We discovered a novel secreted protein by Pseudomonas aeruginosa, PlpD, as a member of the bacterial lipolytic enzyme family of patatin-like proteins (PLPs). PlpD is synthesized as a single molecule consisting of a secreted domain fused to a transporter domain. The N-terminus of PlpD includes a classical signal peptide followed by the four PLP conserved blocks that account for its lipase activity. The C-terminus consists of a POTRA (polypeptide transport-associated) motif preceding a putative 16-stranded β-barrel similar to those of TpsB transporters of Type Vb secretion system. We showed that the C-terminus remains inserted into the outer membrane while the patatin moiety is secreted. The association between a TpsB component and a passenger protein is a unique hybrid organization that we propose to classify as Type Vd. More than 200 PlpD orthologues exist among pathogenic and environmental bacteria, which suggests that bacteria secrete numerous PLPs using this newly defined mechanism.
Proteins secreted by Gram-negative bacteria are transported across the cell envelope, which is composed of the inner membrane, the peptidoglycan-containing periplasm and the outer membrane. To date, six different secretion pathways have been identified (Economou et al., 2006; Filloux et al., 2008; Michel and Voulhoux, 2009). The Type V secretion system (T5SS) comprises diverse branches, the autotransporter (AT or Type Va and Type Vc) pathway and the two-partner secretion (TPS or Type Vb) pathway. The AT family of secreted proteins are supposedly able to translocate by themselves across the outer membrane (hence their name) (Henderson et al., 2004; Linke et al., 2006; Dautin and Bernstein, 2007). ATs are modular proteins consisting of an N-terminal signal peptide, a C-terminal β-barrel (the β-domain) and, in between, a secreted passenger domain. Depending on the AT, the passenger domain may either remain attached to the transporter domain protruding from the bacterial surface, or be cleaved from the β-domain and released into the extracellular medium. Processing by both exogenous and endogenous proteolytic activities has been reported (Hendrixson et al., 1997; Shere et al., 1997; Coutte et al., 2003; van Ulsen et al., 2003). In contrast to ATs, TPS systems consist of two separate proteins, a TpsB transporter involved in the secretion of the TpsA protein across the outer membrane (Hodak and Jacob-Dubuisson, 2007; Mazar and Cotter, 2007). Like the AT passenger domains, TpsAs are large proteins that remain attached at the cell surface or are released into the extracellular medium. Although the AT pathway is regarded as a variant of the TPS pathway, in which the transported protein and the translocator are connected in a single polypeptide, it should be pointed out that there is no sequence or structural similarities between the β-domains of ATs on the one hand and the TpsB transporters on the other hand. Indeed, while the β-domains of ATs present a 12-stranded β-barrel (Oomen et al., 2004), the TpsBs have a β-barrel with 16 β-strands (Clantin et al., 2007; Hodak and Jacob-Dubuisson, 2007). Moreover, the N-terminal portion of TpsBsextends in the periplasm and contains two POTRA (polypeptide transport-associated) domains, one of which is proposed to be involved in the recruitment of the TpsA partner via its N-proximal secretion signal, the TPS domain (Clantin et al., 2004; Hodak and Jacob-Dubuisson, 2007). POTRA domains are also found among other members of the Omp85 protein superfamily to which the TpsB proteins belong, but never in ATs. The Omp85 superfamily includes the Omp85/BamA proteins required for the insertion of β-barrel membrane proteins into the outer membrane of bacteria (Voulhoux et al., 2003; Bos et al., 2007). The periplasmic POTRA domains may have chaperone-like qualities, and together with the β-barrel constitute the functional core of BamA (Bos et al., 2007). Interestingly, the BamA protein has been proposed to insert the β-barrel of the AT into the outer membrane and to transport the passenger domain to the cell surface (Gentle et al., 2005; Hodak and Jacob-Dubuisson, 2007; Jain and Goldberg, 2007), suggesting that the TPS and AT pathways have analogous transporters.
