Reconstitution of phosphatidylserine transport from chemically defined donor membranes to phosphatidylserine decarboxylase 2 implicates specific lipid domains in the process.

Phosphatidylserine (PtdSer) is transported from its site of synthesis in the endoplasmic reticulum to the locus of PtdSer decarboxylase 2 (Psd2p) in the Golgi/vacuole and decarboxylated to form phosphatidylethanolamine. Recent biochemical and genetic evidence has implicated the C2 domain of Psd2p and a membrane-bound form of the phosphatidylinositol binding/transfer protein, PstB2p, as essential for this transport process. We devised a reconstituted system in which chemically defined donor membranes function to transfer PtdSer to the biological acceptor membranes containing Psd2p. The transfer of PtdSer is poor when the donor membranes have a high degree of curvature but markedly enhanced when the membranes are relatively planar (> or =400-nm diameter). PtdSer transfer is also dependent upon both the bulk and the surface concentrations of the lipid, with pure PtdSer vesicles acting as the most efficient donors at all concentrations. The lipid transfer from donor membranes containing either 100% PtdSer or 50% PtdSer at a fixed concentration (e.g. 250 microM PtdSer) differs by a factor of 20. Surface dilution of PtdSer by choline, ethanolamine, glycerol, and inositol phospholipids markedly inhibits PtdSer transfer, whereas phosphatidic acid (PtdOH) stimulates the transfer. Most importantly, the transfer of PtdSer from liposomes to Psd2p fails to occur in acceptor membranes from strains lacking PstB2p or the C2 domain of Psd2p. These data support a model for PtdSer transport from planar domains highly enriched in PtdSer or in PtdSer plus PtdOH.

The mechanisms responsible for transporting newly synthesized glycerophospholipids among the different subcellular compartments within eukaryotic cells remain largely undefined at the genetic and biochemical level (1). In many instances, the data support a process that is relatively rapid and independent of a requirement for ATP or other nucleotides and largely independent of factors essential for vesicle movement that are an integral part of protein traffic between membrane compartments (2)(3)(4)(5)(6)(7)(8)(9). Morphological and biochemical experiments have suggested that one possible mechanism may consist of the formation of zones of apposition between donor and acceptor membranes that facilitate lipid transport between membranes (3, 10 -15). We have developed a series of genetic screens and biochemical assays to study the transport of Ptd-Ser to the loci of PtdSer 1 decarboxylases in an effort to define some of the molecules involved in glycerophospholipid transport processes (16). These screens and assays follow the synthesis of PtdSer in donor membranes and its subsequent decarboxylation to form phosphatidylethanolamine (PtdEtn) in acceptor membranes. The formation of this PtdEtn provides a distinct biochemical marker for the transfer reaction. In yeast, there are two PtdSer decarboxylases, Psd1p and Psd2p, that are located in the mitochondria and the Golgi/vacuole, respectively (17)(18)(19)(20). Combined genetic and biochemical experiments have thus far implicated one protein in transport of PtdSer between a subfraction of the endoplasmic reticulum (known as the mitochondria-associated membrane) and mitochondria (21). This latter protein is a subunit of ubiquitin ligase, Met30p (22,23), and its action affects the ability of mitochondriaassociated membrane to act as a PtdSer donor and mitochondria to act as a PtdSer acceptor (21). Thus far, three proteins have been implicated in the transfer of PtdSer to the locus of Psd2p. One of these proteins is the phosphatidylinositol-4kinase, Stt4p (24,25). A second protein is a phosphatidylinositol (PtdIns) transfer/binding protein, PstB2p (7). The third protein motif is the C2 domain of the Psd2p (26). Our previous data provide some support for a model in which docking reactions occur between donor/acceptor pairs of membranes that enable lipid transport to occur between them (16). These docking reactions may require protein modification (e.g. protein ubiquitination), lipid modification (e.g. phosphorylation of Ptd-Ins), and specific lipid binding reactions, by the actions of PstB2p and the C2 domain of Psd2p and/or other proteins. One current hypothesis is that macromolecular assemblies form on and between both the donor and acceptor membranes to create zones of adhesion, within which specific proteins function to transport phospholipids (16).
