT7Fcc3His6. First, the BamHI-HindIII fragment of pT7Fcc3His6 was replaced by a PCR fragment encoding the C-terminal portion of B. pertussis pertactin obtained using the oligonucleotides prn-CT-Up1 (5′-ggatccgagttgcgcctgaatccg-3′) and prn-CT-stop (5′-aagcttccaagctccaggaaaaactc-3′) as primers, thus yielding pT7prn-Ct. A linker obtained by annealing the partially overlapping oligonucleotides Stag-Up1 (5′-gatcttggtcgcacccgcagttcgagaagg-3′) and Stag-Lo1 (5′-gatcccttctcgaactgcgggtgcgaccaa-3′) was then inserted into the BamHI site of the latter plasmid, and its orientation was checked by sequencing. This linker codes for the affinity Strep-tag, WSHPQFEK. The resulting plasmid was called pT7prn-Ct-Stag. Then, a central portion of sphB1 was amplified by PCR using the oligonucleotides SbX-Up (5′-ggatccacggatacctccacgttcgg-3′) and gn8 (5′-tctcgaggcggtccatccag-3′) as primers. The amplicon was inserted into the pCRII-Topo cloning vector and sequenced. It was then used as a BamHI-XhoI restriction fragment in a three-way ligation with a XhoI-HindIII fragment from pIC-2K-Ct, encompassing the 3′ portion of sphB1 [20], and the HindIII-BamHI vector portion of pT7prn-Ct-Stag. This resulted in pT7SB1-X- β -S.

pT7SB1-α- β -S is a vector for the production in E. coli of the C-terminal β-barrel domain of SphB1 preceded by a short region that comprises the predicted α-helix (prediction by the PSIPRED server, http://bioinf.cs.ucl.ac.uk/psipred/; see below). It was obtained as follows. The 3′ portion of sphB1 was amplified by PCR using the oligonucleotides SB-Ct-Lk (5′-ggatccgcgatgcgcgaactcgatg-3′) and SB-Ct-Lo (5′-aagcttcagtagcggtaagtgaggct-3′) as primers. After cloning in pCRII-Topo and sequencing, the amplicon was inserted as a HindIII-BamHI fragment into the same sites of pT7prn-Ct-Stag, yielding pT7SB1-α-β-S.

pT7SB1-β-S is a vector for the production in E. coli of the β-barrel domain of SphB1 without the predicted α-helix. It was constructed as follows. A shorter 3′ portion of sphB1 was amplified using the oligonucleotides SB-Ct-Lo (see above) and SBnoLk (5′-ggatccgagcgcgtggcggccat-3′) as primers. After cloning in pCRII Topo and sequencing, the amplicon was inserted as a HindIII-BamHI fragment into the same sites of pT7prn-Ct-Stag in replacement of the prn fragment, yielding pT7SB1-β-S.

pFJD158 is a vector for the expression of the entire SphB1_(Ser-Ala) precursor in E. coli, which was obtained as follows. First, a central portion of sphB1 with a point mutation resulting in the replacement of the active Ser residue by Ala was amplified using pIC135 as the matrix [18] and the oligonucleotides SB1-exp1-UP (5′-ggatccgaagtgccggccgca-3′) and SB1-exp1-Lo (5′-tgcgggtcgacgccgtac-3′) as primers. After cloning in pCRII-Topo and sequencing, the amplicon was used as a BamHI-BsiWI restriction fragment in a three-way ligation with a BsiWI-PstI 1200-bp fragment from pIC-SphB1 [20] and the BamHI-PstI vector portion of pFJD138. pFJD138 is a kanamycin-resistant version of pT7SB1-α-β-S, with a His₆ tag (unpublished). In that case, the His-tag was chosen rather than the Strep-tag, because the full-length protein was produced at very low levels. The plasmid resulting from the three-way ligation was called pFJD158.

