Phospholipid-assisted Refolding of an Integral Membrane Protein

Escherichia coli-derived phosphatidylethanolamine (PE) or PE with fully saturated fatty acids was able to correct in vitro a defect in folding in the lipid-dependent epitope 4B1 of lactose permease (LacY) resulting from in vivo assembly in the absence of PE. PE plasmalogen, PE with two unsaturated fatty acids, and lyso-PE, which all do not favor bilayer organization, did not support proper refolding. Proper refolding occurred when these latter lipids were mixed with a bilayer-forming lipid (phosphatidylglycerol), which alone could not support refolding. l-Phosphatidylserine (PS; natural diastereomer) did support proper refolding. PE derivatives of increasing degrees of methylation were progressively less effective in supporting refolding, with phosphatidylcholine being completely ineffective. Therefore, the properties of nonmethylated aminophospholipids capable of organization into a bilayer configuration are essential for the recovery of the native state of epitope 4B1 after misassembly in vivo in the absence of PE. Neitherd-PS (sn-glycero-1-phosphate backbone) nor P-d-S (d-serine in the head group) is competent in supporting proper refolding unless used in binary mixtures with phosphatidylglycerol. The detailed characterization of phospholipid-assisted refolding reported here further supports a specific rather than nonspecific role for PE in structural maturation of lactose permease in vivo (Bogdanov, M., and Dowhan, W. (1998) EMBO J. 17, 5255–5264).

Escherichia coli-derived phosphatidylethanolamine (PE) or PE with fully saturated fatty acids was able to correct in vitro a defect in folding in the lipid-dependent epitope 4B1 of lactose permease (LacY) resulting from in vivo assembly in the absence of PE. PE plasmalogen, PE with two unsaturated fatty acids, and lyso-PE, which all do not favor bilayer organization, did not support proper refolding. Proper refolding occurred when these latter lipids were mixed with a bilayer-forming lipid (phosphatidylglycerol), which alone could not support refolding. L-Phosphatidylserine (PS; natural diastereomer) did support proper refolding. PE derivatives of increasing degrees of methylation were progressively less effective in supporting refolding, with phosphatidylcholine being completely ineffective. Therefore, the properties of nonmethylated aminophospholipids capable of organization into a bilayer configuration are essential for the recovery of the native state of epitope 4B1 after misassembly in vivo in the absence of PE. How a polytopic membrane protein inserts into the membrane and adopts its native conformation remains one of the major unresolved questions of biology. During membrane assembly, integral membrane proteins must interact with other proteins and with the lipid bilayer itself, resulting in their proper conformational maturation. The role phospholipids play in forming a membrane bilayer and a hydrophobic environment for membrane protein folding and assembly is well accepted (1,2); however, the role specific phospholipids play in assisting or directing folding of membrane proteins has not been extensively explored. In order to determine the role that individual phospholipids play in the folding of membrane proteins in Escherichia coli, we are using the lactose permease (LacY) 1 as a model system. LacY is one of the most intensively studied integral membrane proteins for which there is both extensive structural information (3,4) and monoclonal antibodies (mAbs) directed against several extramembrane epitopes whose recognition depends on secondary or tertiary structural organization (5,6). The organization of extramembrane and transmembrane domains of LacY is characteristic of the major facilitator superfamily of transport proteins (7), making results obtained from studies on LacY applicable to a variety of other transport proteins.
