Plasticity of - interactions in the function and topogenesis of the membrane protein lactose permease from Escherichia coli

Mikhail Bogdanova,1, Philip Heacocka, Ziqiang Guanb, and William Dowhana,1

aDepartment of and , University of Texas Medical School at Houston, Houston, TX, 77030; and bDepartment of Biochemistry, Duke University Medical Center, Durham, NC 27710

Edited by William T. Wickner, Dartmouth Medical School, Hanover, NH, and approved July 15, 2010 (received for review May 9, 2010)

Phosphatidylcholine (PC) has been widely used in place of naturally N-terminal two-TM helical hairpin of PheP and GabP are in- occurring phosphatidylethanolamine (PE) in reconstitution of verted with respect to their orientation in PE-containing cells bacterial membrane . However, PC does not support native and the remaining TMs. These permeases do not carry out pro- structure or function for several reconstituted transport proteins. ton-coupled energy-dependent uphill transport of substrate in Lactose permease (LacY) of Escherichia coli, when reconstituted cells lacking PE, but still display energy-independent downhill in E. coli phospholipids, exhibits energy-dependent uphill and transport. LacY reconstituted into total E. coli phospholipids car- energy-independent downhill transport function and proper con- ries out uphill transport with domains C6 and P7 on opposite formation of periplasmic domain P7, which is tightly linked to sides of the membrane bilayer as observed in wild-type cells uphill transport function. LacY expressed in cells lacking PE and (9). Leaving out PE during reconstitution results in only downhill containing only anionic phospholipids exhibits only downhill trans- transport with domains C6 and P7 residing on the same side of port and lacks native P7 conformation. Reconstitution of LacY in the bilayer as observed in PE-lacking cells (9). Reconstitution E. coli the presence of -derived PE, but not dioleoyl-PC, results in into proteoliposomes where dioleoyl-PC (diC18∶1PC) replaces uphill transport. We now show that LacY exhibits uphill transport PE results in wild-type topology and downhill transport (9) but

and native conformation of P7 when expressed in a mutant of not uphill transport (9–11). An ionizable amine or hydrogen bond BIOCHEMISTRY E. coli in which PC completely replaces PE even though the structure donor capacity is also required for uphill transport because is not completely native. E. coli-derived PC and synthetic PC species of the amine of PE progressively reduces activity containing at least one saturated fatty acid also support the native (10, 11). PE but not diC18∶1PC or eukaryotic-derived PC also sup- conformation of P7 dependent on the presence of anionic phos- ports uphill transport by the multidrug transporter (LmrP) of pholipids. Our results demonstrate that the different effects of Lactococcus lactis (12, 13), the leucine permease of Pseudomonas PE and PC species on LacY structure and function cannot be ex- aeruginosa (14), the branched chain transporter of plained by differences in the direct interaction of the lipid head Streptococcus cremoris (15), the ABC transporter HorA from groups with specific amino acid residues alone but are due to more Lactococcus lactis (16), and the Ca2þ ATPase of the sarcoplasmic complex effects of the physical and chemical properties of the lipid reticulum (17). environment on protein structure. This conclusion is supported by Another feature of LacY that is strongly correlated with uphill the effect of different on the proper folding of domain P7, transport is the structure of the extramembrane domain P7 (3) which indirectly influences uphill transport function. that is exposed to the periplasm and connects TMs VII and VIII. The conformation-specific monoclonal (mAb) 4B1 ver 35% of all proteins are integrated into cell membranes, recognizes this domain in membranes and on Western blots of Oand many others interact at the membrane surface. There- LacY that has been assembled in PE-containing but not in fore, determining how lipid-protein interactions govern protein PE-lacking E. coli membranes (7). Binding of mAb 4B1 also in- structure and function is fundamental to understanding cellular hibits uphill transport by LacY (2, 3). recognition is re- processes at the molecular level. Although reconstitution of stored to LacY from PE-lacking membranes by exposure to PE purified membrane proteins with lipids is necessary to study but not diC18∶1PC blotted to the same solid support prior to lipid-protein interactions, the complexity of lipid molecular forms Western analysis (8). Recognition by mAb 4B1 appears to that make up the lipidome has largely been ignored in most be a reliable indicator of the proper topological and structural studies, which have employed lipid mixtures that rarely reflect organization of LacY in the vicinity of domain P7 (Fig. 1) and the composition of host membranes. of LacY uphill transport function. Phosphatidylcholine (PC) has been extensively used in recon- The inability of PC to support the full function of several trans- stitution studies because it spontaneously organizes into bilayers. port systems as well as the native structure of domain P7 of LacY However, use of PC for reconstitution of membrane proteins raises several questions of broad significance to studies of from bacterial sources that do not contain PC may not provide membrane protein structure, function, and lipid interaction in re- reliable information, because there are no data on the in vivo constituted systems. Is lack of functional and structural support function of these proteins in membranes where PC replaces the major phospholipid phosphatidylethanolamine (PE). Therefore, a distinction between a biological preference for PE over PC and Author contributions: M.B., P.H., and W.D. designed research; M.B., P.H., and Z.G. an artifact in a reconstituted system cannot be made. performed research; P.H. and Z.G. contributed new reagents/analytic tools; M.B. and W.D. analyzed data; and M.B. and W.D. wrote the paper. The topological orientation and function of the 12 transmem- brane (TM) domain-spanning membrane integrated permeases The authors declare no conflict of interest. for lactose (LacY), phenylalanine (PheP), and γ-aminobutyrate This article is a PNAS Direct Submission. (GabP) of Escherichia coli are dependent on membrane lipid 1To whom correspondence may be addressed at: 6431 Fannin Street, Suite 6.200, Department of Biochemistry and Molecular Biology, University of Texas Medical School composition (1). In mutants lacking PE (normally ca. 70% of at Houston, Houston, TX, 77030. E-mail: [email protected] or William. total phospholipid) and containing only anionic phospholipids [email protected]. [primarily phosphatidylglycerol (PG) and cardiolipin (CL)], the This article contains supporting information online at www.pnas.org/lookup/suppl/ N-terminal six TM helical bundle of LacY (see Fig. 1) and the doi:10.1073/pnas.1006286107/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1006286107 PNAS Early Edition ∣ 1of6 Downloaded by guest on October 2, 2021 H205 Legionella pneumophila pcsA gene (22) under induction control NT CT by arabinose of the promoter P . PC is synthesized from Q340 araB L5 C4 C6 R285 endogenous CDP-diacylglycerol and choline from the growth G71 C2 R135 C10 medium. The L. pneumophila gene was selected because the Si- G13 C8 L14 norhizobium meliloti or Bradyrhizobium japonicum genes (22) were unstable in E. coli. Full induction by arabinose and supple- mentation of the medium with choline resulted in a level of PC I II III IV V VIVII VIII IX X XI XII (70%, see Fig. S1) equivalent to the level of PE (70%) in wild- type cells (1). The remaining lipids were primarily CL (27%) and

