In vitro insertion and assembly of outer membrane protein PhoE of Escherichia coli K-12 into the outer membrane. Role of Triton X-100.

The assembly of the in vitro synthesized outer membrane protein PhoE into purified outer membranes was investigated. The assembly appeared to be strongly stimulated by the presence of low amounts of Triton X-100 (optimal 0.08%, w/v). The role of Triton X-100 in the in vitro system was further examined. Pretreating outer membranes with Triton X-100 did not make the membranes competent for correct assembly, indicating that the detergent did not act on the membrane but at the protein level. PhoE became assembly-incompetent with a half-life of approximately 12 min and 90 s at 37 degrees C in the absence and presence, respectively, of 0.08% Triton X-100. Apparently, Triton X-100 induces an assembly-competent state in the PhoE protein with a very short half-life. Furthermore, the efficiency of correct assembly of PhoE was greatly reduced when outer membranes of deep rough lipopolysaccharide mutants were used, indicating an important role of lipopolysaccharides in the assembly of the porin.

The assembly of the in vitro synthesized outer membrane protein PhoE into purified outer membranes was investigated. The assembly appeared to be strongly stimulated by the presence of low amounts of Triton X-100 (optimal 0.08%, w/v). The role of Triton X-100 in the in vitro system was further examined. Pretreating outer membranes with Triton X-100 did not make the membranes competent for correct assembly, indicating that the detergent did not act on the membrane but at the protein level. PhoE became assembly-incompetent with a half-life of approximately 12 min and 90 s at 37°C in the absence and presence, respectively, of 0.08% Triton X-100. Apparently, Triton X-100 induces an assemblycompetent state in the PhoE protein with a very short half-life. Furthermore, the efficiency of correct assembly of PhoE was greatly reduced when outer membranes of deep rough lipopolysaccharide mutants were used, indicating an important role of lipopolysaccharides in the assembly of the porin.
The outer membrane (OM) 1 of Escherichia coli K-12 contains three related proteins, OmpF, OmpC, and PhoE, that form general pores through which small hydrophilic solutes can pass (1,2). The functional unit of these porins is a trimer. Intriguing questions are how these proteins reach their final destination in the cell and how they are assembled into trimers. We have described an in vitro system to study the assembly of PhoE protein (3). Using monoclonal antibodies that recognize conformational epitopes, it could be demonstrated that an in vitro synthesized quasi-mature PhoE protein could fold into a monomeric configuration, resembling a native subunit in the trimer. High concentrations of Triton X-100 (TX-100) (optimal 2%, w/v) could induce the formation of heat-stable trimers, resembling the in vivo formed native trimers, whereas the presence of purified OMs drastically increased the kinetics of this process (3,4). The OM components required for this trimerization activity were not identified. Although interactions between outer membrane proteins and lipopolysaccharides (LPS) have been demonstrated (5) and suggested to be implicated in the assembly process (6 -12), purified LPS did not induce assembly of PhoE trimers in this in vitro system (3), showing that LPS is either not involved or not sufficient for the assembly process. In addition, trimers of OmpF (13,14) have been recently reconstituted in the absence of LPS. However, in this case detergents and/or phospholipids were used that might mimic the LPS activity.
In the present work, we attempted to extend the studies on the in vitro assembly of PhoE by developing a system to study the insertion of the protein into the OMs. Recently, it has been shown that low amounts of TX-100 (0.03%) can induce the insertion and assembly of the mature OmpF protein, secreted by spheroplasts (13), into purified OMs in vitro. Furthermore, OmpF protein, synthesized in an in vitro transcription-translation system, was assembled into trypsin-resistant trimers in the presence of OMs (15), suggesting correct insertion of OmpF into OMs. In this latter system, the presence of TX-100 was not required for trimerization. In contrast, in vitro synthesized PhoE protein was not assembled into a trypsin-resistant configuration when incubated with OMs in the absence of TX-100 (16). Here, we demonstrate that low amounts of TX-100 strongly stimulate the assembly of PhoE into purified OMs, and we investigated the role of the detergent and of LPS in this process in more detail.