Employing a novel blotting technique (Eastern-Western) in which proteins are exposed to phospholipids bound to a solid support during renaturation from SDS in the standard Western blotting procedure, we previously presented the following experimental evidence (8,9) that the phospholipid phosphatidylethanolamine (PE) acts as a non-protein molecular chaperone in the folding of LacY: (i) LacY appears to fold in vivo with the assistance of PE as assessed by recognition by a conformation-sensitive mAb directed against periplasmic loop P7 (Phe 242 -Gly 254 ), which is flanged by transmembrane domains VII and VIII and defines epitope 4B1 (Fig. 1); (ii) LacY assembled initially in vivo in the presence of PE either retains nativelike conformation of this epitope throughout or has sufficient "conformational memory" to reform this epitope after SDSpolyacrylamide gel electrophoresis (PAGE) and Western blot analysis in the absence of added lipid; (iii) after SDS-PAGE and Western blotting, no phospholipid can be detected associated with LacY (less than one molecule of PE per 75 LacY monomers); (iv) LacY initially assembled in vivo in the absence of PE can form the native-like conformation of epitope 4B1 during refolding from SDS in the presence of specifically PE but not phosphatidylglycerol, CL, or phosphatidylcholine (PG, CL, and PC, respectively); (iv) extensive denaturation of LacY using urea-SDS eliminates epitope 4B1 in LacY originally assembled in PE-containing membranes and prevents its refolding in the presence of PE; (v) the requirement for PE in the folding of LacY has been demonstrated in an in situ assembly system using in vitro protein synthesis coupled with in vitro phospholipid synthesis in the presence of membranes initially lacking PE; (vi) PE is not required either prior to or concomitant with membrane insertion but is required in a late step of conformational maturation to attain native structure. Thus, PE corrects in vitro a LacY folding defect caused by either in vivo or in vitro assembly in PE-deficient membranes, but once this epitope is formed in vivo, PE is not required to maintain its conformation. Based on these results, we proposed as a general principle that phospholipids can act as molecular chaperones of non-protein origin that specifically mediate the folding of multispanning polytopic membrane proteins, thereby extending the definition of chaperones to other biomolecules in addition to proteins.
What are the minimum requirements of a phospholipid to facilitate membrane protein folding? In this study, we establish a specific requirement for a nonmethylated aminophospholipid of natural chirality and preference for lamellar organization for refolding of LacY into its native conformation in the region defined by epitope 4B1. The proper conformation of epitope 4B1 is required for full function of LacY as a transporter (5).
Bacterial Strains, Plasmids, and Growth Conditions-Strains carrying the pss93::kan null allele cannot synthesize PE and require either growth medium containing 50 mM MgCl 2 for viability or a functional plasmid-borne copy of the pssA gene (plasmid pDD72, temperaturesensitive for replication) to restore wild type phospholipid composition (11). Plasmid pT7-5 (12) carries the intact lacY gene under control of both lacOP and the T7 RNA polymerase promoter (p(T7)) and was used for in vivo high level expression of LacY. Cells were grown in LB-rich medium, and ampicillin (100 g/ml) was included in the growth medium for maintenance of plasmid pT7-5. Since replication of plasmid pDD72 is temperature-sensitive and mutants lacking PE require MgCl 2 for viability, all strains (mutant and wild type) were grown at 30°C in 50 mM MgCl 2 . Overexpression of the lacY gene on plasmid pT7-5 directed by the lacOP promoter was carried out in mutant and wild type strains grown exponentially in the presence of 1 mM isopropyl ␤-thiogalactoside. Strain JA200/pPSD2b carrying a copy of the psd gene (encodes phosphatidylserine decarboxylase) and a tet R marker on a multicopy number plasmid was grown in LB medium containing 25 g/ml tetracycline (13) and was used to prepare membranes enriched in phosphatidylserine decarboxylase.
Enzymatic Preparation of D-(C16:0)PS-E. coli strain JA200/pPSD2b was used to prepare a Triton X-100 membrane extract containing amplified levels of phosphatidylserine decarboxylase as described previously (13). Phosphatidylserine decarboxylase is specific for the L-isomer (sn-glycero-3-phosphate isomer) of PS and is not inhibited by the Disomer (13). A solution of 1.4 mM rac-(C16:0)PS and 8.6 mM Triton X-100 in 0.5 ml of 0.1 M potassium phosphate (pH 7.0) was supplemented with sufficient crude phosphatidylserine decarboxylase (0.2 units) to convert all of the L-PS to PE as judged by pilot experiments using L-(C16:0)PS as substrate. Phospholipids were extracted from the reaction mixture with chloroform/methanol/water (1:2:0.5), and PE was separated from D-(C16:0)PS by silica gel thin layer chromatography as described below.
Membrane preparation and SDS-PAGE-Inside-out membrane vesicles, the source of membrane protein in all experiments, were prepared as described previously (14), using a French press to break cells at 8,000 p.s.i. in the presence of 20 mM MgCl 2 . Protein was determined by the BCA assay according to the manufacturer's suggestions. Inside-out membrane vesicles were adjusted to the concentration of the SDS gel-loading buffer of 2.8% SDS, 10% glycerol, 100 mM dithiothreitol and heated at 37°C for 15 min and centrifuged to remove any insoluble material prior to being subjected to SDS-PAGE in 12.5% polyacrylamide as described previously (8).