A new family of bacterial lipolytic enzymes, called patatin-like proteins (PLPs), has recently been proposed (Banerji and Flieger, 2004). Patatins represent 40% of total soluble potato tuber proteins and, while they are considered as storage proteins, they show lipid acyl hydrolase activity (Shewry, 2003). Such activity may be considered as a possible defence mechanism against plant pathogens and stress (Dhondt et al., 2000). Potato patatin B2 and human cytosolic phospholipase A2 (PLA2) share conserved domains (Hirschberg et al., 2001), including a Serine–Aspartate active-site dyad instead of the more common Serine–Histidine–Aspartate (or Glutamate) triad of lipolytic enzymes (Rydel et al., 2003). In bacteria, genes encoding patatin homologues are highly represented in some animal pathogen and plant pathogen/symbiont genomes, suggesting that they are important during interaction with host (Banerji and Flieger, 2004). The first characterized bacterial PLP was the ExoU protein from the opportunistic pathogen Pseudomonas aeruginosa. ExoU is a substrate for the Type III secretion system (T3SS) of strains PA103 and PA14 and is directly delivered into the cytosol of the eukaryotic cell during the infection process. ExoU is the major cytotoxin secreted by P. aeruginosa. It produces rapid cell death (Finck-Barbancon et al., 1997; Hauser et al., 1998; Sato et al., 2003) and is an important virulence determinant in most animal models of infection (Finck-Barbancon et al., 1997; Hauser et al., 1998; Shaver and Hauser, 2004). ExoU cleaves phosphatidylcholine and phosphatidylethanolamine in vitro (Sato et al., 2005) and requires eukaryotic superoxide dismutase as a cofactor for its phospholipase A2 activity (Sato et al., 2006). ExoU is trafficked to the plasma membrane of the host cell where it undergoes ubiquitinylation, which allows efficient utilization of adjacent substrate phospholipids (Stirling et al., 2006). In Legionella pneumophila, four ExoU homologues were found (Vanrheenen et al., 2006) and they were shown to be injected into target cells via a Type IV secretion system (T4SS). In the social bacteria Myxococcus xanthus, the MXAN_3852 protein contains motifs characteristic of patatins. Furthermore, the purified recombinant protein hydrolysed esters of short-chain fatty acids (Moraleda-Munoz and Shimkets, 2007).
It is well known that the P. aeruginosa strain PAO1, as well as many other P. aeruginosa strains, does not contain the exoU gene but instead carries an exoS gene encoding another T3SS effector (Wolfgang et al., 2003). The basis for the incompatibility between exoU and exoS within the same P. aeruginosa genome is not known, but the carriage of exoS appears more prevalent (Kulasekara et al., 2006).
In this study we identified an ExoU homologue in the PAO1 and PA14 strains of P. aeruginosa. This homologue, named PlpD, has a lipase activity and could be secreted into the extracellular medium. Interestingly it has an N-terminal signal peptide and is likely, in contrast to ExoU, T3SS-independent. Instead, we showed that PlpD secretion involves a novel secretion pathway, which is conceptually close to the Type Vb secretion system. Yet, we revealed drastic structural differences that lead us to propose a novel secretion pathway, which we named Type Vd.
PA3339 encodes a patatin-like protein
We used the ExoU amino acid (aa) sequence to perform a blast search on the P. aeruginosa PAO1 proteome and to identify putative PLPs. The results obtained revealed that the PA3339 protein (GI:15598535) shares 40% of similarity and 25% of identity with ExoU within its N-terminal domain (between residues 23 and 216). PA3339 is a protein of 728 residues, with a predicted molecular weight of 80 899 (Fig. 1A). A Pfam domain search identified a patatin-like domain (PF01734) between residues 27 and 220, within the N-terminal part of PA3339. Usually, bacterial PLPs possess the three conserved domains (blocks I, II and IV) of potato patatin B2 (Banerji and Flieger, 2004) and an additional motif, named block III. Each of these conserved domains is present in the protein coded by pa3339 (Fig. 1A). In Fig. 1B, we show the alignment between the PA3339 blocks with those of three other bacterial PLPs, namely ExoU from P. aeruginosa PA103, VipD from L. pneumophila and MXAN_3852 from M. xanthus. Block I consists of a Glycine-rich region containing a conserved basic residue, Arginine or Lysine (Arg35 in PA3339), which probably serves as an oxyanion hole. Block II is located 10–20 aa downstream block I and comprises a typical lipase motif Gly-X–Ser-X–Gly with the putative active-site Serine (Gly58-X–Ser60-X–Gly62) (Arpigny and Jaeger, 1999). Block III contains a conserved Serine (Ser188), which may be an important structural element. Block IV contains the active-site Aspartate (Asp207) residue that, together with the active-site Serine, forms the catalytic Serine–Aspartate dyad. Finally, the highly conserved Proline residues in blocks III and IV (Pro192 and Pro215) may be important for the proper conformation of the protein. Based on these observations, we concluded that PA3339 belongs to the PLP family and we designated this protein PlpD for Patatin-like protein D.
