- Top of page
- Materials and methods
The assembly of the wild-type and several mutant forms of the trimeric outer membrane porin PhoE of Escherichia coli was investigated in vitro and in vivo. In in vivo pulse–chase experiments, approximately half of the wild-type PhoE molecules assembled within the 30-s pulse in the native conformation in the cell envelope. The other half of the molecules followed slower kinetics, and three intermediates in this multistep assembly process were detected: a soluble trypsin-sensitive monomer, a trypsin-sensitive monomeric form that was loosely associated with the cell envelope and a metastable trimer, which was integrated into the membranes and converted to the stable trimeric configuration within minutes. The metastable trimers disassembled during sample preparation for standard SDS/PAGE into folded monomers. In vitro, the isolated PhoE protein could efficiently be folded in the presence of N,N-dimethyldodecylamine-N-oxide (LDAO). A mutant PhoE protein, ΔF330, which lacks the C-terminal phenylalanine residue, mainly followed the slower kinetic pathway observed in vivo, resulting in increased amounts of the various assembly intermediates. It appears that the ΔF330 mutant protein is intrinsically able to fold, because it was able to fold in vitro with LDAO with similar efficiencies as the wild-type protein. Therefore, we propose that the conserved C-terminal Phe is (part of) a sorting signal, directing the protein efficiently to the outer membrane. Furthermore, we analysed a mutant protein with a hydrophilic residue introduced at the hydrophobic side of one of the membrane-spanning amphipathic β strands. The assembly of this mutant protein was not affected in vivo or in vitro in the presence of LDAO. However, it was not able to form folded monomers in a previously established in vitro folding system, which requires the presence of lipopolysaccharides and Triton. Hence, a folded monomer might not be a true assembly intermediate of PhoE in vivo.
The cell envelope of Gram-negative bacteria consists of two membranes, the inner and the outer membrane, which are separated by the periplasm. To reach their destination, integral outer membrane proteins (OMPs) have to be transported across the inner membrane via the Sec machinery . As hydrophobic sequences would act as stop-transfer signals [2,3], OMPs do not contain such sequences. Hence, the structure of OMPs is entirely different from those of integral inner membrane proteins, which contain hydrophobic α helices as the prominent membrane-spanning segments. Recently, the crystal structures of several OMPs have been solved [4–12]. All these proteins are β barrels, consisting of 8–22 antiparallel, amphipathic β strands that span the membrane, with the hydrophobic side exposed to the lipids.
Whereas the mechanism of the transport of proteins across the inner membrane via the Sec machinery has already been studied intensively in the past decades, the knowledge concerning the later steps in the biogenesis of OMPs is still very limited. After their translocation as linear polypeptide chains across the inner membrane, they probably fold in the periplasm prior to their insertion into the outer membrane . Several periplasmic proteins, including SurA [14,15] and Skp [16–18] that may play a role in the folding of OMPs or in their targeting to the outer membrane have been identified, but their exact role is unknown. In addition, lipopolysaccharide (LPS), a major lipidic component of the outer membrane, has been implicated in the folding of OMPs [19–22]. It is unknown whether OMPs insert spontaneously into the outer membrane or whether a proteinaceous machinery is implicated in this process.
In our laboratory, PhoE protein of Escherichia coli is used as a model for studying the biogenesis of OMPs. The structure of this trimeric porin has been solved . Previously, we have studied the assembly of this porin extensively after its synthesis in vitro in an S135 lysate of E. coli. In the presence of low amounts of Triton X-100 (0.015%, i.e. at the critical micelle concentration), PhoE could be assembled into a folded monomer that displays a higher electrophoretic mobility in SDS/PAGE than the denatured protein; LPS and divalent cations were required for the formation of this form of the protein . Upon addition of outer membranes and increasing the detergent concentration to 0.08%, these folded monomers could be converted into stable trimers, which were incorporated in the membranes, suggesting that the folded monomer represents a bona fide assembly intermediate of the PhoE protein. On the basis of these in vitro experiments, it was suggested that the assembly of PhoE proceeds sequentially in the order: (a) formation of the folded monomer; (b) sorting to assembly sites in the outer membrane; (c) trimerization; and (d) insertion .