MATERIALS AND METHODS
Bacterial Strains-S135 cell extracts were isolated from the E. coli K-12 strain MC4100 (17). Membranes were also isolated from this strain, as well as from E. coli U20 strains containing smooth LPS (18), galU mutant MC1000 containing LPS of chemotype Rd 1 (19), and CE1229, a derivative of MC4100, containing heptose-deficient LPS (20). The LPS content was determined by 3-deoxy-D-manno-octulosonic acid measurements (21) after precipitation of the membranes with acetone to remove the sucrose.
In Vitro Translations and Immunoprecipitations-Isolation of S135 cell extracts, preparation of the membranes, and the in vitro transcription and translation reactions were performed as described previously (22). Shortly thereafter, plasmid pJP370 (16) was used to direct the synthesis of a quasi-mature PhoE protein, containing, instead of the signal sequence, only a methionine and a serine at the N terminus of complete mature PhoE. Correctly folded proteins were immunoprecipitated with mAb PP1-1 (23) and protein A-Sepharose CL-4B. Immunocomplexes were dissociated by incubation in 200 mM glycine/HCl, pH 6.0. Plasmid pMS54 2 was used to synthesize quasi-mature PhoE protein lacking the C-terminal phenylalanine (⌬Phe 330 ). Plasmid pJP29 (25) contains the intact phoE gene and was used to synthesize the precursor form of PhoE protein.
Association and Insertion into the OM-To study association and insertion of PhoE protein into OMs, the protein was synthesized in vitro at 37°C in a 50-l system, and 25 min after initiation of translation, puromycin (10 M) was added. Subsequently, OMs and TX-100 were added, and the solution was for a short time mixed on a Vortex mixer and incubated for another 30 min at 37°C. To study the association of PhoE to the membranes, the total mixture was centrifuged for 30 min * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
To determine the amount of correctly assembled (i.e. protease-resistant) PhoE in the OM, 1 l of trypsin (1400 g/ml in 100 mM Tris-HCl, pH 8.0, 50 mM MgCl 2 ) was added, and the mixture was incubated for 30 min at 37°C unless stated otherwise. When insertion kinetics were determined, 2 l of the trypsin solution was added. This amount of trypsin is sufficient to degrade all trypsin-sensitive PhoE within 10 s (data not shown). After the addition of 1 l of phenylmethylsulfonyl fluoride (in 96% ethanol; 1 mM end concentration) and 5 min of incubation at 0°C to inhibit protease activity, trimers were extracted from the membranes by incubation in 0.5 M NaCl and 10 mM EDTA. After 45 min of incubation at 37°C, proteins were either immunoprecipitated with mAb PP1-1 or directly analyzed by SDS-PAGE.
SDS-Polyacrylamide Gel Electrophoresis and Quantifications-The LPS chemotype was analyzed by SDS-PAGE (18). Proteins were separated on SDS-polyacrylamide gels as described (26). Prior to electrophoresis, protein aliquots were incubated in sample buffer for 10 min at room temperature to detect folded monomers and at 56°C to detect heat-stable trimers. Electrophoresis was performed at 4°C in a temperature-controlled room at 20 mA in those cases where folded proteins had to be detected on the gels. Otherwise, gels were run at 30 mA at room temperature. Proteins in the gels were fixed and stained with 0.1% Coomassie Blue R-250. Gels were incubated for 30 min with Amplify (Amersham Corp.) and dried at 80°C. X-ray films were exposed at Ϫ80°C for appropriate time periods to be within the linear exposure range. Quantifications were performed on an LKB 2222-010 Ultroscan-XL laser densitometer. The amount of heat-stable trimers (56°C) was calculated by using the formula (M 100°C Ϫ M 56°C )/M 100°C ϫ 100%, where M is the amount of PhoE monomers determined at the indicated temperature.
To determine the amount of radiolabeled protein synthesized in vitro, [ 35 S]methionine-labeled proteins were separated by SDS-PAGE. Gels were dried after staining of the proteins as described above. To extract radiolabeled PhoE, gel pieces were incubated for approximately 18 h at room temperature in the dark with Lumasolve/water/Lipoluma as described by the manufacturer (Lumac ‫ء‬ LSC B.V.). The amount of radiolabeled proteins in the original sample was calculated after determining the amount of disintegrations/min in a scintillation counter (Beckman LS6000SE). Data were corrected for background radioactivity levels, which were determined with gel pieces of equal size from the same gel, and for incorporation of non-radioactive methionine, which was present in the S135 extract.
