PtdSer (phosphatidylserine) is synthesized in the endoplasmic reticulum and the related MAM (mitochondria-associated membrane), and transported to the PtdSer decarboxylases, Pds1p in the mitochondria, and Psd2p in the Golgi. Genetic and biochemical analyses of PtdSer transport are now revealing the role of specific protein and lipid assemblies on different organelles that regulate non-vesicular PtdSer transport. The transport of PtdSer from MAM to mitochondria is regulated by at least three genes: MET30 (encoding a ubiquitin ligase), MET4 (encoding a transcription factor), and one or more unknown genes whose transcription is regulated by MET4. MET30-dependent ubiquitination is required for the MAM to function as a competent donor membrane and for the mitochondria to function as a competent acceptor membrane. Non-vesicular transport of PtdSer to the locus of Psd2p is under the control of at least three genes, STT4 [encoding Stt4p (phosphatidylinositol 4-kinase)], PSTB2 (encoding the lipid-binding protein PstB2p) and PSD2 (encoding Psd2p). Stt4p is proposed to produce a pool of PtdIns4P that is necessary for lipid transport. PstB2p and Psd2p must be present on the acceptor membrane for PtdSer transport to occur. Psd2p contains a C2 (Ca2+ and phospholipid binding sequence) domain that is required for lipid transport. Reconstitution studies with chemically defined donor membranes demonstrate that membrane domains rich in the anionic lipids, PtdSer, PtdIns4P and phosphatidic acid function as the most efficient donors of PtdSer to Psd2p. The emerging view is that macromolecular complexes dependent on protein–protein and protein–lipid interactions form between donor and acceptor membranes and serve to dock the compartments and facilitate phospholipid transport.
With the exception of the ER (endoplasmic reticulum), all organelles present within eukaryotic cells lack the biochemical machinery necessary to synthesize the full complement of their constituent phospholipids . This situation imposes a requirement for phospholipid transport for the biogenesis of all organelles, with the principal, but not sole exporter of these molecules being the ER. The essential role for phospholipid transport in membrane biogenesis also makes it a fundamental requirement for all cell growth, division and differentiation. Despite this central role for lipid transport in cell biology, very little is known about the genes and proteins that execute and regulate the phospholipid transport processes, and their mechanisms of action. One general feature of phospholipid transport that is being more broadly appreciated is its divergence from the transport pathways that are characteristic of membrane proteins [2–4]. Historically, a number of studies identified the intermembrane transport of PtdCho (phosphatidylcholine) , PtdEtn (phosphatidylethanolamine) [6,7] and PtdSer (phosphatidylserine) [8–10] as being distinctly different from routes followed by membrane proteins. The phospholipid whose inter-organelle transport is currently best understood is PtdSer. Reconstitution studies with permeabilized yeast cells and yeast organelles reveal that the transport of PtdSer between the ER and mitochondrion, or between the ER and Golgi, occurs independently of the presence of ATP and cytosolic factors [10–12]. These latter two features alone highlight pathways for phospholipid transport that must be uniquely different from the well-characterized vesicular routes followed by membrane proteins. Currently, there is much interest in zones of apposition that form between membranes as possible sites for phospholipid transport. These zones of apposition were first described in electron micrographs, but it was unclear whether these were random associations or stable junctions [13,14]. Reconstitution studies with permeabilized cells provided evidence that a stable, functional association of the ER with the mitochondria was required for PtdSer transport [9,15]. A subdomain of the ER, named the MAM (mitochondria-associated membrane), which formed stable complexes with mitochondria, was isolated and partially characterized [16–18]. The MAM is now recognized as an important functional and structural interface between a specialized region of the ER and mitochondria, but almost nothing is known about the molecules that forge the linkage between the organelles and how the association is regulated. The morphological connections between organelles are now being revisited with renewed interest in their stability and purpose. Additional organelle associations that have been described are ER–plasma membrane , ER–Golgi  and nuclear–vacuolar membrane . The present paper will focus on some of the genetic and biochemical features of the MAM–mitochondria and ER–Golgi interactions that participate in PtdSer transport in the yeast Saccharomyces cerevisiae.