PlpD harbours a C-terminal TpsB-like domain
Interestingly, in contrast to ExoU, PlpD is predicted to contain a type I N-terminal signal peptide when using the PSORTB (Gardy et al., 2003) or SignalP softwares (Nielsen et al., 1997; Bendtsen et al., 2004) (Fig. 1A). Moreover, two Pfam domains could be identified within the C-terminal domain of PlpD. First a POTRA motif from the surface Ag VNR family (PF07244) is located between residues 333 and 404 (Fig. 1A). Second, a bacterial surface Ag domain (PF01103) is located immediately downstream of the POTRA domain, between residue 404 and the C-terminal residue of PlpD. Both domains are typically found in BamA proteins as well as in TpsB proteins (Gentle et al., 2005). Secondary structure predictions of the PlpD POTRA domain indicate three β-strands with a pair of α-helices (order is β−α−α−β−β), in agreement with the POTRA domain fold of BamA (formerly known as YaeT) from Escherichia coli (Kim et al., 2007) (Fig. 1C). The DYF aa sequence (highlighted in Fig. 1C), which is located within a loop between the second α-helix and the second β-strand of PlpD, aligns with the GYF aa signature of the POTRA sequence family. An in-depth analysis of the last 324 C-terminal residues of PlpD, comprising PF01103, using the bomp server (Berven et al., 2004) and a combination of secondary structure prediction tools (see Experimental procedures), predicted 16 putative amphipathic β-strands (shown in black in Fig. 1D) that may form a β-barrel. All together these data suggest that the N-terminal signal peptide of PlpD allows translocation across the inner membrane in a Sec-dependent manner, and PlpD is subsequently inserted in the outer membrane via its C-terminal β-barrel domain. Accordingly, as for most bacterial outer membrane proteins, the last C-terminal amino acid residue of PlpD is a Phenylalanine, which is preceded by a series of hydrophobic residues at positions 5, 7 and 9 from the C-terminus (Fig. 1D) (Struyve et al., 1991). It is worth noting that the PlpD β-barrel (PF01103) is unrelated to the aa sequence motifs found in PF03797 and PF03895, which typically constitute the β-domains of ATs.
All together these in silico data, i.e. (i) a large protein with a signal peptide, (ii) an N-terminal domain with a putative enzymatic activity, and (iii) a putative C-terminal β-barrel domain, suggest that PlpD belongs to the AT family (Type Va). However, since the β-barrel fused with a POTRA domain is similar to a TspB component (Type Vb) we concluded that PlpD might belong to the family of T5SS proteins but of a previously unidentified type.
PlpD structure prediction supports a fold with two independent modules
The three-dimensional structure of PlpD was predicted by using the Phyre server (http://www.sbg.bio.ic.ac.uk/phyre/). The results obtained are summarized in Fig. 1A and E and confirmed an organization of PlpD into two modules. The structure of the N-terminal domain of PlpD is predicted according the potato patatin Pat17 (Fig. S1 in Supporting information) (Rydel et al., 2003). The C-terminal moiety of PlpD was structured with the TpsB protein FhaC, which mediates the secretion of the Bordetella pertussis filamentous haemagglutinin (FHA) (Clantin et al., 2007) (Fig. 1E). This domain of PlpD is predicted to form a β-barrel, composed of 16 antiparallel β-strands (shown in yellow in Fig. 1E and in grey in Fig. 1D). The periplasmic side of the barrel consists of the globular POTRA domain, organized as a three-stranded β-sheet and two α-helices (shown in blue and green in Fig. 1E), confirming the secondary structure prediction (Fig. 1C). In conclusion, we predicted that the C-terminus domain of PlpD is likely a periplasmic POTRA domain and an outer membrane β-barrel. The β-barrel is likely to form a channel in the outer membrane through which the patatin passenger domain can be translocated at the cell surface.
The patatin domain of PlpD is secreted by P. aeruginosa
In order to study the fate of PlpD in P. aeruginosa, we introduced a V5/His6 tag at the C-terminal end of PlpD. The recombinant gene was cloned in a broad-host-range vector (leading to pRS9) under the control of a tac promoter. The pRS9 plasmid was conjugated into the P. aeruginosa strain PAO1, and expression of plpD was induced with IPTG (isopropyl-β-d-thiogalactopyranoside) for 1 or 3 h (Fig. 2A). Protein samples from whole cell extracts and from culture supernatants were prepared as usual (Bleves et al., 1999) and subjected to sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and immunoblotting using anti-V5 tag antibodies. A protein with an apparent molecular weight of 49 kDa was specifically recognized in the whole cell fraction (Fig. 2A), while the anti-V5 tag antibodies did not detect any product from the culture supernatants (data not shown). From these results, we concluded that PlpDV5/His6 is cleaved in P. aeruginosa, with the C-terminal V5/His6 tagged domain remaining associated with the bacterial cell, presumably with the outer membrane, while the untagged N-terminal domain could not be detected.