Treatment of OMs with Detergents-OMs were first isolated from a crude membrane fraction, containing both IMs and OMs, by centrifugation for exactly 5 min, not including the time required to start and stop the rotor, at 40,000 rpm in a Ti-50 rotor (Beckman Instruments) at 4°C. This step provides specific separation of OMs from IMs (16). The pellet containing the OMs was dissolved in buffer L. Subsequently, 10 l of OMs (6.65 nmol of LPS) were mixed with 65 l of buffer L and 25 l of a detergent solution in water. After incubating for 30 min at room temperature, OMs were reisolated by centrifugation in a TLA 100.2 rotor (30 min, 40,000 rpm at 15°C) and resuspended in 100 l of buffer L. Aliquots of pellet (OMs) and supernatant fractions (containing extracted material) were rapidly frozen in liquid nitrogen and stored at Ϫ80°C.

TX-100 Stimulates Insertion and Assembly of PhoE into the
Outer Membrane-In vivo synthesized PhoE is resistant to trypsin degradation when correctly assembled in the OM (27). Trypsin resistance can therefore be used as a criterion to establish whether PhoE is correctly assembled in vitro. To investigate whether low concentrations of TX-100 can induce the correct insertion and assembly of PhoE into OMs, quasi-mature PhoE was synthesized in vitro and incubated with OMs in the absence or presence of TX-100. Subsequently, OMs were pelleted and resuspended in buffer L. After 30 min of incubation at room temperature with trypsin, trimers were extracted from the membranes, and the folded proteins were immunoprecipitated with mAb PP1-1, which recognizes a conformational epitope, and analyzed by SDS-PAGE (Fig. 1). Indeed, trypsinresistant trimers could be extracted from the OMs (Fig. 1). Optimal insertion was observed with 0.06% (lanes g and h) to 0.1% TX-100 (lanes i and j). Only very low amounts of protein were inserted into OMs in the absence of TX-100 (lanes a and b; visible only after long exposure of the film). In the absence of OMs, no protease-protected proteins were detected (lanes q and r). Some insertion was observed when IMs instead of OMs were used (lanes o and p; visible only after long exposure of the film), but this is most likely due to the presence of small amounts of OMs in this membrane fraction.
When the protein samples containing trypsin-resistant PhoE were not heated before electrophoresis (Fig. 1), several distinct forms of PhoE could be detected. Next to the expected forms, i.e. dimers (visible only after longer exposure times) and trimers, a monomeric form was detected, migrating slightly faster (M r of 36,000) than the completely denatured PhoE. This form is not identical to the previously described folded monomer, which has an M r of 31,000 (3). Since this 36-kDa species was immunoprecipitated with the conformation-dependent mAb, it could represent a pool of inserted folded monomers that may be an intermediate in the assembly process. Alternatively, it may arise by denaturation of intermediates (folded monomers, dimers, or unstable trimers) of the assembly process. The 36-kDa species was not always observed in different experiments.
Trypsin-resistant PhoE proteins were not extracted from the OMs after incubation for 30 min with 4 M urea at 0°C in buffer L, indicating that these proteins were integrated into and not peripherally associated with the OMs (data not shown). Analysis of the kinetics of the process indicated that the insertion of PhoE into the OM was completed within approximately 3 min at 37°C (data not shown).
The Efficiency of Insertion into OMs Depends on the OM Concentration-Insertion of PhoE into OMs is expected to depend on the amount of available insertion sites in the OM. Therefore, insertion experiments were performed with various amounts of OMs isolated from strain U20. After trypsin treatment, OMs were pelleted, and both the OMs and supernatant fractions were analyzed (Fig. 2, A and B). With increasing amounts of OMs, a gradual increase in the amount of trypsinresistant PhoE was observed to be inserted in the OM (Fig. 2, A and C) up to a certain maximum when apparently the amount of insertion-competent PhoE becomes limiting. Incubation of the samples at room temperature prior to electrophoresis revealed that trimers, as well as small amounts of folded monomers, were present ( Fig. 2A). The trimers were resistant to denaturation in sample buffer at 56°C, comparable with trimers formed in vivo. Quantification of the fluorograms indicated that approximately 50% of the total amount of trypsinresistant PhoE protein was assembled into heat-stable trimers.