Eastern-Western Blotting-Phospholipids (0.1-0.2 mg) were first developed on silica gel thin layer chromatography plates using chloroform/methanol/water/30% ammonium hydroxide (120:75: (15) and were blotted to either nitrocellulose or polyvinylidene difluoride sheets (Eastern blotting) as follows. The plate was dried thoroughly and then dipped in the blotting solvent of isopropyl alcohol/0.2% aqueous CaCl 2 /methanol (40:20:7, v/v/v) for 20 -30 s (16). First a nitrocellulose or polyvinylidene difluoride sheet and then a 3MM glass microfiber filter sheet (Whatman) were placed over the face-up plate, and the assembly was pressed for 30 s with a heating block at 130 -150°C applied to the filter. The sheet was then air-dried and used in the Western blotting protocol. The presence of phospholipids on polyvinylidene difluoride sheets was verified as described previously (16). In order to make a complex between phospholipid and antibiotic Ro09 -0198, a stock solution (50 mg/ml) in 50% Me 2 SO, 50% water (v/v) was diluted to 0.1 mg/ml in water. Whatman No. 4 filter paper was first immersed in this solution and then was sandwiched between the area of phospholipid on the above solid support and a glass plate for 10 min.
Nitrocellulose or polyvinylidene difluoride sheets with preblotted phospholipids were immersed in distilled water for 10 s. Developed SDS-PAGE gels were pre-equilibrated in Western blotting cathode buffer (see below) for 10 min before electroblotting the proteins to solid support to reduce but not completely remove SDS, which is important for the recovery of epitope recognition. The helical structure (secondary structure) of many membrane proteins is maintained in SDS, but the helix-helix interactions (tertiary structure) are largely disrupted. The optimal amount of SDS remaining in the gel must be empirically determined by varying immersion time and methanol concentration; too low a methanol concentration (Ͻ10%) will not facilitate removal of SDS from the gel and proteins, and too high a concentration can lead to aggregation of membrane proteins in the gel matrix. The gel was overlaid with the area containing LacY lined up with the patch of phospholipid on the face-up solid support. The use of prestained SDS-PAGE protein standards (Bio-Rad wide range) is helpful in aligning the gel and the solid support. Electroblotting for 90 min was carried out using a Milliblot-SDE semidry electroblotting apparatus under the following conditions: constant voltage power supply with initial current density of 2.5 mA/cm 2 of gel area. The following buffers were used: anode buffer 1 (0.3 M Tris-HCl, 10% methanol, pH 10.4), anode buffer 2 (25 mM Tris-HCl, 10% methanol, pH 10.4); and cathode buffer (25 mM Tris-HCl, 40 mM glycine, 20% methanol, pH 9.4). The semidry blotting system appears to be essential to the success of the process. The sheets were then blocked overnight with 5% bovine serum albumin in TBS buffer (10 mM Tris-HCl, pH 7.4, 0.9% NaCl) containing 0.05% Nonidet P-40 at 4°C. The sheets were washed once with TBS/Nonidet P-40 buffer for 15 min and incubated for 1 h in either mAb 4B1 or mAb 4B11 at a final dilution of 1:10,000 in the TBS/Nonidet P-40 buffer. The sheets were washed three times for 15 min each with TBS buffer/ Nonidet P-40 buffer. The washed sheets were incubated for 1 h with peroxidase-labeled anti-mouse antibody from the ECL detection kit (Amersham Pharmacia Biotech) diluted 1:10,000 in TBS/Nonidet P-40 buffer followed by three washes with TBS/Nonidet P-40 buffer and once with TBS buffer only. The sheets were incubated for 1 min with ECL detection reagents (Amersham Pharmacia Biotech) mixed 1:1 and immediately exposed to x-ray film for 10 s, 45 s, or 5 min.
Nitrocellulose sheets were stripped of the first antibody and reprobed with a different antibody as follows. The immunostained and developed sheets were washed twice with TBS-Nonidet P-40 at room temperature and submerged in stripping buffer (100 mM 2-mercaptoethanol, 2% SDS, 62.5 mM Tris-HCl, pH 6.7) and incubated at 50°C for 30 min in a sealed plastic bag with agitation. The sheets were washed twice in TBS/Nonidet P-40 and treated with ECL detection reagents as described above followed by exposure to x-ray film to ensure full removal of the first antibody. The sheets were washed twice again with TBS/ Nonidet P-40 and blocked by immersing in 5% bovine serum albumin in TBS/Nonidet P-40 buffer overnight at 4°C. Finally, immunodetection was performed with the indicated antibody as described above.

