Lysophospholipid flipping across the Escherichia coli inner membrane catalyzed by a transporter (LplT) belonging to the major facilitator superfamily.

The transfer of phospholipids across membrane bilayers is protein-mediated, and most of the established transporters catalyze the energy-dependent efflux of phospholipids from cells. This work identifies and characterizes a lysophospholipid transporter gene (lplT, formally ygeD) in Escherichia coli that is an integral component in the 2-acylglycerophosphoethanolamine (2-acyl-GPE) metabolic cycle for membrane protein acylation. The lplT gene is adjacent to and in the same operon as the aas gene, which encodes the bifunctional enzyme 2-acyl-GPE acyltransferase/acyl-acyl carrier protein synthetase. In some bacteria, acyltransferase/acyl-ACP synthetase (Aas) and LplT homologues are fused in a single polypeptide chain. 2-Acyl-GPE transport to the inside of the cell was assessed by measuring the Aas-dependent formation of phosphatidylethanolamine. The Aas-dependent incorporation of [3H]palmitate into phosphatidylethanolamine was significantly diminished in deltalplT mutants, and the LplT-Aas transport/acylation activity was independent of the proton motive force. The deltalplT mutants accumulated acyl-GPE in vivo and had a diminished capacity to transport exogenous 2-acylglycerophosphocholine into the cell. Spheroplasts prepared from wild-type E. coli transported and acylated fluorescent 2-acyl-GPE with an apparent K(d) of 7.5 microM, whereas this high-affinity process was absent in deltalplT mutants. Thus, LplT catalyzes the transbilayer movement of lysophospholipids and is the first example of a phospholipid flippase that belongs to the major facilitator superfamily.

Membrane phospholipids are critically important in maintaining the structure of biological membranes; in addition, portions of these lipids are used in the synthesis of other membrane-associated macromolecules. In Escherichia coli, the acyl moieties at the 1-position of PtdEtn 1 are metabolically active. The 2-acyl-GPE metabolic cycle begins with the transfer of the 1-position fatty acid of PtdEtn to the N terminus of the major outer membrane lipoprotein ( Fig. 1) (1). Topologically, the prolipoprotein is secreted through the inner membrane, and all of the subsequent processing steps, proteolytic cleavage (2), transfer of the diacylglycerol (3), and N-acylation (4,5), are located on the outer aspect of the inner membrane. The 2-acyl-GPE generated as a product of this reaction is re-acylated by the bifunctional enzyme 2-acyl-GPE acyltransferase/acyl-ACP synthetase (Aas) (6,7). The 2-acyl-GPE acyltransferase can either activate a free fatty acid with ATP-Mg 2ϩ and transfer it to a bound ACP subunit or interact directly with acyl-ACP produced by the biosynthetic pathway to transfer a fatty acid to 2-acyl-GPE and re-generate PtdEtn. In cells lacking PtdEtn, other lipid species are used to acylate the N terminus of the lipoprotein (8).
The missing piece in the 2-acyl-GPE cycle is the understanding of how the 2-acyl-GPE on the outside of the cell is translocated to the cytoplasmic aspect of the inner membrane to be acylated by Aas. One possibility is that the lipid simply diffuses or spontaneously flips across the membrane. However, this scenario seems unlikely, because protein transport systems appear to be required for the movement of phospholipids across cell membranes. The best known transporters are members of the ATP-binding cassette (ABC) transporter superfamily (TC 3.A.1) that possess two ABC domains, which hydrolyze ATP, and energize unidirectional substrate transport through the membrane-spanning domains of these complexes (9,10). This group of transporters translocates phospholipids, sterols, bile acids, and hydrophobic drugs (11). They were first recognized as multidrug transporters that interfere with the control of cancer by mediating the extrusion of cytotoxic drugs from the cell (12). E. coli has homologues to these transporters such as MsbA, which is clearly implicated in the ATP-dependent efflux of lipid A and phospholipids from the interior of the cell (13)(14)(15)(16)(17). Bacteria also possess a family of proton-dependent efflux systems for hydrophobic drugs, polyketides, lipooligosaccharide signaling molecules, and outer membrane lipids known as the resistance-nodulation-cell division superfamily (TC 2.A.6) (18). AcrB of E. coli is a typical member of this class whose substrates include hydrophobic drugs such as acridine (19). Lipid transporters in this superfamily include the MmpL7 (20,21) and MmpL8 (22) proteins that catalyze the export of the unique outer membrane lipids in Mycobacterium tuberculosis. Another class of flippase is the eukaryotic phospholipid scramblase that facilitates the calcium-dependent externalization of phosphatidylserine from the inner leaflet of the plasma membrane (23,24). Bacterial homologues of phospholipid scramblase are not known.