Similar analysis of supernatant fractions (Fig. 2, B and C) revealed that PhoE was also assembled in a trypsin-resistant form that was not pelleted together with the OMs. We therefore conclude that this fraction contains PhoE proteins that were assembled in an insertion-independent manner. Significant amounts of trypsin-resistant PhoE were detected in this fraction after incubation with OMs at concentrations that were suboptimal for insertion (Fig. 2C). At higher concentrations of OMs, the amount of trypsin-resistant PhoE in the supernatant fraction decreased. Both folded monomers (Fig. 2B, M*) and trimers (T), most of which were heat-labile, were detected in this fraction.
The relative efficiency of insertion of PhoE into OMs varied considerably between different experiments, and in some experiments, up to 50% of the total amount of protein synthesized was assembled into OMs. The variation in efficiency might result from the different amounts of PhoE protein synthesized in independent experiments. Quantification of the amount of protein synthesized in vitro in different experiments showed that it varied between 0.6 and 12 fmol of PhoE/50 l of translation volume, which can explain the differences in the relative insertion efficiency at a constant amount of OMs. Therefore, one batch of in vitro synthesized protein was used in those cases where direct comparison of insertion efficiencies was required. Furthermore, the source of the OMs was of importance. Lower amounts of PhoE protein were assembled into OMs of strain MC4100 than assembled in those of strain U20 (Table I). In addition, only very low amounts of PhoE protein were assembled in an insertion-independent manner when incubated with OMs of the former strain. Therefore, OMs derived from strain U20 were used in further experiments. It is remarkable that relatively high amounts of outer membranes are required for insertion and assembly of very low amounts of PhoE protein. Overall, these data suggest either that the OMs used in these experiments contain limiting amounts of insertion sites or that a special, but limiting, subset of the isolated OMs is able to take up the PhoE protein.
Pretreatment of OMs with TX-100 Does Not Make OMs Competent or Incompetent for Insertion-The role of TX-100 in the in vitro assembly of PhoE was further investigated. TX-100 might increase the competence of OMs in taking up outer membrane proteins, for example, by creating insertion sites. To test this possibility, OMs from strain U20 were pretreated with various amounts of TX-100, reisolated, and used in insertion assays with or without 0.08% TX-100 (Table II). To detect possible effects of the pretreatments on insertion of PhoE, OMs were used in amounts supporting suboptimal insertion. Pretreatment of OMs with TX-100 was not sufficient to render them fully competent for insertion and assembly into trimers (Table II; ϪTX-100, PEL). In the absence of TX-100, some trypsin-resistant PhoE protein was inserted into the TX-100pretreated OMs, but the majority of these were not assembled as heat-stable trimers but as folded monomers into the membrane. Addition of TX-100 (0.08%) during the incubation of PhoE with the membranes was again required to obtain correct insertion of PhoE as heat-stable trimers in the TX-100-pretreated OMs (Table II; ϩTX-100, PEL). Similar amounts of PhoE were inserted into OMs not pretreated with TX-100, and in both cases, approximately 50% of the inserted proteins were assembled into heat-stable trimers. These results demonstrate that pretreatment of OMs with TX-100 does not make them fully competent for PhoE assembly, suggesting that TX-100 works at the level of the PhoE protein, rather than on the membrane.
Pretreatment of OMs with TX-100 probably results in the formation of mixed micelles of the detergent with OM components or of OMs loaded up with TX-100. When no additional TX-100 is added, PhoE probably interacts with these TX-100/OM micelles resulting in the formation of trypsin-resistant folded monomers. Mixed micelles are probably also responsible for the presence of trypsin-resistant monomers in the superna- tant fraction after incubation in the absence of 0.08% TX-100 (Table II; ϪTX-100, SUP). These folded monomers might have been released from the micelles and therefore not recovered by centrifugation. Apparently, addition of 0.08% TX-100 to the in vitro synthesized PhoE protein is required for association with the OMs where assembly into heat-stable trimers occurs.