Effect of Lamellar Versus Non-lamellar Forming Phospholipids on LacY Refolding-We previously demonstrated (8) that
LacY assembled in vivo in PE-containing membranes, but not in PE-deficient membranes, subjected to SDS-PAGE and Western blot analysis is recognized by a conformation-specific mAb 4B1 directed against the continuous epitope 4B1 (Fig. 1) within the periplasmic domain P7 (8). However, if E. coli PE or total E. coli phospholipids were blotted to the solid support (Eastern blot) prior to transfer of proteins from the polyacrylamide gel, the LacY from PE-deficient cells regained recognition by mAb 4B1 (8). Thus, LacY lacking recognition by mAb 4B1 can be induced to form epitope 4B1 if partially denatured and then renatured in the presence specifically of PE. We next wished to refine our understanding of the collective physical properties and individual chemical properties of phospholipids necessary to assist in the refolding of LacY improperly assembled in vivo in membranes lacking PE.
As described under "Experimental Procedures," membrane preparations of LacY expressed in PE-deficient strain AD93 were subjected to Eastern-Western blotting using PE's with different fatty acid compositions followed by a determination of the degree of proper refolding of LacY as indicated by recognition by mAb 4B1 and the intensity of the immunoreactive LacY band after Western blotting. Fig. 2 illustrates a few examples of the results from many experiments summarized in Table I. Recognition by mAb 4B1 was restored with E. coli-derived PE (primarily C16 and C18 fatty acids saturated at 1-sn and unsaturated at the 2-sn) or synthetic PEs with C16 or longer saturated fatty acids at both positions. PEs with unsaturated fatty acids at both acyl positions or lyso-PE (any fatty acid composition) did not support proper refolding; (C12:0)PE was effective but at a reduced level.
Phospholipids with fully saturated fatty acids normally adopt a bilayer (lamellar) organization irrespective of hydration, pH, ionic strength, or divalent cation nature and concentration (17). Among unsaturated fatty acid-containing phospholipids, only PE adopts the non-bilayer inverted hexagonal II phase (H II ) at physiological temperatures, pH values, and salt concentrations, while divalent cations induce the H II phase for CL. PEs with combined long chain and more extensively unsaturated fatty acids such as PE plasmalogen adopt the H II phase almost exclusively. Lysophospholipids generally adopt a micellar organization. Phospholipids that adopt the bilayer phase exclusively can stabilize H II -preferring lipids (i.e. unsaturated PE) in an overall bilayer organization in mixed binary systems (18). The proportions of bilayer lipid required to achieve this can vary substantially (usually Ͼ 20%), with 30 mol % of PG mixed with (C18:1)PE being lamellar in its physical state. To distinguish between the potential contributions of individual phospholipid structure and the physical state of the lipid, we further investigated whether this difference between saturated and unsaturated phospholipid in supporting refolding might be explained by different phase-promoting properties of these lipids.
Consistent with the physical state of PE being an important determinant for renaturation, those forms of PE that can readily assume a bilayer or lamellar structure (19) at room temperature ( Fig. 2 and Table I, saturated fatty acids) supported renaturation, while those derivatives that tend to form micellar structures or the non-bilayer H II arrangement did not support renaturation. However, H II -forming lipids in binary mixtures with a lamellar or bilayer-forming lipid such as PG supported renaturation of LacY irrespective of the fatty acid composition of PG. The same was true for lyso-PE, but a higher proportion of PG was required probably to convert micellar to lamellar organization. PE plasmalogen, which also highly favors the H II phase over the lamellar phase, required a higher mol % of PG to be effective. There may be some advantage of saturated fatty acids over unsaturated fatty acids, since binary mixtures with (C18:1)PG were not as effective as with (C16: 0)PG in supporting renaturation by (C18:1) or (C18:2)PE. Although (18:1⌬9trans)PE by itself was not as effective as (C18: 0)PE, it did support significant refolding, especially when compared with (18:1⌬9cis)PE, which was only effective in the presence of PG. The basis for the reduced effectiveness of  (49). mAb 4B1 has been shown to recognize a conformation-sensitive, continuous, PE-dependent epitope (4B1) in the periplasmic hydrophilic loop P7 (5,8); the highlighted amino acids are part of this epitope. mAb 4B11 has been shown (6) to recognize a conformation-sensitive, discontinuous, PE-independent epitope (4B11) formed by the last two cytoplasmic loops, C8 and C10 (6,9). Formation of this epitope is dependent on membrane insertion of LacY (9).