This report describes the identification of a Lysophospholipid Transporter (LplT) that translocates 2-acyl-GPE to the cytoplasmic aspect of the inner membrane. This new lipid transporter does not belong to either the ABC or resistance-nodulation-cell division superfamily of transporters that are known to utilize lipid substrates but rather is a member of the major facilitator superfamily (MFS) (TC 2.A.1), which consists of integral membrane proteins with 12-membrane-spanning segments that are best known for their ability to transport all types of small molecules including carbohydrates, amino acids, and inorganic ions (25). MFS transporters catalyze either cation symport (either proton or sodium), cation antiport, or uniport (facilitated diffusion). The LplT protein is the first member of the MFS family shown to use a lipid substrate and differs from the lipid transporters described to date in catalyzing facilitated diffusion rather than active transport of phospholipids across the bilayer.
Generation of Deletion Mutants-E. coli strains were derivatives of strain BW25113 (26) and are listed in Table I. Bacteria were usually cultured in Luria-Bertani (LB) broth (27) at 37°C, and when appropriate, ampicillin (Ap) or kanamycin (Km) was added to the medium to the final concentration of 100 or 25 g/ml, respectively. Mutations were generated using the methods described by Datsenko and Wanner (26). To prepare competent cells, strain BW25113 carrying plasmid pKD46 (expressing the -Red recombinase) was cultured at 30°C in the presence of ampicillin and 1 mM L-arabinose (28). When the A 600 reached 0.6, cells were harvested, washed three times with ice-cold 10% glycerol, and re-suspended in the same solution to 1% original volume. The kanamycin gene from pKD4 was cloned by PCR using gene-specific primers. Primers D-For-Aas and D-Rev-LplT were used to introduce deletions in aas and lplT simultaneously in strain LJ964. After PCR product purification, DpnI digestion, and re-purification by electrophoresis, the PCR products were transformed into competent BW25113 cells by electroporation (Gene Pulser, pulse controller at 200 megohms, capacitance at 250 microfarads, voltage at 25-kV, 0.1-cm chambers). After electroporation, cells were incubated for 1 h at 37°C with shaking (29) and subsequently plated onto LB agar plates containing Km. The Km r transformants were first purified on Km-LB plates and then grown in LB medium without antibiotics and finally tested for Ap and Km resistance. Replacement of the target gene by the km gene in the Ap s Km r mutants was verified by three PCRs using bacterial DNA as template (26).
The kanamycin cassette was deleted by transforming the mutants with pCP20 (expressing the flippase recombinase), plated on LB agar plates containing Ap, and incubated at 30°C. Colonies obtained were grown overnight at 42°C in LB medium without antibiotics and then tested for sensitivity to Ap and Km. The three PCRs described above were repeated on the Ap s Km s mutants to check for the loss of the kanamycin gene. The fadD::km mutation was transferred to strains LJ962 (⌬aas), LJ963 (⌬lplT), and LJ964 (⌬aas ⌬lplT) by transduction with P1 vir phage, and the km gene was removed to generate strains LJ965 (⌬fadD ⌬aas), LJ966 (⌬fadD ⌬lplT), and LJ967 (⌬fadD ⌬aas ⌬lplT), respectively.
DNA Manipulations and Gene Cloning-Standard methods were used for restriction enzyme digestion, agarose gel electrophoresis, ligation, and transformation (29). Chromosomal DNA and plasmid isolation, as well as the gel extraction and purification of PCR products, were carried out using Qiagen kits. For gene cloning, the aas and lplT genes were amplified from strain BW25113 chromosomal DNA by PCR. After gel purification, PCR products were treated with SacI and XbaI and then cloned into the SacI and XbaI sites of pBAD28 (30) to generate plasmids pBAD28-aas and pBAD28-lplT. Strain DH5␣ was used for cloning.