Pretreatment of OMs with TX-100 resulted in a drastic decrease of the insertion-independent assembly pathway (Table  II). Possibly, some OM components are inactivated or removed from the OMs by the pretreatment. Analysis of the TX-100extracted OMs and their corresponding supernatant fractions by SDS-PAGE revealed that only very low amounts of outer membrane proteins were extracted (data not shown). However, substantial amounts of LPS were removed by TX-100 treatment (Fig. 3). Therefore, the strong decrease in the efficiency of the insertion-independent assembly pathway (Table II) could indeed be due to removal of LPS or other OM components by TX-100 extraction. Thus, the assembly of large amounts of PhoE protein into a trypsin-resistant configuration, without concomitant insertion, proceeds due to the presence of LPS or other OM components extracted from OMs by 0.08% TX-100. Interestingly, extraction of OMs with TX-100 did not remove components required for efficient insertion, since similar amounts of PhoE were correctly assembled in OMs pretreated with or without TX-100.
TX-100 Affects the Assembly-competent State of PhoE-The results described in the previous section suggest that TX-100 acts at the PhoE protein level rather than on the membranes. TX-100 might be required to modulate the conformation of PhoE to obtain an assembly-competent state. We noticed that the amount of assembly-competent PhoE greatly decreased when PhoE was preincubated with TX-100. The half-life of the assembly-competent state of PhoE at 37°C was reduced drastically from approximately 12 min in the absence (Fig. 4A) to approximately 90 s in the presence of 0.08% TX-100 (Fig. 4B). The half-life of the assembly-competent state in the presence of 0.08% TX-100 appeared to be temperature-dependent, since it was substantially longer at 0 than at 37°C (data not shown). These data suggest that TX-100 affects the folding of PhoE, since folding is temperature-dependent.
Recently, we have demonstrated that removal of the C-terminal phenylalanine of PhoE drastically reduces the efficiency of insertion of the protein into OMs in vivo (28) and in vitro (this study). 2 This decreased efficiency appears not to be due to a greatly decreased half-life of the assembly-competent state that was, both in the absence (data not shown) and in the presence (Fig. 4B) of TX-100, similar to that of wild-type PhoE. The signal sequence of PhoE interferes with the in vitro folding of PhoE into a native-like structure (3). In addition, it has been   Fig. 2. The amount of trypsin-resistant PhoE protein present in the OMs pellet (P), the supernatant (S), and the total amount (T) are indicated as percent of the total amount of PhoE synthesized. Results of a representative experiment are shown in which OMs of MC4100 and of U20 were directly compared using a single translation mixture. shown that the signal sequence is able to retard the folding of the precursor of maltose-binding protein into its native state (29). Therefore, it has been suggested that an important role of the signal sequence is to keep the precursors in a translocationcompetent state. However, the presence of the signal sequence did not increase the half-life of the assembly-competent state of the precursor form of PhoE in the presence of TX-100 but only reduced the efficiency of assembly of this protein (Fig. 4B). (30) and other outer membrane proteins (31), suggesting that they are affected in the assembly and/or insertion of these proteins. We therefore investigated whether mutations in the LPS core region affected OM insertion in the developed in vitro system. OMs of strains with various LPS chemotypes were incubated with equal amounts of in vitro synthesized PhoE protein in the presence of 0.06% TX-100, and the amount of correctly assembled trypsin-resistant PhoE was determined (Fig. 5). OMs containing LPS of chemotype Rd 1 and especially of chemotype Re were affected in the efficiency of the in vitro insertion process. DISCUSSION The transport of outer membrane proteins to their final destination can roughly be divided into two steps, (i) transport across the IM and (ii) insertion and assembly into the OM. In the case of PhoE, the first step was previously reconstituted in vitro (22). In the present study, the reconstitution of the second step, which up to now was only partially achieved (3,4), was completed. We demonstrated that the presence of low amounts of TX-100 strongly stimulate the insertion of PhoE protein into purified OMs. The data support the idea that the in vitro synthesized PhoE protein forms an assembly-competent intermediate with a short half-life in the presence of low amounts of TX-100. This assembly-competent intermediate proceeds via either one of two assembly pathways, depending on the amount of OMs present. One of these pathways includes insertion into the OMs, but the other does not. An important question is whether the in vitro assembly resembles the in vivo process. Two arguments can be given that this is indeed the case. First, mutations that affect the structure of the core region of the LPS have been shown to result in decreased amounts of PhoE (30) and other outer membrane proteins in vivo (31). Here, we have demonstrated that such membranes are also in vitro defective in the insertion-dependent assembly of PhoE. Second, point mutations in phoE have been described that affect the efficiency of assembly of the protein into the OM in vivo. In one of these mutants, the C-terminal phenylalanine was deleted (28). This mutant protein was also defective in vitro in TX-100induced insertion into the OM (this study). 2 Thus, the fact that assembly defects in vivo are reflected in the in vitro system strongly suggests that these processes are alike. Therefore, this in vitro system will be very useful in dissecting the assembly process in distinct stages and in studying the roles of individual amino acids of PhoE and of cellular components in the assembly process.