FIG. 2. Effect of PE derivatives on the refolding of epitope 4B1.
LacY from PE-deficient cells was subjected to Eastern-Western blotting either without phospholipid (None) or the indicated phospholipid. E. coli refers to PE derived from E. coli. In binary mixtures, PE and PG were in an equal molar ratio. In A, two sets of samples (6 and 12 g) and in B one sample (12 g) of total membrane protein was subjected to SDS-PAGE and analyzed using mAb 4B1.
(C12:0)PE in the absence of PG is not apparent, but the short acyl chains may not provide a large enough hydrophobic domain to efficiently accommodate LacY.
Effect of Degree of Amine Methylation on Refolding-It has been established that LacY specifically requires a phospholipid with a free primary amine (such as PE or PS) to support active transport in an in vitro reconstituted proteoliposome system (29). Thus, both in vivo (14) and in vitro in the absence of PE or PS, LacY only carries out facilitated diffusion of substrates. PC, which is not present in E. coli, does not substitute for PE in supporting active transport, and we have previously shown that PC will not support proper refolding of LacY originally assembled in PE-deficient membranes (8). To gain more insight into the requirement of the phosphoamino group in the refolding of LacY, the renaturation of partially denatured protein was performed in the presence of progressively methylated derivatives of PE (Table I). As methylation increased or the availability of an ionizable proton decreased (PMME to PDME to PC), the effectiveness of the derivatives in supporting refolding of LacY decreased, with PC being completely ineffective. The nature of the fatty acid chains appears not to be a factor, since none of the derivatives of PC were effective. Although the PMME and PDME derivatives used were C18:1, both have been determined to favor the lamellar state (21). Unlike the ability of PG to rescue non-lamellar derivatives of PE, this lipid could not induce PC to properly refold LacY even when at 85 mol %. Therefore, either an ionizable amine or a small aminecontaining head group is required for the renaturation process.
Even more surprising was the finding that subjecting LacY originally assembled in the presence of PE, and therefore containing epitope 4B1, to the Eastern-Western procedure in the presence of PC resulted in loss of epitope 4B1 (Fig. 3, lanes 7-9  versus lanes 4 -6). Therefore, PC appears to act as an antagonist in the maintenance of native LacY structure. This unexpected finding may explain why native LacY reconstituted in the presence of PC lacks full function (29, 30) (i.e. only carries out facilitated but not active transport of substrate). Also shown in Fig. 3 (lanes 1-3) is an increase in the apparent yield of epitope 4B1 when LacY originally assembled in the presence of PE is subjected to Eastern-Western blotting in the presence of E. coli-derived PG/CL. This result is similar to a previous observation that total E. coli phospholipids were more effective than PE alone (8) although PG/CL cannot in themselves support refolding of LacY assembled in the absence of PE (Ref. 8 and Fig. 2B, lane 3).
Defining the Head Group Specificity of PE-Consistent with the above requirement (29,30) of either PE or PS for reconstitution of full LacY function in vitro, refolding of LacY (assembled in membranes lacking PE) in the presence of bovine brain PS led to recovery of epitope 4B1 (Table II and Fig. 4B, lanes 3  and 4). Therefore, PS is as effective as PE in reconstituting epitope 4B1. These results are also consistent with the lack of active transport mediated by LacY in cells lacking PE (14) but with the ability to actively transport lactose in mutants in which 95% of the PE has been replaced by PS (31).