To analyse the fate of both PlpD domains, the recombinant PlpD protein produced in E. coli was purified and used to raise a polyclonal antiserum (see Experimental procedures). Pseudomonas aeruginosa PAO1 was grown in LB at 37°C for 1.5–3.5 h and protein samples prepared as described above. A protein of ∼47 kDa, similar to the one previously observed, was readily detected in the cell, indicating a significant level of expression from the chromosomal copy of the plpD gene (Fig. 2B). From the PlpD aa sequence, the calculated size of the β-domain including the POTRA motif is 44.97 kDa, which is in agreement with the product detected in cells. Interestingly, another protein, with a slightly smaller molecular weight, was detected in the culture supernatants (Fig. 2B). This secretion is specific and not simply due to cell lysis since the periplamsic DsbA proteins was not found in the same supernatant samples (data not shown). Although the size of the PLP domain after cleavage of the signal peptide is predicted to be 33.10 kDa, which is considerably smaller than the product detected in culture supernatant, this observation suggests that this protein is likely corresponding to the secreted passenger domain of PlpD. This size difference could be due to an aberrant migration or to a post-translational modification of the patatin domain. It is worth noting that neither the chromosomal PlpD nor the recombinant PlpDV5/His6 is detected as mature full-length protein in P. aeruginosa and this even in early time points of growth (Fig. 2B).
The C-terminal domain of PlpD is folded in the outer membrane
We further analysed the intracellular localization of the putative C-terminal β-domain of PlpD. Pseudomonas aeruginosa PAO1 was grown in LB at 37°C up to stationary phase and bacteria were collected by centrifugation. Bacterial cells were then lysed by sonication (see Experimental procedures), and the cell envelope fraction was collected by ultracentrifugation. The inner membrane proteins were solubilized in sodium lauryl sarcosinate (SLS), whereas the outer membrane proteins are recovered in the insoluble fraction. The inner membrane fraction contained the XcpY protein but was totally free of PlpD. In contrast, the C-terminal domain of PlpD was mostly detected in the same fraction as OprF, a P. aeruginosa porin indicating that it localized into the outer membrane (Fig. 3A, lane 3 and 4). Indeed the PlpD C-terminus should essentially be associated with the outer membrane fraction since that is where the cleavage should occur as it is described for the ATs.
Many outer membrane proteins displayed heat modifiability (Dekker et al., 1995; Konieczny et al., 2001; Oomen et al., 2004) that is the correctly folded monomer had a faster electrophoretic mobility in semi-native PAGE than the heat-denaturated form. To assess if the C-terminal domain of PlpD attains a proper β-barrel conformation in the outer membrane, its heat modifiability was tested on semi-native gels as described in Experimental procedures. As shown in Fig. 3B (left panel), PlpD C-terminus is immunodetected as a folded monomeric form that migrates faster in semi-native PAGE than the heat-denatured form. In those semi-native conditions, the P. aeruginosa porin trimers are stable and observable at the top of the gel, while the heat-denatured porins are dissociated into monomers (Fig. 3B, right panel). The C-terminal domain of PlpD thus displays the heat modifiability of many outer membrane proteins that tends to indicate a proper folding of this transporter domain as a β-barrel in the outer membrane.
PlpD possesses lipase activity
To gain insight into the function of PlpD, we assessed the activity of PlpD produced in E. coli since P. aeruginosa itself secretes several lipolytic enzymes. The plpD gene was cloned in the expression vector pET22b(+), yielding plasmid pET22plpD. The plasmid contains the plpD gene under the control of an IPTG-inducible T7 promoter. The recombinant plasmid carrying plpD was introduced in E. coli BL21(DE3), plpD gene expression was induced with IPTG, cells were collected by centrifugation, the total cellular proteins were separated on SDS-PAGE and stained with Coomassie blue. A protein with an apparent size of 90 kDa is specifically abundant in the BL21(DE3) (pET22plpD) cell fraction upon IPTG induction, which likely represents the full-length form of PlpD (Fig. 4A, lane 4). This was confirmed by Western blotting using anti-PlpD antibodies (data not shown and Figs S2 and S3 in Supporting information). In contrast with P. aeruginosa, PlpD is not cleaved in E. coli. The size of the PlpD protein produced in E. coli (∼90 kDa) is higher than the predicted molecular weight (80 899) but is in agreement with the sum of the apparent sizes of the intracellular PlpD C-terminus (∼47 kDa) and the secreted domain (∼45 kDa) observed in P. aeruginosa.
We further examined the activity of PlpD towards lipids. Whole cell extracts of E. coli producing full-length PlpD, or not, were mixed with p-nitrophenyl palmitate and the release of p-nitrophenol was measured (see Experimental procedures). Extracts from bacteria producing PlpD released significantly more p-nitrophenol than samples from the strains containing either the empty vector or pET22plpD for which expression was not induced. PlpD activity in the IPTG-induced E. coli strain was comparable to the P. aeruginosa LipA activity used as a positive control (Fig. 4B). We concluded that the full-length PlpD has a lipase activity in vitro, which further confirms that PlpD is a PLP.