F PG (3%) with the former being higher and the latter being lower F* R than that found in wild-type cells (usually ca. 20–25% PG and I103 T163 T T A312 G* P11 – P1 P3 P5 S P7 P9 5 10% CL). Cells lacking PE contain equal amounts of PG S41 250 F* Q FATG*E and CL (lane 1 of Fig. S1). Mass spectral analysis of total phos- þ − þ 4B1 pholipid extracts from PE and PE PC cells revealed that CT PE and PC of E. coli exhibit a near identical spectrum of species with respect to fatty acid composition (Fig. S2). PE-lacking cells P1 P3 P5 C10 C8 require supplementation with tryptophan for growth in minimal medium due to a defect in uphill transport by aromatic amino acid transporters (23). This defect was corrected by introduction of GlcDAG (23) into −PE mutants and was also corrected by I II III IV V VI VIII IX X XI XII introduction of PC. However, PC did not relieve the requirement for Mg2þ in the growth medium.

P9 Transport Function of LacY in −PE þ PC Cells. VII P11 We analyzed transport C2 function of wild-type LacY expressed from a plasmid in strain C4 AL95 without additional plasmids (−PE) or with either plasmid NT pssAþ þ − þ P7 pDD72 ( , PE) or plasmid pAC-PCSlp-Sp-Gm ( PE C6 PC). Choline and Mg2þ were present in all growth media. Con- − Fig. 1. Topological organization of LacY as a function of membrane lipid sistent with previous results (18), LacY assembled in PE cells composition. TMs (Roman numerals), extramembrane domains (P for peri- was unable to catalyze uphill transport of the nonmetabolizable plasmic and C for cytoplasmic as in þPE cells), N-terminus (NT) and C-terminus substrate methyl-β-D-galactopyranoside (TMG) (Fig. 2A). Sur- (CT) domains are indicated. The approximate positions and names of amino prisingly −PE þ PC cells showed the same rate and level of uphill acids substituted by cysteine and used for topological analysis are indicated transport as þPE cells (Fig. 2A), which were confirmed by inhibi- near the closed and open symbols in a cysteineless derivative of LacY used in tion of accumulation of TMG by a protonophore (Fig. 2A). Sub- – this work. The conformation-sensitive (2, 3), PE-dependent (4 8) epitope in strate entry measured in −PE þ PC cells was via LacY as domain P7 recognized by mAb 4B1 is noted with the amino acids that are 95∕ part of this epitope marked with an asterisk. (Top) Denotes orientation in evidenced by lack of TMG and lactose uptake by AL pAC- þPE cells; (Bottom) denotes orientation in −PE cells with the upper side of PCSlp-Sp-Gm without a plasmid copy of LacY. Comparative each figure being the side that faces the cytoplasm. TM VII resides in the periplasm in −PE cells. Topological orientation of LacY is taken from ref. 1. 8 8 A B an artifact of in vitro reconstitution or the use of an inappropriate PC species? Can PC replace PE in vivo with respect to structure 6 6 and function of LacY? Are conclusions thus far made concerning lipid-protein interactions based on the ineffectiveness of PC valid? To address these questions we report on the orientation 4 4 of TMs, transport function, and the recognition by mAb 4B1 after assembly of LacY in a mutant of E. coli where PC replaces PE. 2 2 –PE +PE nmole TMG/mg protein +PC

Results nmole lactose/mg protein Substitution of PE by PC in E. coli Cells. Previous results demon- strated that LacY either expressed in cells or reconstituted into 0 0 proteoliposomes showed a near absolute dependence for uphill 86420 0 8642 transport activity, but not downhill transport activity, on an ioniz- Time (min) Time (min) able amino-based zwitterionic phospholipid [PE or phosphatidyl- Fig. 2. Transport function of wild-type LacY in whole cells. Uphill transport serine (PS)] (9, 10, 18, 19). LacY expressed in cells in which the of TMG (A) or downhill transport of lactose (B) normalized to total cell pro- neutral glycolipid monoglucosyl diacylglycerol (GlcDAG) (20) tein was determined as a function of time by AL95/pDD72 (closed squares, þ 95∕ − þ but not diglucosyl diacylglycerol (GlcGlcDAG) (21) replaced PE PE), AL pAC-PCSlp-Sp-Gm (open squares PE PC), or AL95 (open circles −PE), all induced for expression of a plasmid-borne copy of the wild-type lacY also carried out uphill transport. However, diC18∶1PC alone or in gene. Uphill transport in −PE þ PC cells was verified by addition of a proto- combination with PG and CL supported downhill but not uphill nophore (carbonyl cyanide p-trifluoromethoxyphenylhydrazone) to the transport of LacY in proteoliposomes (9, 10). of TMG uptake (open diamonds), which reduced the uptake to the level of To rule out an artifact introduced by reconstitution conditions the −PE results (open circles). No uptake of TMG or lactose by AL95∕pAC- − þ Δ and to definitively determine whether or not PC can substitute for PCSlp-Sp-Gm (closed diamonds, PE PC LacY) demonstrated that uphill PE in supporting LacY function, we utilized an E. coli strain in and downhill transport in −PE þ PC cells (open squares) were mediated by which PC replaces PE. Strain AL95 (pss93∷kanRlacY∷Tn9, −PE LacY. Standard deviations for the average of two determinations are shown except for lines where uptake was essentially zero (open diamonds, open cir- cells) is devoid of PE and PS and contains only the negatively cles, and closed diamonds). The reduced level (one-fifth) of LacY in −PE þ PC charged major lipids, PG and CL (6). Introduction of plasmid cells relative to the level of LacY in þPE cells (25 μg of protein per lane) is − þ pAC-PCSlp-Sp-Gm ( PE PC cells) allows expression of the shown by analysis (BInset) using anti-LacY polyclonal antibody.