Previous studies have established conditions for the transport of nascent PtdSer from its site of synthesis to the locus of Psd2p in intact cells, permeabilized cells, and isolated membranes (7,8,24,26). In all cases, the transport of PtdSer required the presence of the PstB2p and the C2 domain of Psd2p on the acceptor membrane. In an effort to further deconstruct and reconstitute the transport reaction, we have now addressed the problem of defining the basic molecular require-ment for producing competent donor membranes. These studies focused upon defining 1) the minimal physical requirements, 2) the lipid compositional requirements, 3) the ionic requirements, and 4) the fidelity of defined donors in the Ptd-Ser transport reaction. Our results demonstrate that planar PtdSer-rich domains function optimally in the transport reaction with the same genetic and molecular constraints observed for the process in vivo.
Preparation of Membranes-Yeasts in 1-liter cultures were grown to an A 660 ϭ 1.8 and harvested by centrifugation at 5000 ϫ g for 5 min at 4°C. The cell pellet was resuspended in 1 liter of deionized water and sedimented by centrifugation at 5000 ϫ g for 5 min at 4°C. The resultant pellet was resuspended at 0.5 g, wet weight, in 1 ml of buffer (100 mM Tris-sulfate, pH 9.4, and 10 mM dithiothreitol) and incubated at 37°C for 10 min. Next the cells were centrifuged at 1000 ϫ g for 5 min at ambient temperature. The cell pellet was resuspended in 50 ml of 1.2 M sorbitol and recentrifuged. The resultant pellet was resuspended at 0.15 g, wet weight, in 1 ml of spheroplasting buffer, containing 1.2 M sorbitol, 1% yeast extract, 2% peptone, and 0.5% glucose. Zymolyase was added at 5 mg of enzyme/g, wet weight, of cells for the purpose of degrading the yeast cell wall. The cell suspension was incubated a 30°C with gentle agitation for 45 min. At the end of the incubation, the spheroplasts were harvested by centrifugation at 2000 ϫ g for 5 min. The spheroplasts were washed in 50 ml of 1.2 M sorbitol and recovered by centrifugation. The final spheroplast pellet was resuspended in 100 ml of regeneration buffer, containing 0.7 M sorbitol, 0.75% yeast extract, 1.5% peptone, and 1% glucose. This spheroplast suspension was incubated at 30°C for 20 min with shaking at 120 rpm to allow the cells to physiologically recover from the spheroplasting procedure. After regeneration, the cells were washed in 50 ml of 1.2 M sorbitol and collected by centrifugation. The pellet was resuspended in 5 ml of cold homogenization buffer containing 50 mM Tris-Cl, pH 8.0; 0.5 mM EDTA; 0.3 M sucrose; 1 mM 2-mercaptoethanol; 10 g/ml antipain, leupeptin, and aprotinin; 1 g/ml pepstatin A; and 1 mM phenylmethylsulfonyl fluoride. The cells were homogenized using 15 strokes with the B pestle of a Dounce homogenizer (Kontes). The homogenate was centrifuged at 1500 ϫ g for 5 min at 4°C, and the supernatant was saved. The pellet derived from the first centrifugation was again homogenized and centrifuged in the same manner to yield a second supernatant, which was combined with the first supernatant. The supernatants were centrifuged at 30,000 ϫ g for 15 min at 4°C. The new supernatant was harvested and centrifuged at 4 ϫ 10 5 ϫ g for 1 h. The membrane pellet was resuspended in 3 ml of homogenization buffer and homogenized using a motor-driven pestle in a glass tube. The final membrane suspension was adjusted to 3.5 ml/liter of starting culture and stored at Ϫ80°C in 200-l aliquots. These membranes remain competent to act as acceptors for lipid transport for at least 6 months.