Putative PlpD-like transporters are found among Gram-negative bacteria
So far bacterial PLPs were shown to be secreted via Type III (Finck-Barbancon et al., 1997; Hauser et al., 1998) and Type IV (Vanrheenen et al., 2006) secretion machinery. To our knowledge, P. aeruginosa PlpD is thus the first example of a PLP using a Type V mechanism to reach the extracellular medium. In order to establish whether this is a unique case or not, we searched for PlpD homologues with a similar domain architecture at NCBI (http://www.ncbi.nlm.nih.gov/) A total of 133 proteins harbouring the three domains (patatin-like, POTRA and 16-stranded β-barrel) were found. Moreover, 99 PlpD homologues containing the patatin domain and the β-domain but no obvious POTRA domain, which may be due to the limited aa sequence identity between POTRA domains, were also discovered. PlpD-like proteins are present among pathogens (Burkholderia species, Vibrio cholerae), symbiotic bacteria (Vibrio fischeri), commensal bacteria (Bacteroides ovatus) and environmental bacteria (Pseudomonas putida, Pseudomonas fluorescens and Shewanella oneidensis) (Fig. 5). The taxonomic distribution of PlpD-like proteins reveals that these proteins are not abundant among bacteria. They are mainly represented in four phyla (Proteobacteria, Bacteroidetes, Fusobacteria and Chlorobi), among which they are restricted to few sublineages. For instance, PlpD is only found in 43 genomes among 239 sequenced genomes of γ-Proteobacteria. This distribution suggests that these lineages have acquired these genes by horizontal gene transfers. The phylogenetic analysis confirms this hypothesis. PlpD acquisition could be due to inter- as well as to intra-phyla transfers. While Bacteroidetes and Chlorobi phyla belong to the same superphyla, their PlpD-like proteins are not regrouped that can be explained by independent acquisition during inter-phyla transfer. Intra-phyla transfers can be illustrated by the monophyletic group (posterior probability = 1.00) that contains Vibrionales (Vibrio sp.) and Alteromonadales (Shewanella sp.). Despites such kind of horizontal transfers, some groups corresponding to taxonomic units are present (i.e. β-proteobacteria phylogenetic group, γ-proteobacteria phylogenetic group). This observation suggests that when the plpD-like gene has been acquired in one particular organism of a group, it has then been transferred within this group. Moreover PlpD-like proteins are mainly found within bacteria leaving in animals (respiratory and gastrointestinal tracts), marine bacteria and bacteria isolated from polluted waters. The strength that can sustain such transfers is thus the same lifestyle or the same ecological niche.
ExoU is one of the most potent toxins produced by P. aeruginosa towards host cells. Its cytotoxic activity depends on an original catalytic dyad, present in vegetal patatins and in human cytosolic PLA2. In this study we discovered a new secreted protein in P. aeruginosa, which, because of its similarity with patatin, was designated PlpD for patatin-like protein D. PlpD is synthesized as a single molecule consisting of a secreted domain fused to a transporter domain. An in-depth in silico analysis predicted that the C-terminal domain of PlpD can form a β-barrel in the outer membrane to allow translocation of the patatin-like passenger domain across the outer membrane. This organization suggests a T5SS mechanism. In agreement with this model, we showed that the passenger patatin-like domain of PlpD was secreted into the extracellular medium by the PAO1 and PA14 strains, whereas the C-terminal domain remained inserted into the outer membrane in a folded form. With this result we definitively demonstrated that PlpD was a novel secreted protein from P. aeruginosa.
We took advantage of heterologous production of PlpD in E. coli as a full-length protein to monitor the enzymatic activity of PlpD. Besides their activity towards phospholipids, PLA2-like patatins usually display esterase and/or lipase activities. Escherichia coli extracts producing PlpD clearly resulted in the hydrolysis of a typical lipase substrate, the p-nitrophenyl palmitate. The full-length PlpD is thus an active lipase even when it remains covalently associated to the β-domain. The lack of processing in E.coli is maybe the consequence of the mislocalization of the protein that was either found in inclusion bodies or stucked in the inner membrane (Fig. S3 in Supporting information). One can also not exclude the possibility that additional factors are required for proper outer membrane targeting and are not present in the heterologous background. Contrary to ExoU (Sato et al., 2003; 2006), PlpD does not require an eukaryotic cofactor to be active. This suggests that the secreted PlpD can be active in the extracellular environment and not only in the context of the target host cell, which can make sense regarding the fact that some environmental bacteria possess PlpD-like proteins.