The importance of the amino group of PE for the proper refolding of LacY was further tested by shielding the phosphoethanolamine head group of PE by the peptide antibiotic Ro09-0198, which accommodates the head group of PE specifically and does not interact with other phospholipids including PS (10). No refolding of LacY was observed when solid support was preblotted with PE treated with Ro09-0198 (Fig. 4A, lanes 1-3  versus lanes 7-9). In a control experiment, the specificity of Ro09-0198 for PE and its lack of interference with antibody detection were demonstrated by the lack of effect of the anti-  7-9). In each set of three samples, 3, 6, and 12 g of total membrane protein was subjected to SDS-PAGE and analyzed using mAb 4B1. biotic on PS-assisted refolding of LacY (Fig. 4B, lanes 5 and 6) and on the detection of LacY assembled in PE-containing cells (data not shown). Stripping of mAb 4B1 as described under "Experimental Procedures" and reprobing with mAb 4B11 verified the presence of ample amounts of LacY in all lanes (data not shown). mAb 4B11 recognizes a discontinuous epitope defined by cytoplasmic domains C8 and C10 ( Fig. 1) with the same efficiency whether or not assembly occurred in the presence of PE (9). Effectiveness of PS Analogues on Refolding-The ability of PS analogues to promote proper refolding of LacY was investigated (Table II). Bovine brain PS and L-(C16:0)PS were able to support refolding, but L-(C18:1)PS was only able to support refolding in binary mixtures with PG although PS normally adopts a lamellar organization irrespective of fatty acid composition (17). The unnatural diastereomers P-D-S and D-PS were also not able to support renaturation unless in binary mixtures with PG (Fig. 5); rac-PS did support renaturation but with about 50% of the efficiency when compared with the natural diastereomer. Although we were not able to investigate the dependence on fatty acid composition extensively, the lack of effectiveness of P-D-S appears not to be due to its fatty acid composition, since it is similar to that of bovine brain PS, which does support refolding. These differences between diastereomers are quite surprising as is the difference between the saturated and unsaturated derivations of PS. These results may be due to subtle differences in the physical properties between diastereomers and between PSs with different fatty acid compositions as discussed below. DISCUSSION The aim of the present study was to explore in more detail the specific properties of the head group, acyl chains, and physical state of phospholipids that support proper refolding of LacY. Although we do not understand the physical and chemical properties of the lipids and proteins once bound to a solid support in the Eastern-Western blotting procedure, several clear relationships from our studies can be made between the requirements to be effective in refolding of LacY and the physical and chemical properties in solution of phospholipids that support refolding: (i) a positively charged head group preferably with a small or ionizable amine; (ii) a fatty acid composition either singly or in binary mixtures that favors an overall bilayer or lamellar organization in solution; and (iii) a distinct preference for naturally occurring diastereomers of the aminolipid in single component systems or the presence of lamellar forming phospholipids of natural chirality in binary mixtures with an unnatural diastereomer of the aminolipid. These relationships clearly extend our previous conclusions supporting a distinct biological specificity for PE in phospholipid-assisted folding of LacY both in vivo and in vitro that goes well beyond simply providing a nonspecific detergent-like or two-phase environment for refolding of LacY.
Since the requirement for naturally occurring PE in the refolding of LacY had been established both in vivo and in vitro, the finding that unsaturated PE as opposed to fully saturated PE cannot facilitate refolding was surprising. This finding raised the question of whether the phase properties of phospholipids were a factor in the refolding process. The refolding effectiveness of the various PE derivatives singly or in binary mixtures with PG is consistent with the importance of the overall macrostructure of the lipid phase and strongly suggests that refolding of LacY is dependent an overall lamellar macrostructure.