2of6 ∣ www.pnas.org/cgi/doi/10.1073/pnas.1006286107 Bogdanov et al. Downloaded by guest on October 2, 2021 rates of uphill transport are difficult to assess especially with mul- Conformation of Functionally Important Domain P7. The native con- tiple copies of lacY because 10% induction of single copy lacY formation of an epitope within periplasmic domain P7 (Fig. 1), as results in 50% of the uphill transport rate after full induction of determined by the conformation-specific mAb 4B1, is strongly single copy lacY (24). However, the rate of downhill substrate correlated with LacY uphill transport (3, 4, 6, 7). Binding of mAb transport was significantly lower than that observed in þPE cells 4B1 to LacY inhibits uphill transport (2, 3). Consistent with uphill (Fig. 2B), which appears to be due to a 5-fold decrease in LacY transport function, LacY assembled in −PE þ PC and þPE cells þ relative to that in PE cells as detected by Western blot analysis (Fig. 4A, right and center lanes, respectively) was strongly recog- (Fig. 2B Inset); the rate of downhill transport by LacY is propor- nized by mAb 4B1 after SDS PAGE and Western blotting. LacY tional to the number of carriers in the membrane (24). Similar assembled in −PE cells (left lane) was very weakly recognized transport results were observed using the H205C derivative of if at all. LacY (Fig. S3).

Topological Orientation of TMs of LacY in −PE þ PC Cells. LacY with A cysteine replacements in otherwise cysteineless LacY (Fig. 1) was Cell Source –PE +PE –PE+PC expressed in −PE þ PC cells and reactivity with 3-(N-maleimidyl- propionyl) biocytin (MPB) in intact cells (periplasmic exposed) or during cell disruption by sonication (cytoplasmic and periplas- B mic exposed) was used to establish TM orientation based on Preblotted None –PE+PC diC18:1PC None derivatization of cysteines in flanking extramembrane domains (25, 26). As shown in Fig. 3, cysteines residing within normal cytoplasmic domains C2, C4, C6, C8, or C10 were labeled only Cell Source +PE –PE after cell disruption indicating cytoplasmic exposure, whereas cysteines in normal periplasmic domains P1, P7, or P9 were la- C diC18:1PC beled whether or not cells were disrupted indicating periplasmic Preblotted +PE +PG+CL –PE+PC EcPC exposure in these cells. However, the cysteines residing within normal periplasmic domains P3 or P5 were not accessible to MPB at pH 7.5. Lack of can be due to location within a

Cell Source –PE BIOCHEMISTRY TM or proximal environmental effects, which affect thiol pKa or accessibility (26). Increasing solution pH should favor alkylation D EcPC of an extramembrane cysteine or disruption of local secondary Preblotted –PE+PC None EcPC +PG+CL structure without exposure of cysteines in TMs (27). P3 and P5 cysteines were fully labeled at pH 9.1 without sonication, indicating periplasmic exposure. The H205C substitution in Cell Source –PE +PE –PE domain C6 remained inaccessible to MBP at pH 9.1 without so- nication, which verified retention of cell impermeability to MBP 16:1t 16:1c None 16:0/18:1 16:0/18:1 E Alone (Fig. S4B). None of the cysteines in the normal cytoplasmic PG+CL PG+CL PG+CL N-terminal (NT) domain were labeled by exposure to pH 7.5, –PE 9.1, or 10.5 either before or during sonication, which indicates Cells 12 34 5 a significant departure from LacY expressed in þPE cells where None 16:0/18:1 18:1/16:0 16:0 labeling occurs at pH 7.5 (6). Presence of sufficient amounts of PG+CL PG+CL PG+CL LacY with the L5C substitution to be detected by MBP labeling was shown using Western blot analysis with polyclonal antibody –PE Cells (Fig. S4A). Although the disposition of domains NT and TM I 67 8 9 remains uncertain, aberrant organization of domains NT through TM II was previously shown not to affect uphill transport (27). Fig. 4. Western and Eastern-Western blotting of LacY used to probe the con- Thus the gross topology of LacY assembled in −PE þ PC cells formation of domain P7. (A) Inner membrane preparations (duplicate sam- ples of 10 μg of protein) from cells with the indicated lipid composition and starting at least from domain P1 is the same as in wild-type cells expressing a plasmid-borne copy of wild-type LacY were analyzed by Western and is consistent with topology determined after reconstitution of blot analysis using conformation-specific mAb 4B1. (B–E) Inner membrane LacY into −PE þ diC18∶1PC proteoliposomes containing PG and preparations from cells with the indicated phospholipid composition and ex- CL that did not support uphill transport (9). pressing a plasmid-borne copy of wild-type LacY were subjected to SDS PAGE. The region of the gel containing LacY (molecular weight 31,000–33,000) was transferred to the area of a sheet onto which the indicated Loop/AA NT C2 C4 C6 C8 C10 phospholipids were transferred from a TLC plate () followed L5 G71 R135 H205 R285 Q340 by Western blotting analysis using mAb 4B1. Single samples contained 10 μg of membrane protein and duplicate samples contained 5 (Left)or Son –+–+–+–+–+–+ 10 (Right) μg of membrane protein. Preblotted phospholipids were as fol- Loop/AA P1 P3 P5 P7 P9 lows: (B) no lipid (none), total lipid extract from −PE þ PC cells, diC18∶1PC S41 I103 T163 F250 A312 (Avanti); (C) PE (Avanti) purified from E. coli (þPE), 70% diC18∶1PC (Avanti) supplemented with 30% of total lipid extract from −PE cells containing ap- Son – +–+ –+–+–+ proximately equal amounts of PG and CL, total lipid extract from −PE þ PC Fig. 3. Mapping the topology of LacY in −PE þ PC cells. All samples were cells, PC isolated from −PC þ PE cells by TLC (EcPC); (D) total lipid extract from treated with MPB without sonication (−) or during sonication (þ). AL95∕pAC- −PE þ PC cells, PC isolated from −PC þ PE cells by DE52 chromatography − þ PCSlp-Sp-Gm ( PE PC) cells expressing single cysteine replacements in the either alone (EcPC) or 70% PC supplemented with E. coli PG (3%) and CL indicated extramembrane loop/amino acid were grown and processed as de- (27%) from Avanti (EcPC þ PG þ CL); (E) preblotted PCs (Avanti) with the in- scribed in Methods subsequent to Western blotting using avidin-HRP to de- dicated fatty acid composition are shown. One fatty acid denotes both acyl tect the biotin moiety linked to cysteines (see Fig. 1 for approximate positions chains are identical. Two fatty acids indicate fatty acids at positions sn-1 and of substitutions) that were accessible to MPB during labeling. MBP labeling sn-2, respectively. PCs were supplemented with E. coli PG plus CL (Avanti) in was performed at pH 7 except for P3/I103C and P5/T163C, which were labeled the ratio as indicated for D. None indicates no lipid preblotted. Fatty acids for at pH 9.1. lanes 1 and 2 were Δ9 trans and Δ9 cis, respectively.