Interestingly, we found that the nature of the PlpD transporter is totally original. Whereas the organization of this large protein with an N-terminal signal peptide, a central passenger domain and a C-terminal β-domain could be reminiscent of an AT protein, the characteristics of the latter domain fit nicely with those of a TpsB transporter. These two pathways are two distinct branches of the T5SS, Type Va/Vc (AT) and Type Vb (TPS) respectively. ATs are secreted proteins supposedly able to translocate by themselves across the outer membrane (Henderson et al., 2004; Linke et al., 2006; Dautin and Bernstein, 2007). The TPS pathway involves two protagonists, the secreted TpsA protein, which crosses the outer membrane thanks to a TpsB transporter (Hodak and Jacob-Dubuisson, 2007; Mazar and Cotter, 2007). The β-domain of PlpD displayed features of TpsB channels with a POTRA motif preceding a putative 16-stranded β-barrel that can be folded accordingly with the structure of the FhaC/TpsB protein of B. pertussis (Clantin et al., 2007). The TpsB protein via its first POTRA domain recognizes the TPS secretion signal of its cognate TpsA, thus allowing the outer membrane translocation of TpsA through the TpsB β-barrel. A hallmark of TpsA proteins is the N-terminal TPS domain, with a highly conserved NPNGI motif (Mazar and Cotter, 2007). This domain is necessary and sufficient for secretion of several TpsA, like ShlA (Schonherr et al., 1993) or FHA (Jacob-Dubuisson et al., 1996) by their cognate TpsB transporters. We could not identify any TPS domain in the PlpD sequence (data not shown), but since the secreted domain is fused to the transporter in the PlpD protein, the recognition between the two partners is not a limiting step in this case and there might be no requirement for a secretion signal.
Maturation by both exogenous and endogenous proteolytic activities has been reported for AT and TPS systems. In order to understand the processing of PlpD we pursued two hypotheses (Fig. S4 in Supporting information). It is known that the outer membrane protease called IcsP is responsible for maturation and release of the Shigella flexneri AT IcsA (Shere et al., 1997). We thus considered that an outer membrane protease of P. aeruginosa could be responsible for PlpD cleavage. We tested the effect of the OprD protein (Yoshihara et al., 1996) and showed that PlpD was correctly processed in its absence. Moreover, as the maturation of an AT can depend on the protease activity of another AT, like for the IgA protease with the serine protease AT NalP in Neisseria meningitidis (van Ulsen et al., 2003) or the serine protease SphB1 for FHA (Coutte et al., 2003), we tested the effect of a mutation in sprS (PA3535), coding a putative Serine protease AT of P. aeruginosa (Ma et al., 2003; Hardie et al., 2009). PlpD was efficiently processed in this mutant. There might be other candidates to be tested and alternatively we could consider an autoprocessing mechanism since the catalytic dyad of the patatin active site is based on a Serine residue like in Serine proteases. Indeed, the passenger domain of the Haemophilus influenzae Hap is released by an endogenous Ser protease activity that mediates an intermolecular reaction (Hendrixson et al., 1997). As neither the intracellular C-terminal domain of PlpD, nor the N-terminal secreted part is detectable with Coomassie blue staining or silver staining (data not shown), we still do not know where the cleavage takes place. But even for well-characterized ATs, the exact cleavage site might be still not known. Indeed to crystallize the C-terminal domain of the NalpP autotransporter, Oomen and colleagues (2004) delimitated and produced the transporter domain on the basis of sequence alignment of 120 autotransporters.
What can be the function of the POTRA domain in PlpD? Considering the targeting role of the POTRA motif in the TpsB transporter, one possibility would be that the PlpD POTRA domain may recruit a partner, perhaps the protease responsible for its processing. Alternatively, the POTRA domain can function as an intramolecular chaperone that assists and limits the folding of PlpD in the periplasm for an efficient translocation across the outer membrane. Accordingly, while structural elements of AT passengers are formed in the periplasm prior to secretion (Skillman et al., 2005), folding in that compartment is limited for TpsA (Hodak and Jacob-Dubuisson, 2007). Indeed, the cytolysin ShlA is not only secreted but also activated through a conformational change by its cognate ShlB/TpsB transporter (Walker et al., 2004). This model has also been proposed for the P-usher, a hybrid transporter of P. aeruginosa, responsible for both fimbrial assembly and TpsA protein secretion. It consists of an outer membrane usher domain fused to a POTRA-like domain possibly involved in folding the secreted CupB5/TpsA protein (Ruer et al., 2008). Finally, a reason why the patatin domain is fused with a TpsB-like domain in PlpD may be that a folded patatin enzyme in the periplasm could be active towards P. aeruginosa phospholipid membranes. Indeed we showed that the full-length PlpD produced by E. coli could be active towards lipids. The fusion of the patatin domain to the translocator domain might thus enable a rapid export out of the periplasm into the extracellular milieu. In agreement with this, the other known bacterial PLPs are secreted by T3SS (Finck-Barbancon et al., 1997; Hauser et al., 1998) or T4SS (Vanrheenen et al., 2006), and in these cases the secreted proteins have no access to the periplasm, are unfolded during the transport and moreover are directly injected into host cells.