Preparation of Liposomes-All lipids were obtained from Avanti Polar Lipids. Multilamellar liposomes were prepared by first removing organic solvents from the stock lipid solutions using nitrogen evaporation. The dried lipid films were hydrated in 10 mM Tris-Cl, pH 7.0, 0.1 M KCl at 30°C, for 1 h. The lipid suspension was subjected to vortex mixing for 5 min at room temperature to create multilamellar liposomes. Unilamellar liposomes were prepared either by sonication or membrane extrusion. For sonication, the multilamellar liposomes were placed in glass tubes and probe-sonicated on ice for 30 s followed by cooling for 30 s. The sonication-cooling cycle was performed three times. The sonicated liposome suspension was briefly centrifuged at 1000 ϫ g for 3 min to remove titanium particles shed from the probe. For membrane extrusion, we used an Avestin Liposofast apparatus. The multilamellar liposome suspensions were converted to unilamellar vesicles of defined size, using 25 passes through polycarbonate filters of either 50-, 100-, 200-, 400-, or 1000-nm diameter, as specified in individual experiments.
Enzyme Assays-The transport-coupled decarboxylation assay was performed in a volume of 40 l containing 25 mM Tris-Cl, pH 7.0, 10 mM 2-mercaptoethanol, 150 mM KCl, and substrates ranging from 50 to 500 M total lipid. Membranes containing Psd2p and 60 g of total protein were added to the incubations containing the liposome substrates. The specific radioactivity of the Ptd[1Ј-14 C]Ser substrate added to the reactions was 0.9 Ci/mol of total lipid. The reactions were performed in gas-tight tubes with an insert well for trapping CO 2 on 2 M KOHimpregnated filter paper. Reactions were performed at 30°C for 20 min and terminated by the addition of 50 l of 0.25 M H 2 SO 4 introduced with a syringe and needle through a septum. The evolved 14 CO 2 was collected at 30°C for 1 h. The filter paper was recovered from each reaction tube and radioactivity quantified by liquid scintillation spectrometry. Psd2p activity that is independent of transport processes was measured using NBD-[Ptd1Ј-14 C]Ser as described previously (8,26).

Donor Membranes with a Low Degree of Curvature Facilitate
PtdSer Transport to Psd2p-Since its discovery, the Psd2p enzyme activity has always been difficult to measure using its native substrate (19,20). Measurement of catalytic activity with Ptd[1Ј-14 C]Ser in the presence of detergents routinely gave very low activity, presumably due to detergent-mediated inactivation of the enzyme. The use of a membrane-partitioning substrate, NBD-Ptd[1Ј-14 C]Ser, provided a means to measure Psd2p activity in the absence of detergents (19,20), but this substrate was not useful for reconstituting transport, because it spontaneously inserts into all membranes (27). Sonicated preparations of Ptd[1Ј-14 C]Ser liposomes routinely proved to be very poor or inactive substrate donors for Psd2p, although biological membranes harboring Ptd[1Ј-14 C]Ser functioned as reasonably good substrates (7,8,26). In addition, liposomes prepared from biological membranes that functioned as substrates were also poor substrate donors for the enzyme. We initially sought to understand the discrepancies between biological membranes and liposomes as substrate donors. In preliminary studies, we discovered that large, multilamellar Ptd[1Ј-14 C]Ser liposomes could function as reliable substrates for the enzyme, whereas the activity from small unilamellar Ptd[1Ј-14 C]Ser liposomes was nearly undetectable. From these initial observations, we refined the biochemical assay using Ptd[1Ј-14 C]Ser liposomes of defined size, prepared by the polycarbonate filter extrusion method. Following the refinements of the assay, we returned to the initial comparisons between small unilamellar and large multilamellar vesicles assayed under optimal conditions. These results are presented in Fig. 1 and demonstrate that at a fixed concentration of 0.2 mM Ptd-Ser, there is a 3-fold difference in the activity of the enzyme measured with the multilamellar liposomes, compared with the unilamellar liposomes. This magnitude of difference is likely to be an underestimate, because the multilamellar vesicles have only their outermost lamellar layer of PtdSer available for transfer to the enzyme, whereas the small unilamellar vesicles contain only one phospholipid bilayer. Thus, with the smaller vesicles, there is far more PtdSer available to interact with the acceptor membranes. A direct comparison of unilamellar lipo-somes of defined diameter, as substrates for the transport reaction, is shown in Fig. 2. These data demonstrate that donor activity varies with vesicle diameter and concentration and that the most efficient substrate donors are liposomes Ն400 nm in size. These findings indicate that membranes with a high degree of curvature are relatively poor substrate donors, whereas relatively planar membranes function more efficiently as substrate donors for Psd2p. This observation is intriguing, since zones of apposition between donor and acceptor membranes observed in micrographs have relatively planar character (13,14,28).