Bogdanov et al. PNAS Early Edition ∣ 3of6 Downloaded by guest on October 2, 2021 Previously, we showed that E. coli-derived PE or total E. coli dependent on divalent cation for viability consistent with the pos- phospholipids when blotted to the surface of nitrocellulose sheets tulated role of divalent cations in inducing the nonbilayer phase (Eastern blot) prior to transfer by (Western blot) for CL (which was highly elevated compared to wild-type cells), of LacY from SDS polyacrylamide gels restored recognition by the only potentially non-bilayer-prone lipid in −PE cells (29). mAb 4B1 of LacY derived from −PE cells (8). DiC18∶1PC with Tryptophan prototrophy of −PE þ PC cells suggests that PC cor- or without PG and CL supplementation showed very weak or rects the uphill transport defect in other secondary transporters no restoration of mAb 4B1 recognition and even resulted in loss (1). Topological organization of LacY was near wild type, which of recognition for LacY initially made in þPE cells (8). However, extended the in vitro results where only the orientation of the recognition by mAb 4B1 of LacY synthesized in −PE cells was C6-TM VII-P7 domain of LacY was probed (9). However, restored in the Eastern-Western blotting procedure using total cysteines in periplasmic domains were less accessible to MPB, phospholipids from −PE þ PC cells (Fig. 4B), which is consistent and the NT domain was completely inaccessible, indicating in- with observed support of uphill transport. However, diC18∶1PC complete native structure. LacY exhibited downhill transport, alone (Fig. 4B) or supplemented with PG and CL (Fig. 4C) but contrary to in vitro results also exhibited uphill transport, showed little or no restoration of recognition by mAb 4B1. To which correlated with proper folding of domain P7 as determined rule out factors other than lipids in these extracts, PC made in by mAb 4B1. PC containing at least one saturated fatty acid, but E. coli was isolated by DE52 column chromatography or after not diC18∶1PC as previously shown (8), in the presence of PG or separation by TLC. As shown in Fig. 4 C and D, purified E. coli CL restored the conformation of domain P7 of LacY assembled PC alone was very weak in supporting proper refolding. However, in PE-lacking cells. this lipid supported proper refolding when mixed with purified PE, PC, GlcDAG (20), and GlcGlcDAG (21) are mostly inter- E. coli PG plus CL (Avanti Polar Lipids) at a ratio mimicking changeable in their ability to support near native topological the phospholipid composition of −PE þ PC cells (Fig. 4D). Syn- organization of LacY. These lipids vary in size, structure, charge, thetic diC18∶1PC weakly supported proper refolding either used and hydrogen-bonding properties of their head groups as well as alone or in ternary mixtures with E. coli PG plus CL (Fig. 4 B or in their effects on the physical properties of the membrane. The C, respectively). common feature among these lipids is a net neutral head group, Next LacY expressed in −PE cells was subjected to Eastern- which has been proposed to dilute the high negative surface Western blotting using synthetic PCs with different fatty acid charge density contributed by PG and CL that results in aberrant compositions. As shown in Fig. 4E, synthetic C16∶0-C18∶1-PC topological organization of LacY, PheP, and GabP in the absence alone (lane 4) was weak in assisting proper refolding of domain of PE (1, 27). PE and GlcDAG are nonbilayer, whereas PC and P7 unless mixed with E. coli PG plus CL (Avanti Polar Lipids) at a GlcGlcDAG are bilayer-prone lipids, and only the head group of ratio mimicking the phospholipid composition of −PE þ PC cells PC cannot act as a hydrogen bond donor, which indicates these (lane 5). DiC16∶0PC (Fig. 4E, lane 9) and diC16∶1Δ9-transPC properties are not important in supporting native topology. (Fig. 4E, lane 1) supported refolding of domain P7 if mixed with The most surprising result was that LacY carried out uphill E. coli PG plus CL, whereas diC16∶1Δ9-cisPC did not (Fig. 4E, lane transport when expressed in cells in which PE was replaced by 2). In order to test the acyl chain positional specificity of an equivalent amount of PC contrary to in vitro results (9–11). phospholipid-assisted refolding of domain P7, C16∶0-C18∶1PC GlcDAG (20) but not GlcGlcDAG (21) also supports uphill and C18∶1-C16∶0PC were compared (Fig. 4E, lanes 7 and 8, respec- transport in vivo. PE is interchangeable with glycolipids in sup- tively) and were shown to be equally effective. CL or PG alone porting uphill transport for the branched chain amino acid trans- 2þ was as effective as PG plus CL in supporting refolding in the porter of S. cremoris (15) and the eukaryotic Ca ATPase (17). presence of C16∶0-C18∶1PC (Fig. S5A), and the latter with anionic These results demonstrate that the ethanolamine head group of lipids was as effective as lipids from −PE þ PC cells or E. coli PE PE is not specifically required for uphill transport carried out by alone (Fig. S5B). several transporters but does not explain why PC does not sub- stitute for PE in reconstituted systems. Discussion The conformation of the P7 domain as indicated by recogni- A major approach to understanding how lipid-protein interac- tion with mAb 4B1 is directly correlated with the ability of LacY tions govern cell function is to study purified membrane proteins to carry out uphill transport that requires protonation of the glua- reconstituted with a variety of natural or foreign lipids differing in mate carboxyl at position 325, which displays a pKa above 9 (30). physical and chemical properties. A limitation of this approach Binding of mAb 4B1 to LacY inhibits uphill transport in spher- has been the inability to verify in vitro observations in an in vivo oplasts and in proteoliposomes (31) concomitant with conforma- system. Development of E. coli strains in which membrane lipid tional changes in several TMs (32). An E325D substitution composition can be systematically manipulated (1) now makes prevents uphill transport and reduces the rate of substrate ex- this possible. A primary aim of this study was to determine the change above pH 8.5 instead of above 9.5 as observed with E325 molecular basis for the ability of the foreign lipid PC to support (33). Binding of mAb 4B1 reduces these exchange rates at one pH native topological organization and downhill transport (9) but not unit lower for the E325D mutant (30). Therefore structural uphill transport (9–11) or the structure of domain P7 of E. coli changes within P7 induced by binding of mAb 4B1, and by extra- LacY (8). The differences between PE and PC in supporting polation folding defects in P7 due to lack of PE, result in a con- function of several secondary transport proteins has focused formational change that lowers the pKa of an essential acidic on the mode of direct interaction between the lipid head groups residue at position 325, which is involved in proton-coupled uphill and specific amino acids of the mature protein during catalysis transport. These results are consistent with studies on LmrP in (12, 13, 16, 28), which, based on our results, appears not to be which the average pKa of acidic residues was 6.5 when reconsti- the case. Our results suggest that the major effect of the lipid tuted in the presence of PE but only 4.5 in the presence of egg environment occurs during folding of the protein into its final yolk PC (12), which does not support uphill transport. There were structure (i.e., P7 conformation) rather than specific lipid-protein also pH-dependent conformational changes associated with interactions after native structure is attained. acidic group ionization in PE but not in PC proteoliposomes. Biosynthesis of PC in an E. coli mutant lacking PE resulted in Using the Eastern-Western procedure, we demonstrated that restoration of the wild-type zwitterionic and anionic phospholipid lipid mixtures that support uphill transport were able to restore levels to 70% and 30%, respectively. The PC species displayed a native conformation of the P7 domain, as demonstrated by regain similar fatty acid distribution as found in PE, which is significantly of recognition by mAb 4B1 of LacY assembled in −PE cells. different from that used in reconstitution studies. Cells remained Strong recognition was observed using total lipids from