We found more than 200 PlpD orthologues in Gram-negative bacteria suggesting that numerous PLPs may be secreted by this original T5SS transporter that we propose to call a Type Vd system, to differentiate it from AT (Type Va and Vc) and TPS (Type Vb). The presence of PlpD-like transporters in pathogenic as well as in non-pathogenic bacteria suggests that PlpD-like proteins involvement is not limited to virulence, but that they may be implicated in functions such as host/symbiont–commensal communication, exchange and cell–cell communication. We have only found patatin domains fused to TpsB-like transporters, but it cannot be excluded that other types of passenger domain may be encountered. In future studies, we will investigate the role of the POTRA domain in this new kind of protein transporter.
Bacterial strains, plasmids and growth conditions
The bacterial strains and plasmids used in this study are described in Table 1. The E. coli DH5α and TGI strains were used for standard genetic manipulations. Recombinant plasmids were introduced in P. aeruginosa using the conjugative properties of pRK2013. Pseudomonas transconjugants were selected on Pseudomonas isolation agar (PIA, Difco Laboratories) supplemented with appropriate antibiotics. The following antibiotic concentrations were used: for E. coli, ampicillin (50 µg ml−1), kanamycin (25 µg ml−1); for P. aeruginosa, carbenicillin (500 µg ml−1). Escherichia coli and P. aeruginosa cultures were inoculated at an optical density at 600 nm (OD600) of 0.1 with overnight cultures, and strains were grown at 37°C with aeration in LB.
The PlpD protein sequence was used to search for proteins with a similar domain organization (PF01734, PF07244 and PF01103) into the NCBI database. A total of 133 proteins with the same organization, including PlpD, and 99 proteins with only two domains (PF01734, PF01103) were found. The PlpD-like homologues were aligned with clustalw (Thompson et al., 1994). The resulting alignment was manually refined with the program ed of the MUST package (Philippe, 1993). Regions where homology between sites was doubtful were removed and 209 amino acid positions were finally kept for the phylogenetic analyses. First, we performed a preliminary phylogenetic analysis of the 232 PlpD-like sequences by Maximum likelihood using the LG model with a Γ law (four rate categories and an estimated alpha parameter) implemented in phyml (Guindon and Gascuel, 2003). The resulting tree was used to choose a subset of 75 sequences representative of the diversity of the PlpD-like homologues as following: 23 from the γ-proteobacteria, 19 from the β-proteobacteria, 17 from the bacteroidetes, 8 from the fusobacteria, 4 from the chlorobi, 2 from the α-proteobacteria, 1 from the verrucomicrobia and 1 from the acidobacteria. Next, we performed a bayesian phylogenetic analysis of these 75 PlpD-like sequences using the program MrBAYES 3 with a mixed substitution model and a Γ law (six rate categories) and a proportion of invariant sites to take among-site rate variation into account (Ronquist and Huelsenbeck, 2003).
Preparation of antisera
The plpD gene from the Pseudomonas gene collection (Labaer et al., 2004) was cloned using the Gateway technology (Invitrogen) in the pET-DEST42 vector under a T7 promoter leading to pRS4 encoding PlpDV5/6His tagged at the C-terminus. The recombinant PlpD protein was produced in E. coli strain BL21(DE3) grown under agitation at 37°C, after induction of the T7 RNA-polymerase gene with 1 mM IPTG. Bacteria were harvested by centrifugation for 10 min at 5000 g, and sonicated (five pulses of 15 s). Proteins found in inclusion bodies were solubilized with 8 M urea for 2 h at room temperature. Purification was performed by affinity chromatography on HiTrap Chelating HP (GE Healthcare). Polyclonal antisera were raised in rabbits (Eurogentec). Two rabbits (NZW) were injected four times using 100 µg per injection during a course of 3 months.