PtdSer-rich Membranes Are the Most Efficient Substrate Donors for Psd2p-We next sought to determine whether the surface concentration of PtdSer played an important role in the interaction of the donor and acceptor membranes. Large unilamellar liposomes of 400-nm diameter were prepared from Ptd[1Ј-14 C]Ser and varying amounts of PtdCho ranging from 0 to 50 mol %. These liposomes were incubated with membranes containing Psd2p, and the transport-dependent decarboxylation of PtdSer was measured. From the data in Fig. 3A, it is apparent that the rate of the coupled transport/decarboxylation reaction is dependent upon the concentration of the added lipid. With pure 100 mol % PtdSer liposomes, the rate appears saturable with increasing substrate concentration and gives a hyperbolic initial velocity versus substrate concentration curve. The addition of PtdCho has a dramatic effect upon the initial rate of the reaction. The introduction of 15 mol % PtdCho reduces the initial reaction rate by more than 50%. Further increases in the mol % of PtdCho also appear to have a disproportionate effect upon the function of the donor membranes. Data provided below demonstrate that this is not an inhibitory effect of PtdCho upon the Psd2p but rather appears to be a direct effect on the efficiency of the PtdSer transported to the enzyme. The data shown in Fig. 3B is a replot of the data in Fig.  3A, in which the reaction rate is expressed as a function of the concentration of PtdSer added to the reaction. This form of the data is especially revealing, since it allows for simple comparison of rates among donor liposomes of differing composition but at fixed PtdSer concentration. For example, at 250 M PtdSer, the difference in rate between liposomes of 100 mol % PtdSer and 50 mol % PtdSer varies by a factor of 20. These data demonstrate that PtdSer transfer to Psd2p occurs at a much higher rate from planar membranes highly enriched in PtdSer when compared with membranes with lower PtdSer content. Fig. 3C is another replot of the data that expresses the transportdependent decarboxylation of PtdSer as a function of the surface concentration of PtdCho. This analysis of the data demonstrates the differences in rates with respect to the surface concentration of PtdCho and PtdSer at different bulk concentrations of total lipid. The striking decline in transport activity occurs in an exponential manner with respect to the reduction in surface concentration of the PtdSer and corresponding increase in PtdCho. The exponential decline in activity occurs systematically and in parallel for all of the concentrations of total lipid examined in the reaction. These data indicate that the surface concentration of PtdSer is a critical parameter in regulating the transport of this lipid to the locus of Psd2p. In Fig. 3D, the data from Fig. 3B are shown as a double reciprocal plot. This latter treatment of the data demonstrates that the apparent V max for the transport-coupled reaction is dramatically affected by reduction in the surface concentration of Ptd-Ser, but the apparent K m of the process is not significantly altered (see inset in Fig. 3D). These kinetic parameters are consistent with the Psd2p being substrate-limited as a consequence of reduced PtdSer transport to the enzyme. Collectively, the information in Fig. 3 supports a mechanism in which Ptd-Ser domains in the donor membrane are recognized by proteins present in the acceptor membrane that transport the substrate to Psd2p.