Especially in view of the requirement for detergent, it is not clear whether the results of the in vitro experiments indeed reflect the events in the PhoE assembly process in vivo. Therefore, any conclusion of these in vitro data has to be verified by in vivo experiments. The existence of a folded monomer of PhoE could be confirmed in vivo in pulse–chase experiments, followed by immunoprecipitation . However, because of the slow kinetics with which these folded monomers disappeared during the chase, it was not immediately clear whether this form of the protein indeed represents an assembly intermediate.
In the present work, we have studied the assembly of PhoE in vivo more extensively. The goal was to detect assembly intermediates of the protein and to determine their subcellular localization. Furthermore, an in vitro system with purified components would be extremely beneficial to gain insight in the assembly process. Hence, in this paper, we describe an efficient procedure for the folding of PhoE after its production in inclusion bodies. In addition to the wild-type protein, the assembly of two mutant proteins was studied in vivo and in vitro.
- Top of page
- Materials and methods
The assembly of outer membrane proteins in E. coli is a multistep process. We investigated the assembly of wild-type PhoE porin and of two mutant PhoE proteins in vivo and in vitro, in order to dissect the assembly process in separate steps.
Two pathways for the assembly process in vivo were revealed with the wild-type protein. Approximately 50% of the wild-type PhoE molecules synthesized followed very fast assembly kinetics. Directly after pulse-labelling of the cells, these molecules were recovered in the membrane fraction in a stable trimeric and trypsin-resistant conformation, which can be considered as the native conformation of the protein. The other half of the wild-type PhoE molecules followed a slower kinetic pathway. A possible explanation for the detection of two kinetically different assembly pathways is that the PhoE protein was overproduced and, as a consequence, was titrating out a putative chaperone. Consistent with this hypothesis is the observation that the proportion of the PhoE molecules that followed the fast kinetics increased when the synthesis level was reduced (data not shown). Because the fast kinetic assembly pathway appeared to be severely disturbed in the ΔF330 mutant protein, we propose that the C-terminal Phe, which is conserved among most bacterial OMPs, is (part of) the targeting signal that is recognized by the putative chaperone. This targeting signal is not necessary for progress in the slow kinetic assembly pathway. The folding of the ΔF330 mutant protein was not affected in the established in vitro system with LDAO, indicating that the protein is intrinsically able to adopt a native-like structure. This observation underscores the hypothesis that the ΔF330 mutant protein is defective in the fast kinetic assembly pathway because of the deletion of (part of) a targeting signal rather than being disturbed in folding itself. The TolA and TolB proteins have been reported to interact with folded porin molecules [44,45]. Hence, these Tol proteins might be the chaperones that target PhoE to the outer membrane.
Three different intermediates could be distinguished in the slow kinetic assembly pathway. Small amounts of a soluble form were detected. This form is probably located in the periplasm, as the signal sequence was cleaved off. The observation that the periplasmic DsbA protein could introduce disulfide bonds in a cystine-containing mutant form of PhoE  is consistent with the idea that PhoE passes through the periplasm on its way to the outer membrane. Furthermore, as the DsbA protein creates an element of tertiary structure, the mutant PhoE was at least partially folded prior to insertion into the outer membrane . Consistently, we detected unfolded as well as folded forms in the periplasm. However, a proportion of the proteins detected in the supernatant fraction appeared to be associated with lipids. Because of the small quantities of these proteins, we were not able to establish definitely whether the lipid-associated PhoE molecules corresponded to the trypsin-resistant folded forms, but this seems a plausible assumption. A water-soluble monomeric intermediate has also been reported for the OmpF protein . This form was secreted by spheroplasts and could be converted to trimers upon the addition of cell envelopes and small amounts of detergent, indicating that it was a real assembly intermediate. Furthermore, the periplasmic SurA protein has been observed to be involved in the conversion of unfolded monomers of LamB into folded monomers , consistent with the idea that folding is a periplasmic or periplasmically exposed membrane-associated event.