Preparation of 2-Acyl-lysophospholipids-2-Acyl-lysophospholipids were prepared by digestion of the corresponding phospholipids with Rhizopus arrhizus lipase essentially as described (31). The method was modified as follows. Approximately 2 mol of phospholipid were dispersed in 0.1 ml of diethylether, 0.1 ml of 0.1 M CaCl 2 , 0.8 ml of 50 mM bis-Tris, pH 5.6, and 1 mg/ml R. arrhizus lipase (Sigma, catalog number L-4384) were added, and the whole mixture was incubated at 37°C for 1 h with shaking. After incubation, 1 ml of methanol was added to the cooled reaction mixture and the fatty acids liberated were washed three times with 4 ml of petroleum ether/diethylether (1:1, v/v). The lower phase was combined with 0.2 ml of 0.5 M citric acid and extracted with a mixture of 0.5 ml of methanol and 2.2 ml of chloroform. The solvent was evaporated under nitrogen, and the residue was dissolved in ethanol at the desired concentration. Thin-layer chromatography (Silica Gel G plates developed with chloroform/methanol/acetic acid/water, 65:25:8:3 (v/v)) was used to monitor the reaction, which proved to be completed under the conditions described above. 2-Acyl-lysophospholipid concentrations were estimated according to the method of Stewart (32) using the corresponding phospholipid as a standard.
Fatty Acid Incorporation Assays-Cells were grown in M9 minimal medium (29) containing glycerol (0.4%), casein hydrolysate (0.2%), and Brij-58 (0.5%). When the A 600 reached 0.5, [ 3 H]palmitate from a stock solution (1 mCi/ml, 0.16 mM) was added to 1 ml of culture to a final concentration of 0.8 M and 100-l aliquots were periodically removed and applied to 0.45-m Millipore filters. Filters were washed twice with 10 ml of ice-cold 50 mM Tris buffer, pH 7, containing 0.5% Brij-58 and dried. The radioactivity was measured by scintillation counting with 10 ml of Biosafe NA scintillation fluid (Research Products International Corp., Mt. Prospect, IL). The data were normalized to nmol/g dry weight using the specific radioactivity of the palmitic acid, and the number of cells per sample was determined from the optical density using 1.4 A 600 ϭ 450 g dry weight. Preparation and Uptake of Fluorescent 2-Acyl-GPE by Spheroplasts-Strains LJ966 (⌬fadD, ⌬lplT), LJ961 (⌬fadD), and LJ965 (⌬aas ⌬fadD) were grown to a density of 5 ϫ 10 8 cells/ml in M9 minimal medium (27) supplemented with 0.1% casamino acids, 0.4% glycerol, and 0.0005% thiamin. Spheroplasts of all three strains were made using the basic lysozyme-EDTA method (33). The spheroplasts were washed twice with phosphate-buffered saline and re-suspended in the same buffer supplemented with 0.4% glycerol. 2-Dodeceyl-NBD-GPE was added to the spheroplasts to a final concentration of 10 M and incubated at 37°C. After 15, 30, and 60 min, 1-ml aliquots (representing 3 ϫ 10 9 cells) of spheroplasts were removed and harvested by centrifugation. The spheroplasts were washed twice with M9 medium and resuspended in 100 l of water. The lipids were extracted as described by Bligh and Dryer (34). Lipid samples were spotted onto Silica Gel G thin-layer plates and developed with chloroform/methanol/acetic acid (65:25:10, v/v/v). The phospholipids on the dried plate were visualized using a Typhoon 9200 (Amersham Biosciences) in fluorescence mode with an excitation wavelength of 532 nm (green laser) and an emission filter of 555 nm with a bandwidth of 20 nm. The fluorescent PtdEtn was quantitated with ImageQuant 5.2 (Amersham Biosciences) and normalized to cell number. The fluorescent PtdEtn signal from strain LJ965 (⌬aas ⌬fadD) was used as the background for calculations. 32 P Labeling-Bacterial strains were grown in rich medium that contained 100 Ci/ml [ 32 P]orthophosphate to a density of 8ϫ10 8 cells/ ml. The cells from 10 ml of culture were harvested, and the lipids were extracted as described by Bligh and Dryer (34). The same total counts of 32 P-labeled lipids (10 5 cpm) were loaded onto Silica Gel G thin-layer plates and developed with chloroform/methanol/acetic acid (65:25:10, v/v/v) to determine the percentage of the total lipids that were acyl-GPE. The dried plate was exposed to a phosphorimaging screen for 16 h. The phospholipid bands were visualized using the Typhoon 9200, and acyl-GPE band was quantified with ImageQuant 5.2.