PtdSer Transport to Psd2p Is Not Generally Enhanced by Anionic Phospholipids-The preceding data demonstrate that donor membranes highly enriched in PtdSer, efficiently transfer this lipid to Psd2p. We next sought to determine whether this was a general property of anionic lipids or if it was specific only for PtdSer. We utilized liposomes containing 70 mol % PtdSer and 30 mol % of other phospholipids and compared the transport activity with that of 100 mol % PtdSer liposomes. The results of these studies are shown in Fig. 4. As described above, optimal transport-dependent decarboxylation occurred with 100 mol % PtdSer liposomes, and this activity was markedly diminished (ϳ75%) with liposomes containing 30 mol % Ptd-Cho. Substitution of PtdEtn, phosphatidylglycerol (PtdGro), PtdIns, PtdIns-4-phosphate, and PtdIns-4,5-P 2 for PtdCho failed to restore the transport activity to the level of 100 mol % PtdSer. However, there were some modest, reproducible increases in the transport, relative to PtdCho addition, when the added lipid was PtdGro or PtdIns 4-phosphate. When the data are expressed as a percentage of the activity obtained with pure PtdSer vesicles, the added lipids and their corresponding activities are as follows: PtdSer, 100%; PtdCho, 25%; PtdOH, 104%; PtdEtn, 40%; PtdGro, 48%; PtdIns, 26%, PtdIns-4-P, 54%; PtdIns-4,5-P 2 , 26%. These data clearly demonstrate that the transport of PtdSer is not generally stimulated by the presence of other anionic phospholipids, or non-choline-containing lipids within the donor membrane. However, when the liposomes contained 70% PtdSer and 30% PtdOH, PtdSer transport to Psd2p was fully restored to the levels found for 100% PtdSer. Thus, the PtdSer transport reaction does not display an absolute requirement for 100 mol % PtdSer in the donor membrane to reach maximum efficiency, but only PtdOH appears to function as a useful substitute for PtdSer.
The Transport System Distinguishes between Homotypic and Heterotypic Donor Membranes-The inhibition of PtdSer transport by lipids other than PtdOH could simply reflect the sensitivity of the reconstituted system to the addition of extraneous lipids or it could be a direct effect of surface dilution of PtdSer. In order to examine the nature of the inhibition, we compared the effects of mixtures of homotypic liposomes and heterotypic liposomes upon the coupled transport/decarboxylation reaction. For homotypic conditions, we measured the effects of adding different populations of liposomes, each composed of 100 mol % of the lipids studied, upon PtdSer transport.  composed of 70/30% PtdSer/PtdCho, and 70/30% PtdSer/Ptd-OH show the expected effect on transport-dependent decarboxylation of PtdSer described in Fig. 4. The addition of PtdCho to create heterotypic liposomes inhibits the reaction, whereas the addition of PtdOH stimulates the reaction. Under the homotypic conditions, 100 mol % PtdSer liposomes give the expected signal for transport. Furthermore, admixture of 100 mol % PtdSer liposomes with 100 mol % PtdCho or 100 mol % PtdOH liposomes does not significantly alter the transport of PtdSer to Psd2p. From these findings, we conclude that PtdCho is not an inhibitor of PtdSer transport when it is present in separate membrane domains. This demonstrates that PtdCho is not a direct inhibitor of the transport process. The PtdCho only acts as an inhibitor of transport when it is incorporated into the same liposome as, and dilutes the surface concentration of, the PtdSer.