Production and purification of the recombinant proteins

The various plasmids were introduced into E. coli BL21(DE3)omp5 [22]. The bacteria were grown in liquid LB medium at 37°C under agitation until the cultures reached absorbencies of 0.8 to 1 at 600 nm. The expression of the recombinant genes was then induced by the addition of 1 mM IPTG, after which the bacteria were grown in the same conditions for 3 h. They were then harvested by centrifugation and resuspended in 1/30 volume 20 mM sodium phosphate (NaPi) (pH 7) containing 10 µg/ml DNAse I and a tablet of Complete® antiprotease cocktail (Roche). The bacteria were broken by 2 passages in a French pressure cell. Bacterial debris were harvested centrifugation at 6000 g for 20 min, and the membrane fractions were recovered in the pellet after a 50-min ultracentrifugation at 100,000 g, 6°C and stored at −80°C until further use.

Differential detergent extraction was performed by resuspending the total membrane fractions in 20 mM NaPi (pH 7) containing 0.8% OG and 250 µg/ml AEBSF (Roche). The suspensions were rocked for 1 h at room temperature, after which the insoluble material containing the recombinant proteins was harvested by ultracentrifugation as above. The pellets were resuspended in 20 mM NaPi (pH 7)/0.5% dodecylmaltoside and rocked for 1 h at room temperature. The recombinant SphB1 proteins were solubilized by the detergent. Ultracentrifugation was performed again to discard the remaining insoluble material, while the recombinant proteins were found in the supernatants. The supernatant fractions were applied onto Strep-Tactin® resin (IBA GmbH, Göttingen, Germany) for affinity chromatography according to the manufacturer's instructions. The recombinant proteins bound to the matrix, and following a washing step with 100 mM Tris-HCl (pH = 8)/ 0.1% DDM/ 150 mM NaCl/ 1 mM EDTA, they were eluted in the same buffer containing 2.5 mM desthiobiotin.

Electrophoretical, enzymatic and analytical methods

Samples for SDS-PAGE were separated on 7%/12.5% (stacking/resolving) SDS/polyacrylamide gels, and the proteins were stained with Blue Coomassie G-250. Protein concentrations were determined using the DC-protein assay from Bio-Rad (France). Enzymatic digestion was performed at 37°C with proteins solutions of 0.350 mg/ml and Trypsin (Promega, France) at 100 u/ml, in a 4/1 (protein/enzyme) ratio. Digestion was stopped by adding 0.1 mM of N⁻p-Tosyl-l-Lysine (TLCK, from Sigma, France), and the samples were kept at –20°C before analysis. N-terminal sequences were obtained using an Applied Biosystems Procise 492A sequencer after protein electroblotting onto polyvinylidene difluoride membranes (Millipore, Bedford, USA).

Circular dichroism spectroscopy

Proteins were dialyzed overnight at 4°C against 0.1% DDM/ 150 mM NaCl/ 10 mM Tris-HCl (pH 8) before measurements. CD spectra were recorded at room temperature on a Chirascan dichrograph (Applied Photophysics) in a quartz cell with an optical path of 0.1 mm.

Five scans were performed between 195 and 260 nm. The protein concentration was 0.250 mg/ml. A blank sample (buffer plus detergent) was analyzed in the same conditions. The percentage of the various secondary structures was calculated with the DICHROWEB software [23].

Dynamic Light Scattering (DLS)

Dynamic light scattering (DLS) measurements were performed at 25°C using a Zetasizer Nano ZS (Malvern Instruments, UK), fitted with a 4mW He-Ne 633 nm laser. The scattered light was detected at an angle of 173° with respect to the incident laser beam. DLS measurements were carried out on purified SphB1-X-α-β and SphB1-α-β obtained after dialysis against 0.1% DDM/ 150 mM NaCl/ 20 mM HEPES (pH 7.8). The final concentrations of SphB1-X-α-β and SphB1-α-β were varied from 0.5 mg/ml to 3 mg/ml. The DLS measurements were performed in the absence and in the presence of SDS. The concentration of SDS was systematically varied from 0.1 to 1% by adding incremental volumes of a 10% (w/v) SDS stock solution. All solvents were prepared with pure Milli-Q water and prefiltered through a 0.2 µm Millipore membrane before use. For measurement purposes, a 50 µl UVette cell from Eppendorf (Hamburg, Germany) was used.