4of6 ∣ www.pnas.org/cgi/doi/10.1073/pnas.1006286107 Bogdanov et al. Downloaded by guest on October 2, 2021 −PE þ PC cells and purified E. coli-derived PC reconstituted protein folding, current results do not support such an interaction with E. coli PG plus CL, whereas weak or no recognition was ob- directly in support of native transport function. Therefore, the served with diC18∶1PC in all lipid mixtures or with E. coli-derived molecular basis for lipid requirements by LacY is more complex PC alone. Further analysis using PCs with different fatty acid than previously thought and is not based solely on the nature of compositions showed that reconstitution of the epitope occurred the lipid head group but involves the physical and chemical with PCs (always in the presence of E. coli PG and CL) containing properties of a multicomponent membrane bilayer. Although at least one saturated fatty acid or Δ9 trans fatty acids; cis unsa- E. coli PC supports LacY function, full native structure is still turated fatty acids impart distinctly different physical properties not attained, emphasizing the need to use native lipid composi- to phospholipids as compared to saturated or trans unsaturated tion in reconstituting membrane proteins. The availability of fatty acids. Unlike E. coli PC, bilayer-prone PE species (E. coli PE mAb 4B1 has demonstrated that different lipid environments and PE species with saturated fatty acids) alone as opposed to have a dramatic effect on the structure of protein domains non-bilayer-prone PE (those with two unsaturated fatty acids) and that use of nonnative lipid mixtures can seriously compro- alone supported epitope formation (8). Mixing of non-bilayer- mise structural studies on membrane proteins. Our results also prone PE species with bilayer-prone PG or CL supported epitope point out the shortcomings in current molecular simulations of formation. Although functional PC species are all bilayer prone, lipid bilayers that include only single lipid species and not more PG or CL was still required to support epitope formation. The complex lipid mixtures. missing component contributed by the anionic lipids might be the ability of their head groups to be hydrogen bond donors in Methods the case of PC or provide an overall bilayer phase in the case Additional procedures and results can be found in SI Text.