PtdSer Transport Is Enhanced by Mn 2ϩ Ions-Previous studies with permeabilized cells demonstrated that the translocation of nascent PtdSer to Psd2p required the presence of Mn 2ϩ ions, and other divalent cations (Ca 2ϩ , Cu 2ϩ , Ni 2ϩ , Mg 2ϩ , and Zn 2ϩ ) would not substitute (8). We tested the effects of Mn 2ϩ upon the transport-dependent decarboxylation reaction as shown in Fig. 6. These data demonstrate that Mn 2ϩ stimulates PtdSer transport ϳ2-fold at concentrations in the range of 200 M. In additional studies with the NBD-Ptd[1Ј-14 C]Ser substrate, we determined that Mn 2ϩ does not directly affect the catalytic properties of Psd2p (data not shown). We next tested whether Mn 2ϩ could overcome the inhibition of PtdSer transport caused by surface dilution of the PtdSer by PtdCho and other lipids. The results of these studies are shown in Fig. 7. The data reveal that the inhibition of PtdSer transport caused by PtdCho, PtdEtn, PtdGro, PtdIns, PtdIns-4-P, and PtdIns-4,5-P 2 is slightly reduced but not reversed by the presence of Mn 2ϩ . This conclusion is evident when the data are expressed as a percentage of the values obtained with pure PtdSer vesicles and compared with similar analyses performed with the data from Fig. 4. The lipids and the relative transport activities in the presence of Mn 2ϩ are as follows: PtdSer, 100%; PtdCho, 47%; PtdOH, 97%; PtdEtn 48%, PtdGro, 45%; PtdIns, 47%; PtdIns-4-P, 60%; and PtdIns-4,5-P 2 , 62%. The Mn 2ϩ is not absolutely required for events at the acceptor membrane but can increase the apparent efficiency of PtdSer transport. The action of Mn 2ϩ is insufficient to overcome the inhibitory effects of the surface dilution of PtdSer.
PtdSer Transport from Defined Liposomes to Psd2p Is Constrained by the Same Genetic and Molecular Factors Observed in Vivo-In vivo studies identify PstB2p and the C2 domain of Psd2p as essential elements for the transfer of nascent PtdSer to the locus of Psd2p (7,8,26). These requirements are recapitulated in permeabilized cell systems used for examining PtdSer transport (8,26). If the reconstituted transport system from defined donor membranes described in this report is a valid model for interorganelle PtdSer transport, it should also conform to the requirements seen in vivo. In order to test the validity of the transport reactions described above, we prepared acceptor membranes from strains harboring a C2 domain deletion (psd2-C2⌬) in Psd2p and from strains with a null allele (pstB2⌬) for PstB2p. The units of Psd2p activity present in these membrane samples were determined independently of lipid transport using NBD-Ptd[1-14 C]Ser. Subsequently, we determined the transport-dependent decarboxylation of Ptd[1- 14 C]Ser using the 200 M liposome donors (400-nm diameter) described above. Fig. 8, A and B, show the results of the experiments. The findings are expressed as the ratio of transportdependent decarboxylation to decarboxylation determined with the NBD substrate. Since NBD-Ptd[1Ј-14 C]Ser spontaneously partitions into the acceptor membrane (27,29), it serves as a transport-independent measure of Psd2p activity. By using this ratio, it is possible to correct for any variation in the amount of Psd2p present in mutant strains relative to the corresponding wild type strains. The transport of PtdSer from liposomes to acceptor membranes containing the mutant Psd2p, lacking a C2 domain, is negligible. Likewise, the transport of PtdSer from liposomes to membranes lacking PstB2p is also at the limits of detection. The data clearly show that transport of PtdSer from liposomes to Psd2p requires the presence of both the C2 domain of Psd2p and PstB2p on the acceptor membranes. The data further demonstrate that the transfer of PtdSer to the acceptor membranes is a specific process and not a consequence of nonspecific fusion events between donors and acceptors. From these data, we conclude that the acceptor membrane properties that are required for PtdSer transport in vivo and in permeabilized cells are also required for lipid transfer from defined donor membranes to biological acceptors in vitro.

DISCUSSION
In this paper, we describe and characterize a defined donor membrane system that is competent to transport PtdSer to the locus of Psd2p for decarboxylation to form PtdEtn. The establishment of this donor membrane system is an important first step toward elucidating the minimum required components involved in interorganelle PtdSer transport and identifying their mechanisms of action. This system now provides important new insights into structural elements of both the donor and acceptor membranes that are important for membrane recognition and PtdSer transport.