SDS-PAGE, semi-native PAGE and immunoblotting
Bacterial cell pellets were resuspended in loading buffer (Laemmli, 1970). Exoproteins from culture supernatants were precipitated with 10% (w/v) trichloroacetic acid (TCA), washed with acetone, and resuspended in loading buffer. The protein samples were boiled and separated on SDS gels containing 10% (w/v) acrylamide. Protein samples analysed in semi-native conditions were resuspended in SDS-PAGE loading buffer containing only 0.2% SDS and no β-mercaptoethanol. These samples were not heated, but kept at 4°C before electrophoresis. Electrophoresis was performed at 100 V in polyacrylamide gels without SDS. For Western blotting, proteins were transferred from gels onto nitrocellulose membranes. After 30 min to overnight saturation in Tris-buffered saline (TBS) (0.1 M Tris, 0.1 M NaCl, pH 7.5), 0.05% (v/v) Tween 20 and 5% (w/v) skim milk, the membrane was incubated for 1 h with anti-V5 (diluted 1:5000), anti-PlpD (1:1000), anti-XcpY (1:5000) or anti-OprF (1:2500); washed three times with TBS, 0.05% Tween 20; incubated for 45 min with goat anti-rabbit immunoglobulin G (IgG) antibodies (Sigma) diluted 1:5000; washed three times with TBS, 0.05% Tween 20; and then revealed with a Super Signal Chemiluminescence system (Pierce).
P. aeruginosa fractionation
Bacteria were grown until the stationary phase (OD600 = 3.5–4), harvested with low-speed centrifugation, and pellets were washed and resuspended in 10 mM Tris-HCl (pH 7.2). The cells were then disrupted by sonication (four pulses of 20 s each) in the presence of 1 mM phenylmethylsulfonyl fluoride. Unbroken cells and cellular debris were removed by low-speed centrifugation. The soluble and membrane fractions were separated by ultracentrifugation for 40 min at 72 000 g (Beckman, TLA-55 rotor). Cytoplasmic and periplasmic proteins in the supernatant were precipitated with 10% (w/v) TCA. Inner and outer membrane proteins were separated by differential solubilization in SLS. Cell envelope fractions were incubated with 2% SLS for 15 min at 4°C with gentle shaking. SLS-insoluble outer membrane proteins were separated from soluble inner membrane proteins by ultracentrifugation (72 000 g, 40 min). Inner membrane proteins were then precipitated with 10% (w/v) TCA. Inner and outer membrane protein pellets were resuspended in sample buffer.
Determination of lipase activity
The plpD gene was amplified with chromosomal DNA of P. aeruginosa PAO1 as a template using Pfu DNA polymerase (Promega) with oligonucleotides plpD Up (5′-AAACATATGCGCCGCCTGCTG-3′) and plpD Down (5′-ATGAGCTCTCAGAAGTTCTGCCC-3′), creating NdeI and SacI sites at each end respectively. The resulting PCR product (2184 bp) was hydrolysed with NdeI and SacI and cloned into the NdeI and SacI sites of the pET22b(+) vector producing pET22plpD. Escherichia coli BL21(DE3) transformed with the pET22b(+) and pET22plpD plasmids were grown in LB medium with 0.4% (w/v) glucose at 37°C until OD600 = 0.8, and expression of plpD was then induced with IPTG at 0.4 mM final concentration for 2 h. The cells were collected by centrifugation for 2 min at 10 000 g and resuspended in 100 mM Tris-HCl (pH 8). Cell disruption was carried out by sonication (two times for 3 min) and inclusion bodies were separated from soluble proteins by centrifugation at 2600 g for 10 min.
Lipolytic activity was measured using the p-nitrophenyl palmitate as substrate according a modified Winkler and Stuckmann method (Winkler and Stuckmann, 1979). The substrate solution was prepared by mixing 5 ml of p-nitrophenyl palmitate stock solution (8 mM in propan-2-ol) with 50 ml of solution containing 50 mM Na2HPO4, 50 mM KH2PO4, 5 mM sodium deoxycholate and 0.1% (w/v) gum arabic. The measurement was started by adding 200 µl of substrate to the 10 µl aliquot of cell lysate in 96-well plate. The absorbance of the p-nitrophenol was spectrophotometrically measured at a wavelength of 410 nm in 20 s intervals for 20 min at 30°C. In parallel with every measurement, the blank sample (buffer used instead of cell sample) was used to set a zero point to avoid incorrect measurements as a result of the spontaneous hydrolysis of the substrate. Each sample was measured in triplicate.
This work is dedicated to the memory of Benjamin Toubiana. We thank members of the ‘Pseudomonas’ group for fruitful discussions, A. Hachani for helping during PlpD purification for antibodies production and Pascal Hingamp for advices in bioinformatic analysis. R.S. was financed with a PhD fellowship from the French Research Minister. F.K. is financed with a PhD fellowship in frame of ‘Marie Curie Actions Antibiotarget’ project financed from European Union. S.W. is financed by Deutsche Forschungsgemeinschaft in the priority program SPP1170. A.F. is supported by the Royal Society. This work was supported by grants-in aid for scientific research from ‘EuroPathoGenomics’ REX (LSHB-CT-2005512061-EPG), ‘Antibiotarget’ (MEST-CT-2005-020278) and ‘Pathomics’ ERA-net PATHO (ANR-08-PATH-004-01).