The experimental DLS data were analyzed with Dispersion Technology Software (DTS), package version 4 provided by Malvern Instruments. The volume size distributions were estimated from the experimental correlation curves after fitting by the General Purpose algorithm implemented in the DTS package.

Chemical crosslinking of recombinant proteins

Cross-linking reactions of the recombinant proteins were performed with DSS [24] at pH 6 in 0.1% DDM/ 150 mM NaCl/ 20 mM NaPi, or at pH 8 in 0.1% DDM/ 150 mM NaCl/ 20 mM HEPES. The experimental conditions of the cross-linking reactions, i.e., protein concentrations of approximately 10 µM and incubation time shorter than 2 h, were optimized to prevent excessive hydrolysis of the cross-linker. DSS was solubilized in dimethyl sulfoxide to a concentration of 20 mM (stock solution). 2 mM of DSS were mixed to 4 µg of protein, supplemented or not with 0.1 or 1% SDS (w/v), and incubated for 1 h at room temperature. Cross-linking reactions were terminated by the addition of 1 M Tris solution.

Planar lipid bilayer experiments

Virtually solvent-free planar lipid bilayers were formed over a 125–200 µm hole in a polytetrafluoroethylene film (10 µm thick) pretreated with a mixture of 1:40 (v/v) hexadecane/hexane and sandwiched between two half glass cells as described [21]. Phosphatidylcholine from soy beans (azolectin from Sigma type IV S), dissolved in hexane (0.5%) was spread on the top of the electrolyte solution (1 M KCl/10 mM HEPES, pH 7.4) in both compartments of the measuring cell. Bilayer formation was achieved by lowering and raising the level up in one or both compartments and monitoring capacity responses. Voltage was applied through an Ag/AgCl electrode in the cis-side and the trans-side was grounded. The purified SphB1 proteins were added to the cis-side (5–100 ng/ml). Single channel currents were recorded with a BLM 120 amplifier (Bio-Logic, Claix, France) and stored on a CD recorder (DRA 200, Bio-Logic, Claix, France) for off-line analysis. CD data were then analysed by the WinEDR and Biotools softwares (Bio-Logic, Claix, France). In macroscopic conductance measurements, the doped membranes were subjected to slow ramps of potential (10 mV/s), and transmembrane currents were fed into an amplifier (BBA-01, Eastern Scientific, Rockville, USA). Current-voltage curves were stored on a computer and analysed with Scope software (Bio-Logic, Claix, France). All measurements were performed at room temperature.

Results and discussion

Preparation of the SphB1-derived recombinant proteins

To refine the initial predictions of the domain structure of SphB1, the protein sequence was subjected to in silico analyses using the Psi-Pred and Betawrap softwares (http://bioinf.cs.ucl.ac.uk/psipred/ and http://betawrappro.csail.mit.edu/). Schematically, those analyses yielded the following results, with position 1 corresponding to the first residue of the mature protein (Cys37 in [18]). The passenger domain of SphB1 is predicted to consist of two distinct structural units, i.e. the subtilisin domain that begins after the N-proximal Pro-rich region and is composed of both α and β structure (approximately residues 129 to 404), and the X domain that is predicted to form a β-helix and extends approximately from residues 426 to 620. The secondary structure of the region between residues 620 and 650 is not well predicted. That region is followed by an α- rich ‘linker’ region. The N-terminal sequence of the translocator domain found in the outer membrane of B. pertussis following the autoproteolytic maturation of SphB1 in vivo was reported to be S⁶⁸⁰RKVL [25]. Thus, the translocator includes the long predicted α-helix (S⁶⁸⁰ to H⁷¹⁷) most likely corresponding to that located inside the β-barrel in the NalP structure [10]. The β-barrel domain of SphB1 is predicted to span the region from A⁷²⁷ to Y¹⁰⁰³(Figure 1).