The first set of experiments in this study was designed to address discrepancies among permeabilized cells, isolated organelles, and liposomes in the transport-dependent decarboxylation of PtdSer. Successful donor/acceptor membrane systems have previously been described for nascent PtdSer transport to the locus of Psd2p using biological membranes (7,8). However, liposomes derived from the donor membranes consistently failed to function as donors in the transport reaction. These prior results along with the lack of success using small unilamellar liposomes of PtdSer, prepared by sonication, suggested that specific protein components were required in the donor membranes to effect PtdSer transport. However, further manipulation of the liposomes revealed that both the curvature and the composition of the donor membrane were important factors in producing competent donor membranes. The data in Figs. 1 and 2 demonstrate that the optimum donors are PtdSer liposomes with diameters of Ն400 nm. There is a significant reduction in PtdSer transport when the diameters of the liposomes are reduced. The dependence of the transport upon donor membrane diameter was unanticipated. However, this preference for a more planar surface on the donor membrane may reflect the donor/acceptor membrane alignments found in other membranes that occur at zones of apposition seen between the mitochondria and ER (3,10,14,28), between the plasma membrane and ER (3), or between the Golgi and the ER (30). In yeast, these zones of apposition appear to be up to 40 nm in length between ER and mitochondria and 40 -100 nm in length between ER and the plasma membrane. For spherical donor membranes with a 50-nm diameter, a zone of apposition of 40 nm would require alignment with 25% of the crosssectional circumference that would be described by an arc of 90°. In contrast, for spherical donor membranes of 400-nm diameter, a zone of apposition of 40 nm would require alignment with 3% of the cross-sectional circumference described by an arc of just 10°. These physical aspects of donor and acceptor membranes may be important features that affect the recognition, binding affinity, and transfer of lipids and are plausible explanations of the vesicle curvature-dependent differences in transport that we observe in our reconstituted system. A second unanticipated finding described in this report is the relatively stringent dependence of the transport system upon the surface concentration of PtdSer. Our studies using homotypic and heterotypic liposomes described in Fig. 5 provide strong evidence that surface concentration and dilution of Ptd-Ser are critical properties affecting lipid transport and that the process is not subject to nonspecific inhibition by the addition of extraneous lipids to the in vitro reaction. As shown in Fig. 3C, the decline of PtdSer transport occurs exponentially with the surface dilution of the lipid. The data implicate high local concentrations of PtdSer as important features of the lipid transport process. These high concentrations could be essential requirements for acceptor membrane recognition of donors and for docking of the membranes prior to transport of the lipid. Alternatively, high surface concentrations of PtdSer may be required to drive the action of inefficient transporters that are In A, the acceptor membranes were generated from yeast strains harboring psd1⌬ psd2⌬ alleles and expressing PSD2 (wild type) or PSD2/C2⌬ from a YEp352 plasmid as indicated. In B, acceptor membranes were generated from strains harboring a psd1⌬ allele and a chromosomal copy of PSD2 in conjunction with either a chromosomal copy of PSTB2 (wild type) or the corresponding null allele pstB2⌬. Values are means Ϯ S.E. for three experiments. assembled on the acceptor membrane. It is noteworthy that studies of PtdSer metabolism to PtdEtn in mammalian cells describe the preferential utilization of newly synthesized Ptd-Ser for transport between the endoplasmic reticulum and mitochondria (31,32). It is possible that newly synthesized pools of this lipid are highly concentrated in specialized subdomains that are uniquely competent for transport. Although surface dilution of PtdSer by many lipids inhibits transport, PtdOH can replace up to 30% of the PtdSer without loss of transport efficiency. It is not yet clear if PtdOH can simply substitute for PtdSer or if it plays a more important role in promoting interactions between donor and acceptor membranes. PtdOH is centrally positioned in the regulation of membrane trafficking events insofar as its production is regulated by ARF1 through stimulation of phospholipase D (33,34), and its increased production stimulates the production of PtdIns-4,5-P 2 (35), which in turn can regulate secretory and endocytic vesicle traffic (36). Both PtdOH and PtdIns-4,5-P 2 also play an important role in maintaining the Golgi apparatus and preventing its fragmentation (37). Our current results with PtdOH suggest that it may also play an important role in regulating PtdSer traffic.