Distribution of LplT in Bacteria-
The transporter classification database includes representatives of all of the functionally characterized transporters as well as many putative transport proteins (35,36). We used the BLAST search tool to screen all of the proteins in the transporter database against the proteins in a database of prokaryotic protein fusions called FusionDB (igs-server.cnrs-mrs.fr/FusionDB/) (37). Among the hits retrieved were proteins that possessed N-terminal domains equivalent to a permease belonging to the MFS and C-terminal domains related to either acyltransferase superfamily (PlsC) or a acyltransferase domain fused with a member of the acyl-CoA synthetase family (ACS) (Fig. 1B and Table II). These proteins are derived from the ␣, ␥, ␦, and ⑀ subgroups of the proteobacteria. The MFS permease homologues identified in this search comprise a novel family within the MFS, which we term the LplT Family (TC subfamily 2.A.1.42).
An important connection was the realization that the PlsC-ACS domain associated with the LplT domain in these organisms are homologous to and probably orthologous to the 2-acyl-GPE acyltransferase (Aas) of E. coli (7,38) (Fig. 1B, 1). This bifunctional enzyme participates in the 2-acyl-GPE cycle outlined in Fig. 1A. In some bacteria, the bifunctional enzyme is fused to LplT as a single polypeptide (Fig. 1B, 2, and Table II), and in other bacteria, the LplT domain is fused to either a PlsC domain (Fig. 1B, 3, and Table II) or in the same operon as a PlsC domain (Fig. 1B, 4, and Table II). A key fact arising from this analysis was that Gram-negative bacteria possess these LplT-PlsC-ACS-associated systems, whereas Gram-positive bacteria do not. This finding is consistent with the function of LplT in a metabolic cycle responsible for the formation of the major lipoprotein of the outer membrane (Fig. 1A). These strong associations suggested that the LplT protein/domain functions with Aas in the 2-acyl-GPE cycle. In Chlamidiae, LplT is associated with a sequence of unknown function (Fig.  1B, 5, and Table II). This domain, referred to as the Pfam-B24103 domain (39), was C-terminal to the full-length LplT.
Average hydropathy and similarity plots were generated (40) for all of the LplT permeases and permease domains (data not shown). The hydropathy plot revealed 12 clear peaks of hydrophobicity as anticipated for members of the MFS. The multiple sequence alignments upon which the hydropathy plot was based revealed discrete regions of sequence conservation between members of the LplT family (www-biology.ucsd.edu/ ϳmsaier/transport/). The predicted cytoplasmic loop regions between transmembrane segments 2 and 3, 4 and 5, 8 and 9, and 10 and 11 were well conserved within the LplT group. An RK-rich sequence RKFKRK occurred between transmembrane segments 2 and 3 where a related motif RKXGRK 2 is found in many MFS permeases with variations from family to family (25). None of these residues was fully conserved in the LplT family members. Between transmembrane segments 4 and 5 and within segment 5 was a motif that was conserved in the  Table II for a list of the organisms where these protein types are found.
LplT group: FGPXKY(G/S)(L/I/V) 2 PX 2 LX 3 NX 13 G, where underlined residues are fully conserved, X represents any residue, and residues within parentheses represent alternative possibilities at one position. This sequence motif and the greater degree of similarity within the N-terminal halves of these permeases, as compared with their C-terminal halves, are characteristics that define the LplT family.