Figure 1.  Schematic representation of the proteins used in this work. The precursor of SphB1 is shown on the upper line, with SP, S, X and β representing the signal peptide, subtilase domain, ‘X’ domain and predicted β-barrel domain, respectively (approximately drawn to scale). The white boxes represent linker regions between the various domains, without well-predicted secondary structures. The hatched box represents a long predicted α-helix most likely corresponding to that of NalP. The arrowhead indicates the natural maturation site, and the sequence below corresponds to the N-terminus of the translocator in the outer membrane. The truncated versions of SphB1 were produced with an E. coli signal peptide (not represented) followed by an affinity Strep-tag (black box). The sequences underneath the various derivatives indicate the first residues of the SphB1 portion in each construct. The vertical arrows indicate the identified trypsin cleavage site in SphB1-X-α-β and SphB1-α-β. Its N-terminal sequence corresponds to residues 3–9 of the N-terminus of the translocator domain naturally found in the outer membrane of B. pertussis and arising from autoproteolytic cleavage of the protein.

Several truncated derivatives of SphB1 that include the predicted β-barrel domain, plus increasing portions of the preceding domains, were constructed. All were produced with a signal peptide in E. coli and prepared from membrane extracts in the presence of a non-denaturing detergent, dodecylmaltoside (DDM). We initially attempted to obtain the proSphB1 protein as a model for the complete autotransporter by constructing a proteolytically inactive version of SphB1 to avoid the proteolytic maturation of the proprotein between the passenger and translocator domains [18]. For unknown reasons, however, E. coli produced extremely low amounts of the recombinant protein, which could not be purified by affinity, suggesting that the N-terminal tag was not accessible. For those reasons, that recombinant protein was not included in our analyses.

Therefore, we produced a truncated form of SphB1 containing a shorter portion of the natural passenger domain. Based on the prediction of two domains for the passenger of SphB1, we engineered a recombinant form of SphB1 comprising the X domain, the ensuing region constituted by the long predicted α-helix and the predicted β-barrel domain. The resulting protein was called SphB1-X-α-β (Figure 1). That recombinant protein was produced and purified in good quantities. It was thus used as a model autotransporter protein, including the C-terminal domain of the natural passenger and the translocator domain.

Another form of SphB1 was constructed that comprises the α-helical-rich region and the β-barrel domain, called SphB1-α-β (Figure 1). That recombinant protein was produced and purified in large amounts. It was thus used as a model for the complete translocator domain.

We also constructed a recombinant protein comprising only the β-barrel without the predicted α-helix, which was called SphB1-β. Very small amounts of that recombinant protein could be purified, indicating that it may be rather unstable. Of note, production of the incomplete translocator domain of NalP devoid of its α-helical linker was also reported to be problematic [9]. SphB1-β served as a model for an incomplete translocator domain, i.e., one that includes only the β-barrel without the linker region. Although the very low quantities of SphB1-β precluded analyses of its secondary structure content and self-association, they were sufficient for the study of its channel properties (see below).

Characterization of the SphB1-derived recombinant proteins

The recombinant proteins were produced and purified to homogeneity (Figure 2A). Then, Edman degradation analyses were performed to determine their N-terminal sequences. This revealed that the signal peptide had been cleaved properly, generating the mature forms of SphB1-X-α-β and SphB1-α-β (not shown). Because they were purified from membranes using a non-denaturing detergent rather than refolded from inclusion bodies, the proteins were expected to be in native conformation, with a properly folded β-barrel. To address this issue, we took advantage of a property common to outer membrane proteins, their heat-modifiable migration on SDS-PAGE [26], [27]. Non-heated SphB1-X-α-β and SphB1-α-β subjected to electrophoresis appeared to migrate faster than predicted by their calculated molecular masses (Figure 2A, lanes 2 and 4). When they were denatured by heating at 95°C for 10 min in the presence of SDS (1% final concentration in the Laemmli buffer), SphB1-X-α-β and SphB1-α-β migrated as expected as approximately 40-kDa and 60-kDa proteins, respectively (Figure 2A, lanes 1 and 3). These differential migrations indicated the presence of a β–barrel in both proteins, suggesting that they were folded properly.