of unsaturated non-bilayer-prone PE species. The requirement 3 32 Reagents. TMG ([6- H]methyl) came from Moravek Biochemicals. ½ PPO4 for PG or CL appears to be their anionic or hydrogen-bonding and lactose [D-glucose-1-14C] came from American Radiolabeled Chemical properties rather than specific structural features (Fig. S5A), Inc. Polyclonal antibody directed against the C terminus of LacY was made which indicates that the high CL content of −PE þ PC cells is by ProSci Inc. Avidin conjugated with HRP (avidin-HRP) and Super-Signal West not a factor. Cells in which GlcDAG replaced PE display the na- Picochemiluminescent substrate was from Pierce. Pansorbin cells were from tive conformation of the P7 domain (Fig. S6) and carry out uphill Calbiochem, and Lubrol (type-PX) was from Nacalai Tesque, Inc.. MPB was ∕ transport by LacY (20). GlcGlcDAG (Fig. S6) only weakly sup- purchased from Invitrogen-Molecular Probes. RbCl CaCl2 Transformation ports the P7 native conformation and does not support uphill Salts were purchased from MP Biomedicals, USA. H. R. Kaback (UCLA) pro- vided mAb 4B1. Nitrocellulose sheets for immunoblotting were purchased

transport (21), which may be due to its larger head group. BIOCHEMISTRY from Schleicher and Schuell. Silica gel TLC plates were purchased from Merck. The lack of phospholipid head group specificity and the varia- E. coli PE, PG, and CL and the following synthetic PCs containing the indicated bility with respect to fatty acid composition in support of uphill fatty acids were purchased from Avanti Polar Lipids: 1,2-dioleoyl- (diC18∶1PC), transport and the structure of the P7 domain strongly indicate 1,2-dipalmitoyl- (diC16∶0PC), 1-palmitoyl-2-oleoyl- (C16∶0-C18∶1PC), 1-oleoyl-2- that specific lipid head group protein interactions in the folded palmitoyl- (C18∶1-C16∶0PC), 2-dipalmitoleoyl- (diC16∶1Δ9-cisPC), and 1,2-dipalmi- state are not the defining lipid requirement. Rather the attain- telaidoyl- (diC16∶1Δ9-transPC) sn-glycero-3-phosphocholine. ment of native structure of LacY facilitated during folding by the proper physical and chemical properties of the lipid environ- Bacterial Strains and Plasmids. LacY derivatives containing single amino acids ment is of primary importance. Native folding of domain P7 (re- replaced by cysteine in a derivative of LacY (see Fig. 1) in which natural sulting in recognition by mAb 4B1), which has long-range effects cysteines were replaced by serine (38) were constructed by site-directed mu- on structure, appears to occur by transient interaction with a sub- tagenesis (39). All derivatives exhibited 60% or more of wild-type transport activity in þPE cells (38). Native LacY and the above LacY derivatives were set of lipid species (now including certain species of PC) during a R OP carried on plasmid pT7-5 (Amp ) and expressed under tac regulation as late step of initial assembly or postassembly (4, 6). Conforma- previously described (27). tional memory within P7 remains after SDS treatment and Strain AL95 (pss93∷kanRlacY∷Tn9) with (þPE) or without (−PE) plasmid removal of SDS and PE during electroblotting, which indicates pDD72 (pssAþ camR) was grown as previously described (6). Strain AL95 car- P R R a transient molecular chaperone role for lipid during protein rying plasmid pAC-PCSlp-Sp-Gm ( araB-pcs Sp Gm ) (see SI Text) was capable folding (5, 7, 8). The epitope recognized by mAb 4B1 consists of making PC instead PE (−PE þ PC) due to expression of PC synthase encoded of Phe-247, Phe-250, and Gly-254 (Fig. 1) on one face of a short by the Legionella pneumophila pcsA gene under control of an arabinose in- P α-helical segment (3). Although PE, PC, and GlcDAG have di- ducible promoter ( araB). Cells competent for transformation were prepared using RbCl∕CaCl2 Transformation Salts as described in SI Text. Cells were verse structural and chemical properties, they share the potential π grown in LB broth containing 50 mM MgCl2 (although only required by to interact through bonding with the face of aromatic residues. −PE and −PE þ PC cells) supplemented with ampicillin (100 μg mL) when Primary (PE) and quaternary (PC) amines have been observed to LacY-expressing plasmids were present and at either 30 °C for cells containing engage in π-bonding with the indole ring of Phe in the structure of pDD72, (which is temperature sensitive for replication and therefore PE cytochrome c oxidase (34) and within a cage formed by three aro- synthesis) or 37 °C for cells lacking pDD72 (unable to make PE). To express matic residues in the structure of human PC transfer protein (35). LacY in the presence of PC, overnight cultures of strain AL95∕pAC- Protein- interactions also rely on aromatic stacking PCSlp-Sp-Gm harboring plasmids carrying derivatives of LacY were diluted interactions with sugar rings (36). Interestingly the analogous P7 to OD600 of ca. 0.025 into 0.2% arabinose, 2 mM choline, and 1 mM isopro- pyl-ß-D-thiogalactoside and grown to a final OD600 of 0.5–0.7. The growth domain of LmrP is also rich in Phe (37) and may also be respon- requirement for tryptophan was determined by plating cells for single sible for transient interaction with the head group of PE during colonies on agar plates composed of minimal medium with or without folding. Taken together these results strongly suggest that the tryptophan (23). lipid environment affects the conformation of the P7 domain of LacY during initial folding, which has long-range indirect Transmembrane Protein Topology Mapping. Topological determination of TMs effects on critical acidic residues involved in proton-dependent is based on the controlled membrane permeability of the thiol-specific uphill transport of substrate. reagent MPB and its reactivity with diagnostic cysteine residues in extramem- Our results emphasize the necessity of verifying conclusions brane domains of LacY as previously described (25, 26). Briefly whole cells about lipid-protein interactions derived from in vitro observa- (periplasmically exposed but cytoplasmic and TM protected cysteines) or tions under in vivo conditions. The physical and chemical pro- sonicated cells (periplasmic and cytoplasmic exposed but TM cysteines pro- tected) were treated with MPB (containing biotin) at pH 7.5 or 9.1. After perties of the lipid bilayer determine the proper overall confor- membrane solubilization, polyclonal antibody precipitation of LacY, and mation of LacY that supports native transport. Although SDS PAGE, Western blot images (generated by Avidin-HRP) were acquired direct interaction between specific lipid head groups and protein using a Fluor-S Max MultiImager (Bio-Rad Laboratories) and processed as functional groups may be important, possibly transiently during described (27).