A critical test of the significance of any in vitro reconstitution system is its fidelity to the in vivo situation and permeabilized cell systems. In previous work, we described mutant strains of yeast that are defective in transporting nascent PtdSer to the locus of Psd2p (7,8,24,26). These studies identify the PtdIns transfer/binding protein PstB2p as essential for PtdSer transport (7). Additional studies also identify the C2 domain of Psd2p as essential for PtdSer transport in vivo (26). In vitro studies with permeabilized cells corroborate the in vivo findings about PstB2p and the C2 domain of Psd2p. Furthermore, studies with isolated organelles reveal the same properties for PstB2p found in both permeabilized cells and living cells (8). The studies described above have led to a model for PtdSer transport that requires the presence of PstB2p and the C2 domain of Psd2p on the acceptor membrane, as essential elements required for PtdSer transport to the Psd2p enzyme for catalysis (16). In the current study, we have endeavored to replace the biological donor membrane with defined liposomes. Our data demonstrate that synthetic donor membranes with relatively planar characteristics and a high surface concentration of PtdSer can productively interact with biological acceptor membranes in the transfer of lipid. The fidelity of this artificial donor system to the characteristics of previously described in vivo, permeabilized cell, and isolated organelle systems was demonstrated in several experiments. Both the permeabilized cell system and the isolated organelle system require Mn 2ϩ for optimal PtdSer transport. In addition, all of the PtdSer transport studies involving Psd2p described previously require the presence of the C2 domain of Psd2p and PstB2p on the acceptor membrane (8,26). In Fig. 8, we demonstrate that the transport of lipid from the liposomes to Psd2p requires the presence of both the C2 domain of Psd2p and the presence of PstB2p. These latter results are extremely important for multiple reasons. First, the findings emphatically demonstrate that there is no artifactual fusion occurring between the donor and acceptor membranes. We know this to be true, because we can quantify the transport-independent activity of Psd2p with the NBD-Ptd[1Ј-14 C]serine substrate. Since the NBD-modified substrate spontaneously inserts into the acceptor membrane without a requirement for the transport machinery, it provides us with a measure of activity to be expected if spontaneous fusion occurs between donor and acceptor membranes. Second, these results make it extremely unlikely that Psd2p is acting upon substrates while they are resident in the donor membrane. This conclusion is deduced from the findings that neither the C2 domain nor PstB2p are required for catalysis by Psd2p, and the activity of the enzyme is normal when the NBD substrate is inserted into the acceptor membrane. Despite this normal activity, the C2 domain and PstB2p are essential for catalysis of substrates that must be transported from other membranes to Psd2p either in vivo or in vitro (8,26).
We have put forward the hypothesis that PstB2p and Psd2p act in concert on the acceptor membrane to form part of a recognition and transport module that interacts with the donor (16). We anticipate that additional proteins and lipids on both the acceptor and donor membranes can also contribute to this complex. The previously demonstrated requirement for the PtdIns-4-kinase, Stt4p (24), suggests that polyphosphoinositides are required for some step in the overall transport process. In addition, this report also suggests a role for PtdOH in the process. We hypothesize that the net result is to form a macromolecular assembly that transiently bridges the two membranes and executes lipid transport. Morphological and biochemical studies support such a model for lipid transport between the ER and mitochondria (3, 10 -12, 14, 15) as well as the ER and plasma membrane (13). The findings in this report implicate specialized PtdSer-rich domains in the donor membrane as required elements for productive lipid transfer to the acceptor. These specialized domains might be a consequence of the action of other effectors such as lipid-binding proteins present in the donor membranes. The formation of PtdSer-rich domains may also be regulated by the polyphosphoinositides or PtdOH present in either the donor or the acceptor membrane.
In summary, we describe a new system for examining the interorganelle transport of PtdSer between defined synthetic donor membranes and biological acceptor membranes. The findings implicate PtdSer domains within the donor as critical elements required for transport of the lipid. PtdOH may also act in concert with PtdSer to facilitate docking and recognition between membranes. The in vitro studies also demonstrate that planar membranes function optimally in lipid transport, and these may further facilitate the formation of zones of apposition between donor and acceptor membranes. This system recapitulates the in vivo lipid transport process and is constrained by the same genetic and molecular elements required in living cells.