Figure 2.  Folding state of purified SphB1-X-α-β and SphB1-α-β proteins. (A) Heat modifiability and trypsic digestion of the purified proteins. Lanes 1 and 3: purified SphB1-X-α-β and SphB1-α-β, respectively. The samples were heated for 10 min at 95°C. Lanes 2 and 4: SphB1-X-α-β and SphB1-α-β, non-heated samples. The proteins were subjected to SDS-PAGE, and the gels were coloured with Coomassie Blue. (B) CD spectra of SphB1 derivatives. Solid line, SphB1-α-β; dashed line, SphB1-X-α-β.

This was further investigated by circular dichroism (CD) experiments carried out to examine the secondary structure of the purified recombinant proteins. Both CD spectra showed a maximum ellipticity between 215 and 220 nm, characteristic of proteins having a substantial amount of β-sheet structure. Thus these measurements confirm that the proteins were folded properly, consistent with their heat-modifiable migration in SDS-PAGE.

The two purified recombinant proteins were then subjected to a complete tryptic digestion. This enzymatic cleavage yielded the same proteolytic fragment of apparent molecular masses of 30 kDa for both proteins (results not shown). N-terminal sequencing of this degradation fragment yielded two sequences that were identical for both SphB1-X-α-β and SphB1-α-β: a major K⁶⁸²VLQDNL sequence and a minor V⁶⁸³LQDNLY sequence (vertical arrow in Figure 1). These sequences start at the second third and fourth residues, respectively, of the S⁶⁸⁰RKVL sequence corresponding to the N-terminus of the translocator domain of SphB1 found in the outer membrane of B. pertussis following its automaturation in vivo [25]. Importantly, the N-terminus of this trypsin-resistant fragment is also that of the long predicted α-helix preceding the β-barrel domain. In the crystal structure of NalP_ß, the corresponding α-helix is positioned in the β-barrel, with its N-terminus exposed to the extracellular side of the membrane [10]. Resistance to proteolytic degradation of the portion K⁶⁸⁰-Y¹⁰³³ of SphB1 in both recombinant proteins is a strong indication that the β-barrel is folded with the α-helix inside, as in the NalP structure.

Analysis of the self-associating properties of the recombinant proteins

In a first approach, dynamic light scattering (DLS) was used to explore the self-associating properties of SphB1 solubilized in DDM. DLS provides a measurement of the average translational diffusion coefficient of the protein-micellar particles in solution, from which their average hydrodynamic diameter can be computed assuming that they are spherical: the larger the size of the protein complexes, the larger the size of the corresponding protein-micellar particles. We systematically investigated the average size of the mixed micelles of SphB1-α-β/DDM and SphB1-X-α-β/DDM, with or without SDS (Figure 3 A, B). A significant difference between the two recombinant proteins was observed in the absence of SDS, with higher average sizes of the protein-micellar particles for SphB1-X-α-β/DDM than for SphB1-α-β/DDM (Figure 3C).

Figure 3.  Dynamic light scattering analysis of the SphB1 recombinant proteins. (A and B) Volume size distribution of SphB1-α-β/DDM (solid line) and SphB1-X-α-β/DDM (dashed line) micelles without (A) and with 1% SDS (B). (C) Variation of the average diameter of SphB1-α-β/DDM (open symbols) and SphB1-X-α-β/DDM (filled symbols) micelles as a function of the SDS concentration. The lines are drawn as guides for the eye. The concentration of the protein is 1.2 mg/mL and 1.4 mg/mL for SphB1-α-β and SphB1-X-α-β, respectively.