Exogenous Fatty Acid Uptake Defect in lplT Mutants-The function of the LplT family of permeases was investigated using the E. coli model system. As described under "Experimental Procedures," we constructed single, double, and triple knock-out mutants of the fadD, lplT, and aas genes. All of the mutations were deletions that would not alter the expression of other adjacent genes. The lplT and aas genes are located in an operon with lplT as the leading gene. We confirmed using the Aas biochemical assay of isolated inner membranes that the ⌬lplT mutants retained the same Aas-specific activity as the wild-type cells and that the ⌬aas mutants lacked this enzyme activity. In the presence of FadD, exogenous [ 3 H]palmitate label was converted to intracellular acyl-CoA and incorporated into all of the phospholipid classes by the de novo pathway via PlsB and PlsC. This system is a high-capacity pathway that allows a much greater rate of incorporation than the Aas/LplT pathway. Moreover, phospholipid labeling in cells with a functional FadD was not appreciably affected by the presence of the ⌬aas or the ⌬lplT mutations (data not shown). This finding was essentially the same as reported previously (38). In a ⌬fadD genetic background, the acyl-CoA pathway is eliminated, reducing the rate and extent of [ 3 H]palmitate incorporation ϳ10-fold ( Fig. 2A) and the only labeled phospholipid was PtdEtn as determined by thin-layer chromatography (38). The loss of Aas eliminated palmitate incorporation (Fig.  2A). These data with the ⌬fadD ⌬aas double mutant were as reported previously (38). Palmitate incorporation into PtdEtn in the ⌬fadD ⌬lplT double mutants was one-third the rate in ⌬fadD controls, suggesting that LplT functions in the same pathway as Aas (Fig. 2A). The ⌬fadD ⌬lplT ⌬aas triple mutant gave the same result as the ⌬fadD ⌬aas double mutant, showing essentially no PtdEtn labeling (Fig. 2A). These results suggest that LplT provides substrates to the Aas acyltransferase, although it cannot be responsible for the entire flux through the Aas acylation pathway. This may be due either to the generation of lysophospholipids inside the cell or to the movement of exogenous 2-acyl-GPE across the inner membrane by an LplT-independent process.
The ⌬lplT mutants were complemented by the introduction of plasmids bearing the wild-type lplT gene under a regulated promoter. Complementation of the ⌬fadD ⌬lplT double mutant by the cloned lplT gene increased the incorporation of [ 3 H]palmitate into PtdEtn over what was observed for the ⌬fadD mutant, showing that the defect in PtdEtn labeling via Aas was reversed by lplT expression (Fig. 2B). Also, the presence of the plasmid bearing the wild-type lplT gene enhanced [ 3 H]palmitate labeling by the ⌬fadD mutant bearing an intact chromosomal lplT gene (Fig. 2C). Control complementation analyses were also conducted. Inclusion of the lplT gene in trans in the ⌬fadD ⌬aas double mutant or of the trans aas gene in the ⌬fadD ⌬lplT double mutant exerted no stimulatory effect on palmitate incorporation. Expression of the aas gene in trans in the ⌬lplT ⌬aas double mutant marginally stimulated palmitate labeling (data not shown). These results are consistent with the conclusion that lplT encodes a transport function that provides the 2-acyl-GPE precursor to the Aas-catalyzed 2-acyl-GPE acyltransferase reaction. However, the fact that PtdEtn labeling was not eliminated in the lplT mutants means that LplT does not provide the only source of Aas substrates. Uptake of [ 3 H]palmitate was studied in energy-proficient cells as a function of substrate concentration. To calculate the kinetic constants, the differences in uptake for the ⌬fadD single mutant and the ⌬fadD ⌬lplT double mutant were calculated at a series of palmitate concentrations. A double reciprocal Lineweaver-Burk plot (1/ versus 1/s) gave a straight line.