PtdSer metabolism as a biochemical and genetic gateway for analysing phospholipid transport
PtdSer, PtdEtn and PtdCho are collectively known as the aminoglycerophospholipids, a group that comprises most of the phospholipids present in most eukaryotic membranes . In both mammalian and yeast cells, the aminoglycerophospholipids account for approx. 70–80% of the structural phospholipids found in organelle membranes. In yeast, the PtdSer synthase gene, PSS1, encodes the enzyme Pss1p that catalyses the formation of PtdSer from CDP-diacylglycerol and serine . The resultant PtdSer can be decarboxylated by either Psd1p (encoded by the PSD1 gene that complements psd1 mutations) or Psd2p (encoded by the PSD2 gene that complements psd2 mutations) [24,25]. The decarboxylation reactions produce PtdEtn. The PtdEtn contains a primary amine that can be methylated by the PtdEtn-methyltransferases, Pem1p and Pem2p, using S-adenosylmethionine as the methyl donor . The methylation reactions sequentially produce the mono-methyl [PtdEtn(Me)], di-methyl [PtdEtn(Me)2] and tri-methyl (PtdCho) derivatives of PtdEtn. In wild-type cells, PtdEtn(Me) and PtdEtn(Me)2 are present as trace level intermediates. 1 outlines these biosynthetic reactions.
Outline of aminoglycerophospholipid synthesis and transport
Although only three relatively simple classes of enzyme reactions define the entire aminophospholipid synthetic pathway (serine transfer, decarboxylation and methyl transfer), the overall process is made considerably more complex because the enzymes are geographically separated within the cell . The separation of substrates and enzymes for sequential steps in the synthetic pathway necessitates the transport of the phospholipids between organelles. As shown in 1, PtdSer synthesized in the ER/MAM must be transported to the Psd1p in the mitochondria or Psd2p in the Golgi, to be decarboxylated. Subsequently, the PtdEtn generated in either of these organelles must be exported to the ER to serve as a substrate for Pem1p and Pem2p methyltransferases to synthesize PtdCho. This level of complexity within the aminoglycerophospholipid synthetic pathway is actually an immense benefit for studying inter-organelle phospholipid transport because the transformation of one phospholipid to another (e.g. PtdSer to PtdEtn) provides a specific biochemical indicator of the transport processes. For example, in cells that are genetically manipulated to eliminate Psd2p (i.e. psd2Δ mutants), the only enzyme capable of decarboxylating PtdSer is the Psd1p found in mitochondria. Thus, in psd2Δ mutants incubated with [3H]-serine, the formation of [3H]PtdSer indicates lipid synthesis in the ER, and the subsequent formation of [3H]PtdEtn reports the transport of the PtdSer to the mitochondria. Furthermore, the ensuing formation of [3H]PtdCho reports the export of [3H]PtdEtn from the mitochondria to the ER. Conversely, in strains lacking Psd1p (i.e. psd1Δ mutants) incubated with [3H]serine, the formation of [3H]PtdEtn reports the transport of nascent [3H]PtdSer to the Golgi. In the same psd1Δ cell, the subsequent formation of [3H]PtdCho reports the export of PtdEtn from the Golgi to the ER.
The spatial separation of aminoglycerophospholipid transport to and from Psd1p and Psd2p has also facilitated a genetic analysis of the processes. For simplicity, the transport of PtdSer to Psd1p is generally referred to as the PSTA (PtdSer transport A) pathway; and the transport of the lipid to Psd2p is referred to as the PSTB (PtdSer transport B) pathway. Likewise, the transport of PtdEtn out of the mitochondria is referred to as the PEEA (PtdEtn export A pathway from mitochondria to the ER); and the transport of the lipid out of the Golgi is referred to as the PEEB (PtdEtn export B pathway from the Golgi to the ER). Genetic dissection of aminoglycerophospholipid transport is made possible by two conditions. The first condition is the discovery that PtdEtn is an essential lipid for growth of yeast [28,29]. The second condition is the existence of an alternative pathway for PtdEtn synthesis that utilizes Etn (ethanolamine) for the synthesis of PtdEtn . This latter pathway was first described by Kennedy and Weiss  almost 50 years ago and proceeds through the intermediates phosphoethanolamine and CDP-Etn. In principle, the Kennedy pathway enables one to disrupt the transport-dependent synthesis of PtdEtn by mutation under conditions where the PtdEtn requirement is met by Etn supplementation. Mutants in the PSTA, PSTB, PEEA and PEEB pathways can be identified in populations of cells that are auxotrophic for Etn.