The organization state of the two proteins without SDS remained similar over a range of concentrations, indicating that their self-association was largely concentration-independent (Supplementary Figure S1 – online version only). Measurements of the average diameter of the protein-micellar particles as a function of SDS concentration (Figure 3C) revealed a decrease of the size of the particles with increasing SDS concentrations for both recombinant proteins, with the particle sizes remaining at 6 nm at SDS concentration superior to 0.5%. This value is only slightly larger than the diameter of the DDM/SDS mixed micelles, which is 4 nm in 1% SDS (data not shown). These results indicate that the two recombinant proteins have a tendency to associate in solution, and the addition of SDS causes the dissociation of these complexes. Although the exact size distribution of the samples cannot be determined by this technique, the X domain clearly promotes monomer-monomer interactions, since SphB1-X-α-β forms larger complexes. Therefore, the structural organization of the autotransporter depends both on the presence of the X domain and on the conditions of solubilization.

To further approach the size distribution of these proteins complexes, cross-linking experiments were performed. Results displayed in Figure 4 (lane 2) showed that SphB1-α-β was mainly present as a monomer. Similar cross-linking experiments performed on the outer membrane β-domain of the temperature-sensitive hemagglutinin (Tsh) from E. coli [24] also supported a monomeric organization for the translocator domain. However, for the SphB1 translocator domain, a weaker protein band of ≈ 80 kDa probably corresponding to the formation of a small proportion of dimer in solution was also present (Figure 4, lane 2), which did not appear when 0.1% or 1% SDS was added to the sample before the cross-linking reaction (Figure 4, lanes 3 and 4). Cross-linked SphB1-X-α-β appeared to partially form heterogeneous and larger protein complexes (Figure 4, lane 6) that were progressively disrupted with increasing SDS concentrations and disappeared at 1% SDS (Figure 4, lanes 7 and 8).

Figure 4.  Cross-linking of purified SphB1-X-α-β and SphB1-α-β. The proteins were incubated for 1 h with the DSS cross-linker at 2 mM. Lanes 1–4: SphB1-α-β. Lanes 5–8: SphB1-X-α-β. ‘+’ and ‘−’ indicate the addition of cross-linker to the proteins and no addition, respectively. Lanes 2 and 6: the cross-linking reaction was performed in the absence of SDS, while lanes 3 and 7 and lanes 4 and 8 correspond to cross-linking experiments performed in the presence of 0.1% and 1% of SDS, respectively. ‘*’ labelled the presumed dimers. The 12.5% SDS-PAGE gel was stained with Coomassie blue.

Collectively, DLS and cross-linking data indicate that both SphB1 derivatives may self-associate in vitro. This is not unprecedented, since another AT protein, AIDA-1, was shown previously to also form a certain proportion of dimers suggested to contribute to the stability of the translocator domain and thus to facilitate passenger translocation [28]. The dissociation of these SphB1 protein complexes by low concentrations of SDS indicates that the interactions are rather labile, suggesting that complexes are in equilibrium with monomers and likely form in a transient manner. Another finding of this work is that the presence of the X domain markedly promotes interactions between monomers.

Channel activity of the SphB1 recombinant proteins

We next investigated the influence of the passenger domain of SphB1 on the channel-forming properties of the translocator when reconstituted in a lipid bilayer, experimental conditions that should mimic more closely the original membrane environment of the AT.

The channel activity of the SphB1 derivatives in planar lipid bilayers was analyzed by performing a large number of electrophysiology recordings, to determine the conductance of the channel and its sensitivity to the applied potential.

In Figure 5 (panels AI and AII), current-voltage recordings of multiple channels showed that the conductance of the pores of SphB1-X-α-β and SphB1-α-β was constant over the entire voltage range tested, i.e., from 0 to±200 mV, and was identical at positive and negative potentials. These characteristics indicate no voltage dependence of the channels. In contrast, the shorter recombinant protein SphB1-β displayed a non-linear current-voltage curve, indicating a voltage-dependence of the conductance (Figure 5, panel AIII). Moreover, the latter curve was not as smooth as the two previous ones but displayed noisy currents, suggesting that the SphB1-β protein forms less stable channels than those of SphB1-X-α-β and SphB1-α-β.