Several reports have documented a preference or specificity for a particular physical organization of phospholipids. Protein translocation across the E. coli inner membrane lacking PE could be restored by (C18:1)PE (non-lamellar) but not (C14: 0)PE or (C18:1)PC (both lamellar) (32). (18:1⌬9cis)PE (nonlamellar) but not (18:1⌬9trans)PE (lamellar) significantly enhanced the respiratory control ratio of ubiquinol-cytochrome c reductase and ATP-induced membrane potential of the H ϩ -ATPase (33). The photochemical function of rhodopsin (34) and the ATPase activity of E. coli SecA protein (50) are enhanced by non-lamellar PE derivatives. However, mammalian Ca 2ϩ -ATPase activity (35) is stimulated significantly more by lamellar phase-than H II phase-forming derivatives of PE. Finally, lipopolysaccharide derivatives that tend to organize in nonbilayer structures were better facilitators of folding of PhoE protein in situ as it transits the inner membrane of E. coli (36).   3 and 4). Ro09-0189 was applied as indicated prior to Western blotting. In each set of samples, 3, 6, and 12 g (A) or 6 and 12 g (B) of the total membrane protein was submitted to SDS-PAGE followed by analysis using mAb 4B1. PS, which is only found in trace amounts in wild type E. coli, may present a more complex special case. Unlike PE, PS contains two chiral centers, one at the 2-sn-position and the other in the serine moiety. Consequently D-PS and L-PS are diastereomers and not enantiomers much like the glycerol-based glycolipids that have multiple chiral centers in the glycerol and sugar moieties. L-(C18:1)PS did not support refolding although it presumably favors a lamellar organization and contains a primary amine. Furthermore, the unnatural diastereomers at either chiral center did not support refolding, but all derivatives of PS were effective in binary mixtures with PG. Previous results have noted different effects of chirality between mixtures versus single systems. Maximal activation of protein kinase C requires the synergistic action specifically of PS (natural diastereomer), Ca 2ϩ , and diacylglycerol. This activation displays no stereoselectivity or strict chemical specificity within the head group in the absence of diacylglycerol; i.e. P-L-S, P-D-S, and even PG are equally effective but at a reduced level in the absence of diacylglycerol (37). 2 H and 31 P NMR studies have revealed small differences in the head group conformations of P-L-S and P-D-S (38). It was suggested that such structural differences might affect packing interactions and alter the effective area individual lipids occupy in respective single component bilayers of these diastereomers (39) that is not apparent in binary mixtures. Finally, the natural 1,2-sndiastereomer of the glycolipid monoglucosyl didodecylglycerol has a higher propensity to form the lamellar phase than the unnatural 2,3-sn-diastereomer (40) that adopts the H II phase. Moreover, the 1,2-rac mixture of this glycolipid displayed phase properties very similar to the natural diastereomer, indicating that mixtures of diastereomers are not simply additive in their final physical properties. All of these studies including our results support as biologically important the physical properties of the phospholipid macrostructure as well as the chemical properties of individual phospholipid species in biological processes.
The failure of PC to assist proper refolding of LacY assembled in the absence of PE is not surprising, since PC cannot support active transport of LacY in an in vitro reconstituted system (29,30). However, the fact that PC actively denatures or destabilizes epitope 4B1 during refolding of LacY originally assembled in the presence of PE now explains why native LacY does not display full function in PC-containing lipid vesicles. This result should be taken as a general caution in designing experiments involving membrane-associated proteins in which endogenous phospholipids are replaced by phospholipids not normally present in vivo.
Still puzzling is why anionic phospholipids in the absence of PE do not support full function in reconstituted systems even for LacY originally assembled in the presence of PE (14,29) and still maintaining epitope 4B1 (Fig. 3, lanes 1-3). Although previous evidence (8,9) strongly suggests that retention of epitope 4B1 after subjecting native LacY to Western blotting is not due to residual PE, the presence of epitope 4B1, although required, may not be sufficient to support full LacY function in active transport in the absence of PE. Our results cannot rule out that PE per se is required for either maintenance of proper structures of domains other than 4B1 or is directly involved in net turnover of the permease required for energy-dependent active transport of substrate. The molecular basis for this PE requirement might be direct involvement in the mechanism of coupling of uphill movement of substrate with downhill proton movement. The ionizable phosphoamino group of PE may be located in close proximity to functionally important residues such as Glu 269 of transmembrane helix VIII (41) or the salt bridge (Asp 240 -Lys 319 ) between transmembrane helices VII and X (42) and could alter their pK a values, thus affecting net movement of substrate (4).
There is growing evidence that phospholipids can facilitate the folding of membrane proteins or proteins that interact with membranes as part of their biogenesis. CL may selectively influence refolding of rhodanase as it passes through the mitochondrial membrane without the aid of protein chaperones (43)(44)(45). OmpA, an outer membrane protein of E. coli, was unfolded in 8 M urea without detergent, and refolding was initiated by rapid dilution of urea in the presence of PC vesicles that were required for regeneration of its native structure (46). Recent reports of specific effects of PG on folding of E. coli periplasmic DegP protein (47) and of the glycolipid portion of lipopolysaccharide on the assembly of E. coli outer membrane protein PhoE (36,48) indicate that the phenomenon of lipidassisted folding may be widespread and general among proteins that come in contact with membranes during their assembly. Heightened awareness of lipid-assisted folding of proteins hopefully will stimulate other investigators in designing experiments to investigate this process in many other systems.