Bogdanov et al. PNAS Early Edition ∣ 5of6 Downloaded by guest on October 2, 2021 Eastern-Western Blotting. Inner membrane vesicles were prepared as de- Transport Assays. Transport of radiolabeled lactose (0.5 μCi∕ml) or TMG scribed previously (18), using a French press to break cells at 8,000 psi (0.1 μCi∕ml) at final concentration of 0.1 mM was assayed in intact cells har-

in the presence of 20 mM MgCl2. Inner membrane fractions were subjected boring the indicated plasmids as described previously (18). to SDS PAGE as described previously (18). Phospholipids (0.1–0.2 mg) were − þ first spotted in blocks on silica gel TLC plates from a chloroform/methanol Isolation of Phospholipids. Lipids were extracted from PE PC cells and (2∶1) solution, dipped in the blotting solvent of isopropyl alcohol/0.2% applied to a DE52 (Whatman) column as described earlier (9). E. coli PC was collected in the run-through fraction. Phospholipids were also separated aqueous CaCl2∕methanol (40∶20∶7,vol∕vol∕vol) for 60 s and then trans- by TLC using chloroform/methanol/acetic acid (65∶25∶10, vol∕vol∕vol) as ferred to nitrocellulose sheets (Eastern blotting) at 130 °C and pressure solvent, and E. coli PC was recovered by extraction with chloroform/methanol 8 for 60 s using a TLC Thermoblotter AC-5970 (ATTO Bio-Instrument) (2∕1,vol∕vol). (8). Proteins were transferred from the SDS gels to nitrocellulose sheets preblotted with phospholipids (phospholipid side facing the acrylamide ACKNOWLEDGMENTS. We thank Dr. H. R. Kaback for supplying mAb 4B1, Heidi gel) by electroblotting for 90 min using a semidry electroblotting system Vitrac for purification of used in this work, and Dr. Otto Geiger for (Labconco Semi-Dry blotting system, W.E.P, Company) as described pre- providing clones of the pcsA gene. The mass spectrometry facility in the viously (7, 8). By combining the Eastern with Western blotting, denaturated Department of Biochemistry of the Duke University Medical Center and Dr. Ziqiang Guan are supported by the LIPID MAPS Large Scale Collaborative proteins are exposed to hydrated phospholipids on the surface of the ni- Grant GM-069338 from National Institutes of Health (NIH). This work was trocellulose sheet as they exit the SDS gel and begin to refold as SDS is supported by NIH Grant R37 GM20478 and funds from the Dunn S. Dunn removed. Research Foundation awarded to W.D.