Figure 5.  Channel activity of the SphB1 recombinant proteins. (A) Current /voltage recordings, with triangular ramps applied at a rate of 10 mV/sec to azolectin planar lipid bilayers containing around 0.5 µg of purified derivatives. (I) SphB1-X-α-β; (II) SphB1-α-β; (III) SphB1-β. (B) Current traces from single channels and associated amplitude histograms. (I) SphB1-X-α-β, applied voltage: 160 mV; (II) SphB1-α-β, applied voltage: 60 mV; (III) SphB1-β, applied voltage: 20 mV. The buffer used was 1 M KCl, 10 mM Hepes, pH 7.4.

Figure S1.  Variation of the average diameter of SphB1-α-β/DDM (open symbols) and SphB1-X-α-β/DDM (filled symbols) micelles as a function of the protein concentration determined by DLS technique. These measurements were performed without SDS in the medium.

The three SphB1 recombinant proteins were further characterized by single channel recordings. To this end, very small amounts of each protein were introduced into the cis-compartment bathing the lipid bilayer. In the presence of SphB1-X-α-β, homogeneous current fluctuations were readily observed 10 minutes after the incorporation. The single transitions recorded gave a conductance value of 155±10 pS in 1 M KCl, as illustrated in Figure 5 (panel BI). This low conductance and homogeneous conductance correspond to a small channel, most likely formed by an individual translocator domain. This is in agreement with recent molecular dynamic simulations showing that the translocator domain of NalP embedded into a DMPC bilayer is stable as a monomer [29].

The incorporation of SphB1-α-β into bilayers in the same conditions produced a much more heterogeneous distribution of the channel conductance size compared with SphB1-X-α-β. Besides a major 150 pS conductance value (frequency of events: 34%) most likely corresponding to that of SphB1-X-α-β, single channel recordings revealed the presence of various other ion channel conductance steps. These various conductance levels ranged from 65 pS (7% of events) to 2400 pS (3% of events) in 1 M KCl, including values of 245 (15% of events), 330 (12% of events), 450 (7% of events), 540 (8% of events), 780 (9% of events) and 1650 pS (5% of events). SphB1-α-β channels with a conductance of 780 pS are displayed in Figure 5 (panel BII). To determine if these heterogeneous conductance values observed with SphB1-α-β might be due to channel formation by various oligomeric forms of the translocator domain, SphB1-α-β channel activity was examined following SDS treatment to dissociate potential self-association (see above). The channel activity of SDS-treated SphB1-α-β was not different from that of the non–treated protein, with similarly heterogeneous conductance values (data not shown). This demonstrates that the conductance heterogeneity of the untreated SphB1-α-β protein was not caused by self-association but most likely by distinct conformations of the same monomeric channel.

This absence of the SphB1 self-association may result from the extremely low protein concentrations (10 ng/ml) used in the single-channel measurements together with the relative lability of the SphB1 assemblies as detected by DLS or crosslinking. In such low concentration conditions that are far from reflecting the molecular crowding of the outer membrane in vivo, only highly stable complexes would be detected.

Interestingly, channels with heterogeneous conductance values have also been described for the NalP translocator domain [10]. The authors have suggested that the conductance step of 150 pS observed for NalP results from ion fluxes through the hydrophilic pore occupied by the α-helix, whereas larger conductance steps could be due to a displacement of the α-helix from the pore. Similarly, the heterogeneity of the conductance steps of the SphB1-α-β channels could be due to movements of the predicted α-helix inside and/or from the pore. The complete displacement of the helix from the pore is probably induced by the applied voltage and might not occur under physiological conditions.

In contrast to SphB1-α-β, the channel activity of SphB1-X-α-β was very homogeneous. Thus, the C-terminal portion of the passenger domain present in SphB1-X-α-β modulates the channel activity of SphB1. It is likely that it does so by stabilizing and limiting the movements of the predicted α-helix within the channel.