In the ⌬fadD single mutant, the Michaelis-Menten constant (K m ) was 3.3 Ϯ 0.3 M and the maximal rate of uptake (V max ) was 0.33 Ϯ 0.04 mol/g dry weight/h. Uptake was also quantitated as a function of palmitate concentration in the ⌬fadD ⌬lplT double mutant. The K m and V max values were 5.7 Ϯ 0.6 M and 0.31 Ϯ 0.02 mol/g dry weight/h, respectively. The optimal pH for the uptake of [ 3 H]palmitate was 8, both for the ⌬fadD single mutant and for the ⌬fadD ⌬lplT double mutant. Thus, the kinetic parameters characterizing fatty acid movement into the cells were not affected by eliminating LplT. We also investigated the energetics of the palmitate labeling in ⌬fadD mutants by performing experiments in the presence of arsenate to block ATP production or carbonyl cyanide m-chlorophenylhydrazone to collapse the membrane electrochemical gradient. Arsenate (10 mM) inhibited ATP synthesis and completely blocked palmitate uptake, whereas 1 mM arsenate inhibited 74%. This result was consistent with the activity of the Aas acyltransferase dependent on ATP for activation of the fatty acid (6). LplT is a member of the MFS and therefore could function either by proton symport and be dependent on the proton motive force or by uniport (facilitated diffusion) and be independent of the proton motive force. Proline transport (known for its dependence on the proton motive force) was strongly inhibited by the carbonyl cyanide m-chlorophenylhydrazone in ⌬fadD mutants; however, palmitate incorporation into PtdEtn was not (data not shown). These data suggest that LplT transport activity is independent of the proton motive force, pointing to a facilitated diffusion mechanism.
Lysophospholipid Metabolism in ⌬lplT Mutants-Aas is one of several pathways for the metabolism of 2-acyl-GPE, and the inactivation of this gene leads to a modest increase in cellular acyl-GPE (38). Therefore, we performed metabolic labeling experiments with [ 32 P]orthophosphate and measured the cellular levels of acyl-GPE (Fig. 3). The ⌬fadD ⌬lplT mutant accumulated significantly more acyl-GPE (0.76% total) than either the ⌬fadD (0.13%) or ⌬fadD ⌬aas (0.28%) mutants. These data clearly illustrate a defect in acyl-GPE metabolism in ⌬lplT mutants, consistent with a role for LplT in delivering acyl-GPE substrates to Aas. The fact that more acyl-GPE accumulated in ⌬lplT mutants than in ⌬aas mutants suggests that LplT function is also required for alternate pathways for acyl-GPE processing, such as the PldB pathway for acyl-GPE degradation. The reactions catalyzed by this enzyme are responsible for the degradation of lysophospholipid via hydrolysis and/or transacylation (42,43).
Intact E. coli cells are very inefficient in the uptake and acylation of exogenous lysophospholipids. Nonetheless, small amounts of exogenous 2-acyl-GPC are taken up and acylated by Aas (44). 2-Acyl-GPC uptake and acylation were reduced to approximately one-third in the ⌬lplT mutant compared with the wild-type control (data not shown). Aas mutants were unable to form labeled phosphatidylcholine from exogenous 2-acyl-GPC (44). These data are consistent with the [ 3 H]palmitate labeling (Fig. 2) and with the idea that LplT catalyzes the transfer of lysophospholipids to Aas. However, it does not provide the only mechanism for lysophospholipid movement across the inner membrane.
In the preceding experiments, lysophospholipid transport was assessed based on the acylation of lysophospholipids with exogenous fatty acids. As a more direct assay, we prepared fluorescently labeled 2-acyl-NBD-GPE to study the uptake and acylation of the lysophospholipid with endogenous fatty acids (Fig. 4). We reasoned that a major impediment to the uptake of exogenous lysophospholipids is the outer membrane. Thus, we prepared right-side out spheroplasts from ⌬fadD, ⌬fadD ⌬aas, and ⌬fadD ⌬lplT mutants to study the uptake and acylation of the fluorescent lysophospholipid. Spheroplasts derived from the ⌬fadD mutant were able to transfer 2-acyl-NBD-GPE into the Aas-accessible intracellular compartment (Fig. 4A). The formation of fluorescent PtdEtn from 10 M 2-acyl-NBD-GPE was linear with time and was reduced to one-third of the rate in spheroplasts derived from the ⌬fadD ⌬lplT mutant (Fig. 4A). This is the same result as observed by two previous experiments using fatty acid labeling (Figs. 2 and 3), illustrating that LplT provides an important but not the only mechanism for the transfer of extracellular lysophospholipids to Aas.