A second assembly intermediate detected was a non-native, trypsin-sensitive form, which was loosely associated with the membranes. As the proteins in the fractions obtained after sucrose gradient centrifugation had to be recovered by trichloroacteic acid precipitation, which denatures all folded forms, we were not able to identify whether this non-native form was associated with the inner or outer membrane. It is possible that the non-native intermediate is associated with the inner membrane and precedes the soluble intermediate. In accordance with this supposition is the observation that processed intermediates in the assembly of LamB have been localized in the inner membrane . Furthermore, an immature processed form of OmpA that accumulated after overproduction was probably associated with the inner membrane .
The third assembly intermediate detected in vivo was a trypsin-resistant metastable trimer, which dissociated in folded monomers in 2% SDS. This form was integrated into the membranes, as it could not be extracted with urea, and it was converted to the stable trimeric configuration with relatively slow kinetics. Previously, metastable trimers of LamB have been detected, which were stabilized with a half-time of several min . Also in the case of OmpF, the existence of a metastable trimeric intermediate, localized in the periplasm, has been reported . However, as this form was detected in the periplasmic fraction by immunoprecipitation in the presence of 1% Triton X-100, the possibility that this form was only formed in vitro during the immunoprecipitation procedure from an unfolded intermediate can not be excluded. As suggested for the LamB protein , the stabilization of the trimers might be induced by the increased interaction with other outer membrane components, such as LPS. Consistently, the stabilization of OmpF and LamB trimers appeared to be affected in deep-rough mutants of E. coli.
Importantly, although we did detect folded monomers on gels, most, if not all, appeared to be generated by the dissociation of metastable trimers. Probably, the folded monomers of PhoE and LamB that were previously detected in vivo[15,24,51] were derived from SDS-sensitive trimers as well. In contrast, the folded monomers that are formed in vitro by incubation of de novo synthesized protein with LPS and a low concentration of Triton X-100  represent a real monomeric form, as incubation in 0.05% instead of 2% SDS prior to electrophoresis did not reveal any oligomeric bands on SDS/PAGE (H. de Cock, unpublished observation). Moreover, because this in vitro formed folded monomer could subsequently be trimerized upon the addition of outer membranes, it is an assembly intermediate, at least in vitro. However, the failure to detect a folded monomer in vivo in the pulse–chase experiments could mean that such a form of PhoE does not exist as a bona fide assembly intermediate in vivo, and that folding of the porins is initiated by the assembly of the trimeric subunit interface, as has been suggested previously . Consistent with this hypothesis are the results obtained with the G185E mutant. The introduction of a charged residue at the hydrophobic side of β strand 9 appeared to be tolerated in vivo and during folding in vitro in LDAO. However, the G185E mutant protein failed to form folded monomers with LPS and Triton after its synthesis in vitro. These results again strongly suggest that the folded monomer is not a true folding intermediate in vivo.
We therefore propose the following model for the assembly pathway of PhoE. After translocation across the inner membrane, the signal sequence is cleaved off, producing a mature PhoE protein, which is very rapidly converted to the native trimeric conformation in the outer membrane, probably in a chaperone-mediated process. The ΔF330 mutant protein is blocked in this process. This mutant protein and, probably as a consequence of overproduction, a proportion of the wild-type molecules, follow slower kinetics. We suppose that the different stages detected in the slower kinetic pathway still reflect the stages in the fast kinetic pathway, but that the process is retarded due to limited interactions with one or more chaperones or with assembly sites in the outer membrane. Three intermediates in this slower pathway were detected in the pulse–chase experiments. We propose that the membrane-associated, non-native monomer is still associated with the inner membrane, from where it is released to create the soluble non-native monomer. Subsequently, the periplasmic intermediate is sorted to the outer membrane, where it is converted to a metastable trimer, which inserts into the membrane and is converted within minutes to the stable trimeric configuration. Further analysis of an array of PhoE mutants that is available in our laboratory in the developed in vivo and in vitro assembly assays will be helpful to verify the proposed working model.