At last, we looked at the single channel activity of the β-barrel domain of SphB1 alone by reconstituting SphB1-β into artificial membranes. Incorporation of this derivative into bilayers was quite tedious. The recording displayed in Figure 5 (panel BIII) illustrates the large and noisy current fluctuations obtained for this SphB1 derivative. The noisy behavior observed argues that this recombinant protein forms quite unstable channels in lipid bilayers. Of note, SphB1-β was extremely difficult to extract from membranes and to purify (see above), indicating that this derivative may be structurally perturbed. In agreement with these results, Konieczny et al. [30] have suggested that the alpha helical linker of the translocator domain of the AIDA-I AT may have a role in stabilizing the β domain in the outer membrane. Moreover Khalid et al. showed by molecular dynamic simulations on the translocator domain of NalP that removal of the internal α-helix induces a degree of flexibility in the β-barrel pore [29]. From the current amplitude histogram (Figure 5 panel BIII), a mean conductance value of ≈2800 pS in 1 M KCl was estimated for the SphB1-β channels. This large value is close to the upper conductances displayed by SphB1-α-β (2400 pS). This value might correspond to a largely open hydrophilic channel, most likely due to the absence of the α-helix in the pore. Khalid et al. [29] showed that removal of α-helix from the NalP transmembrane domain allows the pore to fill with water. This last observation is in good agreement with the increase of the ion flux (i.e., a larger conductance value) we measured for the SphB1-β derivative. Similarly Oomen et al. [10] found that the removal of the α-helix of NalP translocator domain increased the pore activity as assessed with an in vivo antibiotic sensitivity assay.

In summary, the analysis of the channel-forming properties of the SphB1 derivatives support the conclusion that the conducting unit represented by the major 150–155 pS conductance level obtained with both SphB1-X-α-β and SphB1-α-β, most likely corresponds to the translocation unit, i.e., a monomeric pore formed by the β-barrel itself. The presence of the α helix in the pore limits the conductance. When it moves out, the channel conductance increases markedly and becomes much more variable, most likely because the channel is destabilized and its inner wall becomes much more flexible. The presence of the X domain stabilizes the helix within the pore, hence the channel activity becomes low and very stable.

In conclusion, we have shown in this work that the passenger domain significantly influences the in vitro properties of the AT protein SphB1. First, the X domain of the passenger appears to promote the self-association of SphB1. Second, it modifies the conductance properties of the translocator domain. We proposed that both properties are likely to be relevant for the in vivo maturation and stability of AT proteins.

The organization of the SphB1 AT as observed in solution provides information on the ability of the various domains of the protein to promote interactions between monomers. The passenger domain of SphB1 carries a subtilisin activity, which mediates the maturation of both the FHA adhesin and the SphB1 precursor itself at the bacterial surface [18]. Because the automaturation of the SphB1 precursor occurs in trans [20], at least transient contacts between SphB1 monomers must take place during or after secretion. Our data indicate that the X domain of the passenger promotes such interactions, as well as possibly the translocator domain, albeit to a lesser extent. The self-association of SphB1 that we observed in vitro may thus occur in vivo in the course of its automaturation.

We have also shown that the passenger domain of SphB1 maintains the channel in a homogeneous, low-conductance conformation. It is likely that the folding of the passenger domain at the surface constrains the movements of the α-helical linker in the pore. This effect might contribute to limiting the permeability of the outer membrane barrier after the insertion of the translocator and translocation of the passenger domain. This effect could only apply transiently in the case of SphB1, because its passenger and translocator domains are separated by proteolytic maturation at the cell surface. However, it may occur for the numerous ATs that are not processed after secretion, and for those such as IgA protease and AIDA, where the natural maturation leaves a significant portion of the passenger domain attached to the β-domain.

Acknowledgements

We thank C. Locht for his interest in this work and J. L. Rigaud for helpful discussions. A. C. Meli is the recipient of a predoctoral fellowship from the Ministère de l'Education Nationale, de la Recherche et de la Technologie. F. Jacob-Dubuisson is a researcher of the CNRS. This work was supported by the Ministère de l'Education Nationale, de la Recherche et de la Technologie (Grant ACI BCMS2004 to FJD).