We next examined 2-acyl-NBD-GPE uptake as a function of concentration (Fig. 4B). Spheroplasts from the ⌬fadD mutant exhibited non-linear uptake and acylation of 2-acyl-NBD-GPE, consistent with the presence of a saturable system. Uptake and acylation in the ⌬fadD ⌬lplT mutant increase linearly as a function of 2-acyl-NBD-GPE concentration, indicating that LplT-independent transport was not saturable, possibly because it was not protein-mediated. The PtdEtn formed by the ⌬fadD ⌬lplT mutant was subtracted from the data with the ⌬fadD mutant to evaluate the LplT-mediated component. These data were fit to an equation for single-site binding (r 2 ϭ 0.98), which yielded a calculated K d of 7.5 Ϯ 2 M for 2-acyl-NBD-GPE. DISCUSSION The key finding of this research is the identification of the lplT gene product of E. coli as a lysophospholipid transporter. The best known lipid transporters are members of the ABC family of ATP-dependent translocators that drive the unidirectional movement of phospholipids and related hydrophobic molecules across the membranes. MsbA is an example of this class of transporter in E. coli and functions to export lipid A and phospholipids in order to assemble the outer membrane (13)(14)(15)(16)(17). Because Ϸ75% lipid molecules are localized to the outer membrane and the outer leaflet of the inner membrane, energy-dependent unidirectional transport is required to move the lipids from their site of synthesis inside the cell to their extracellular destinations. The function of the LplT-Aas system proved to be independent of the membrane electrochemical gradient, suggesting that LplT catalyzes facilitated diffusion of lysophospholipids across the membrane. This is a characteristic of several MFS transporters that facilitate the bidirectional movement of molecules across membranes. In many cases, these facilitators are coupled to a second enzyme that "traps" the transport substrate inside the cell to achieve net uptake. An example of such a trap in E. coli is the glycerol facilitator (glpF) that shuttles glycerol across the membrane where it is The acylation of the substrate in spheroplasts from strain LJ965 (⌬fadD ⌬aas) was used as the background for calculation. In both experiments, product formation was quantitated by lipid extraction followed by thin-layer chromatography as described under "Experimental Procedures." trapped within the cell by phosphorylation by glycerol kinase (GlpK) (45). The glpF-glpK genes are located in an operon with the facilitator as the leading gene (46). The LplT-Aas system has the same gene organization and overall functional characteristics as the GlpF-GlpK system. LplT acts as the facilitator that flips 2-acyl-GPE into the cell and works in concert with intracellular Aas, which functions as an acylation trap for the lysophospholipid (Fig. 1A). The existence of gene fusions encoding LplT-Aas and LplT-PlsC proteins (Fig. 1B) strongly supports the conclusion that the transporters function in concert with the acyltransferases in the same metabolic process. A provocative observation is that in the chlamydial kingdom LplT-like facilitators are fused to a domain of unknown function (Fig. 1B). It will be interesting to determine the function of this domain.
The accumulation of acyl-GPE in ⌬lplT mutants (Fig. 3) establishes LplT as an important catalytic element in the metabolism of this intermediate in vivo. We propose that LplT functions in the 2-acyl-GPE cycle that exists to reutilize the 2-acyl-GPE generated outside the cell in the maturation of the major outer membrane lipoprotein (Fig. 1A). However, our metabolic labeling studies (Fig. 2) and fluorescent acyl-GPE and 2-acyl-GPC uptake experiments (Fig. 4) consistently showed some internalization of 2-acyl-GPE in the ⌬lplT mutant strains. Therefore, LplT does not provide the sole mechanism for the movement of lysophospholipids into the cells. We do not think this residual activity represents another lysophospholipid transporter because we have no evidence for saturable binding. Rather, the residual activity in ⌬lplT mutants may represent the spontaneous rate of flip-flop for the lysophospholipid.
Aas is not the only pathway for lysophospholipid metabolism. 2-Acyl-GPE is degraded by lysophospholipases (31,47) and converted to PtdEtn by transacylation reactions involving another lysophospholipid (41,42). The reduced accumulation of acyl-GPE in vivo in ⌬aas mutants compared with ⌬lplT mutants (Fig. 3) suggests that these processes also occur inside the cell. These activities are substantially reduced in pldB mutants (42), and accordingly, aas pldB double mutants accumulate more acyl-GPE than aas mutants (38). The differences in 2-acyl-GPE accumulation in vivo in the various mutant strains indicate that 2-acyl-GPE metabolism occurs inside the cell and that LplT is necessary to effectively deliver the substrate to these enzyme systems.