Mutations in LPS Core Region Decrease the Efficiency of the Assembly of PhoE into OMs-Deep rough LPS mutants have been shown to contain reduced amounts of PhoE
An important question concerns the role of TX-100 in the insertion and assembly of PhoE. The finding that TX-100 induces an assembly-competent state in PhoE suggests that the detergent might act like a molecular chaperone. Interestingly, in this respect, detergents or mixed micelles between detergents and lipids have been proposed to provide both polar and non-polar interaction sites for unfolded proteins, thereby reducing the probability for intra-or intermolecular protein interactions that lead to misfolding or aggregation (32). Thus, TX-100 in the in vitro system might substitute for a molecular chaperone that is active in the folding of PhoE in vivo. Alternatively, a putative chaperone that associates with PhoE during the in vitro synthesis may be dissociated from PhoE by TX-100. Such a chaperone could be required to stabilize an assembly-competent state of PhoE in the periplasm and for targeting of the protein to the OM. After removal of such a chaperone by TX-100, the folding of PhoE would proceed via a short-lived insertion-competent state into an incorrectly folded protein. We have investigated whether other compounds could substitute for TX-100 in stimulating the insertion of PhoE into the OM. Only the non-ionic detergents, like n-octyl ␤-D-glucopyranoside, seem to support the insertion of PhoE into a trypsin-resistant form (data not shown). A lysophospholipid C16 -1 c -lysophosphatidylcholine, which can be considered as a biological detergent, or purified LPS was not active (data not shown).
Recently, we have shown that high concentrations of TX-100 (0.5% or more) can induce the folding of PhoE into heat-stable trimers (4). Other non-ionic detergents could not substitute for TX-100 unless they contained, like TX-100, a phenyl ring. Folded monomers of PhoE were detected rapidly after the addition of the detergent, but trimers appeared only very slowly and not at the expense of the amount of folded monomers. Therefore, the folded monomer did not seem to be a precursor of the trimer, and the trimer appeared to be formed from a folding intermediate with a long half-life. At the low detergent concentration used in the present study (0.06 -0.08%), entirely different processes appear to have occurred. At these low detergent concentrations, heat-stable trypsin-resistant trimers can be formed, but the presence of OM components is essential. Furthermore, trimerization occurs via a folding intermediate with a very short half-life, and finally, other non-ionic detergents could substitute for TX-100.
The reduced efficiency of assembly of PhoE into OMs of LPS mutants might be explained by reduced amounts of insertion sites in these OMs. Freeze-fracture electron microscopy has revealed that the OM contains particles that probably consist of protein and LPS complexes, stabilized by divalent cations (33,34). However, particles can also be formed by divalent cations and LPS together (34,35). Interestingly, OMs of LPS mutants contain reduced amounts of particles (35). It was proposed that particles in the outer monolayer of the OM form hemimicelles that make complementary impressions in the inner monolayer (24). These impressions create a local increase in surface curvature at the periplasmic side of the OM. Further research will be required to determine whether these particles represent insertion sites.