1. Dowhan W, Bogdanov M (2009) Lipid-dependent membrane protein topogenesis. function of the polytopic membrane protein lactose permease. J Biol Chem Annu Rev Biochem 78:515–540. 281:19172–19178. 2. Sun J, Frillingos S, Kaback HR (1997) Binding of monoclonal antibody 4B1 to homologs 21. Wikstrom M, et al. (2009) Lipid-engineered Escherichia coli membranes reveal critical of the lactose permease of Escherichia coli. Protein Sci 6:1503–1510. lipid headgroup size for protein function. J Biol Chem 284:954–965. 3. Sun J, Wu J, Carrasco N, Kaback HR (1996) Identification of the epitope for monoclonal 22. Martinez-Morales F, Schobert M, Lopez-Lara IM, Geiger O (2003) Pathways for antibody 4B1 which uncouples lactose and proton translocation in the lactose per- phosphatidylcholine biosynthesis in bacteria. Microbiology 149:3461–3471. mease of Escherichia coli. Biochemistry 35:990–998. 23. Wikstrom M, et al. (2004) Monoglucosyldiacylglycerol, a foreign lipid, can substitute 4. Bogdanov M, Dowhan W (1998) Phospholipid-assisted protein folding: Phosphatidy- for phosphatidylethanolamine in essential membrane-associated functions in Escher- lethanolamine is required at a late step of the conformational maturation of the ichia coli. J Biol Chem 279:10484–10493. polytopic membrane protein lactose permease. EMBO J 17:5255–5264. 24. Maloney PC, Wilson TH (1973) Quantitative aspects of active transport by the lactose 5. Bogdanov M, Dowhan W (1999) Lipid-assisted protein folding. J Biol Chem transport system of Escherichia coli. Biochim Biophys Acta 330:196–205. – 274:36827 36830. 25. Bogdanov M, Heacock PN, Dowhan W (2010) Study of polytopic membrane protein 6. Bogdanov M, Heacock PN, Dowhan W (2002) A polytopic membrane protein displays a topological organization as a function of membrane lipid composition. Protein reversible topology dependent on membrane lipid composition. EMBO J Secretion, Methods in Molecular Biology, ed A Economou (Springer, New York), – 21:2107 2116. Vol 619, pp 79–101. 7. Bogdanov M, Sun J, Kaback HR, Dowhan W (1996) A phospholipid acts as a chaperone 26. Bogdanov M, Zhang W, Xie J, Dowhan W (2005) Transmembrane protein topology – in assembly of a membrane transport protein. J Biol Chem 271:11615 11618. mapping by the substituted cysteine accessibility method (SCAM™): Application to 8. Bogdanov M, Umeda M, Dowhan W (1999) Phospholipid-assisted refolding of an lipid-specific membrane protein topogenesis. Methods 36:148–171. integral membrane protein. Minimum structural features for phosphatidylethanola- 27. Bogdanov M, Xie J, Heacock P, Dowhan W (2008) To flip or not to flip: Protein-lipid – mine to act as a molecular chaperone. J Biol Chem 274:12339 12345. charge interactions are a determinant of final membrane protein topology. J Cell Biol 9. Wang X, Bogdanov M, Dowhan W (2002) Topology of polytopic membrane protein 182:925–935. subdomains is dictated by membrane phospholipid composition. EMBO J 28. Lensink MF, Govaert C, Ruysschaert JM (2010) Identification of specific lipid-binding 21:5673–5681. sites in integral membrane proteins. J Biol Chem 285:10519–10526. 10. Chen CC, Wilson TH (1984) The phospholipid requirement for activity of the lactose 29. Rietveld AG, et al. (1994) Regulation of lipid polymorphism is essential for the viability carrier of Escherichia coli. J Biol Chem 259:10150–10158. of phosphatidylethanolamine-deficient Escherichia coli cells. J Biol Chem 11. Seto-Young D, Chen CC, Wilson TH (1985) Effect of different phospholipids on the re- 269:28670–28675. constitution of two functions of the lactose carrier of Escherichia coli. J Membr Biol 30. Frillingos S, Kaback HR (1996) Monoclonal antibody 4B1 alters the pKa of a carboxylic 84:259–267. acid at position 325 (helix X) of the lactose permease of Escherichia coli. Biochemistry 12. Gbaguidi B, Hakizimana P, Vandenbussche G, Ruysschaert JM (2007) Conformational 35:10166–10171. changes in a bacterial multidrug transporter are phosphatidylethanolamine- 31. Herzlinger D, Viitanen P, Carrasco N, Kaback HR (1984) Monoclonal antibodies against dependent. Cell Mol Life Sci 64:1571–1582. the lac carrier protein from Escherichia coli 2. Binding studies with membrane vesicles 13. Hakizimana P, Masureel M, Gbaguidi B, Ruysschaert JM, Govaerts C (2008) Interactions and proteoliposomes reconstituted with purified lac carrier protein. Biochemistry between phosphatidylethanolamine headgroup and LmrP, a multidrug transporter: 23:3688–3693. A conserved mechanism for proton gradient sensing? J Biol Chem 283:9369–9376. 14. Uratani Y, Aiyama A (1986) Effect of phospholipid composition on activity of sodium- 32. Frillingos S, Wu J, Venkatesan P, Kaback HR (1997) Binding of or monoclonal dependent leucine transport system in Pseudomonas aeruginosa. J Biol Chem antibody 4B1 induces discrete structural changes in the lactose permease of – 261:5450–5454. Escherichia coli. Biochemistry 36:6408 6414. 15. Driessen AJ, Zheng T, In’t Veld G, Op den Kamp JA, Konings WN (1988) Lipid require- 33. Herzlinger D, Carrasco N, Kaback HR (1985) Functional and immunochemical ment of the branched-chain amino acid transport system of Streptococcus cremoris. characterization of a mutant of Escherichia coli energy uncoupled for lactose trans- – Biochemistry 27:865–872. port. Biochemistry 24:221 229. 16. Gustot A, Smriti, Ruysschaert JM, Mchaourab H, Govaerts C (2010) Lipid composition 34. Shinzawa-Itoh K, et al. (2007) Structures and physiological roles of 13 integral lipids of – regulates the orientation of transmembrane helices in HorA, an ABC multidrug bovine heart cytochrome c oxidase. EMBO J 26:1713 1725. transporter. J Biol Chem 285:14144–14151. 35. Roderick SL, et al. (2002) Structure of human phosphatidylcholine transfer protein in 17. Navarro J, Toivio-Kinnucan M, Racker E (1984) Effect of lipid composition on the complex with its ligand. Nat Struct Biol 9:507–511. calcium/adenosine 5'-triphosphate coupling ratio of the Ca2þ-ATPase of sarcoplasmic 36. del Carmen Fernandez-Alonso M, Canada FJ, Jimenez-Barbero J, Cuevas G (2005) reticulum. Biochemistry 23:130–135. Molecular recognition of saccharides by proteins. Insights on the origin of the carbo- 18. Bogdanov M, Dowhan W (1995) Phosphatidylethanolamine is required for in vivo hydrate-aromatic interactions. J Am Chem Soc 127:7379–7386. function of the membrane-associated lactose permease of Escherichia coli. J Biol Chem 37. Bolhuis H, et al. (1995) The Lactococcal lmrP gene encodes a proton motive force- 270:732–739. dependent drug transporter. J Biol Chem 270:26092–26098. 19. Hawrot E, Kennedy EP (1975) Biogenesis of membrane lipids: Mutants of Escherichia 38. Frillingos S, Sahin-Toth M, Wu J, Kaback HR (1998) Cys-scanning mutagenesis: A novel coli with temperature-sensitive phosphatidylserine decarboxylase. Proc Natl Acad Sci approach to structure function relationships in polytopic membrane proteins. FASEB J USA 72:1112–1116. 12:1281–1299. 20. Xie J, Bogdanov M, Heacock P, Dowhan W (2006) Phosphatidylethanolamine and 39. Ho SN, Hunt HD, Horton RM, Pullen JK, Pease LR (1989) Site-directed mutagenesis by monoglucosyldiacylglycerol are interchangeable in supporting topogenesis and overlap extension using the polymerase chain reaction. Gene 77:51–59.

6of6 ∣ www.pnas.org/cgi/doi/10.1073/pnas.1006286107 Bogdanov et al. Downloaded by guest on October 2, 2021