Mutagenesis of yeast strains harbouring a psd2Δ allele, and recovery of Etn auxotrophs enabled the identification of a new mutant defective in PtdSer transport to the mitochondria . This mutant had the characteristics of a defect in the PSTA pathway, and was named pstA1 (mutation affecting PtdSer transport in the A pathway). The lesion in the pstA1 mutant was localized to transport events between the MAM and the outer mitochondrial membrane. Mitochondria isolated from the mutant had decreased levels of PtdEtn and a decreased phospholipid/protein ratio. The altered phospholipid/protein ratio of mutant mitochondria causes them to sediment at much higher density than wild-type mitochondria, when analysed by sucrose-density-gradient centrifugation. Reconstitution studies demonstrated that the defect in lipid transport observed in intact cells was also reproduced in transport assays with isolated organelles. Additional reconstitution studies using different combinations of MAM and mitochondria derived from wild-type and mutant cells revealed that the pstA1 mutation affected both organelles. Thus MAM from pstA1 cells was an incompetent donor of PtdSer to wild-type mitochondria; and mitochondria from pstA1 cells were incompetent acceptors of PtdSer generated in wild-type MAM. Thus the pstA1 mutation appears to cause intrinsic defects in both MAM and mitochondria that render them incompetent for lipid transport. The gene complementing the pstA1 defect was MET30 , which encodes a substrate recognition subunit of a multiprotein SCF (Skip1/Cullin/F box protein components) ubiquitin ligase . The pstA1 strain was shown to have a point mutation in the substrate recognition domain of Met30p (a ubiquitin ligase subunit encoded by the MET30 gene that complements the met30 mutation affecting methionine biosynthesis). In addition, Met4p (a transcription factor encoded by the MET4 gene that complements the met4 mutation affecting methionine biosynthesis), which regulates the expression of multiple genes in methionine biosynthesis, has previously been known to be a substrate for Met30p-directed ubiquitination . Analysis of pstA1 mutants reveals that they are also defective in the ubiquitination of Met4p . The action of Met30p causes ubiquitination of Met4p and its inactivation as a transcription factor. Loss of ubiquitination of Met4p causes it to be constitutively activated. These findings are consistent with a model for PtdSer transport from MAM to mitochondria that is negatively regulated by active Met4p (see Figure 1). Inactivation of Met4p, by Met30p-mediated ubiquitination, alleviates the inhibition of PtdSer transport. Conversely, relevant mutations in Met30p prevent the inactivation of Met4p, and thus fail to alleviate the inhibition of PtdSer transport.
Met30p regulation of PtdSer transport between MAM and mitochondria
A summary of the protein and lipid participants in the PSTB pathway is shown in Figure 2. The mutagenesis of strains containing a psd1Δ allele and recovery of Etn auxotrophs provided a method for isolating new strains with defects in the PSTB pathway [34–36]. All of the mutants discovered were defective in converting newly synthesized PtdSer into PtdEtn, but had normal Psd2p activity. Two mutants, pstB1 and pstB2, were identified by the genetic screen, and a variant of psd2 lacking its C2 (Ca2+ and phospholipid binding sequence) domain (i.e. psd2C2Δ) was deliberately constructed. The pstB1 mutant was complemented by the gene STT4, which encodes a phosphatidylinositol 4-kinase . The pstB2 mutant was complemented by the gene, PSTB2, which encodes a PtdIns binding protein with homology to the PtdIns/PtdCho transfer protein Sec14p (PtdIns/PtdCho transfer protein that is encoded by the SEC14 gene that complements sec14 mutant affecting protein secretion) . The PSTB2 gene has also been identified in other genetic screens and given the designations PDR17 (pleiotropic drug resistance 17) and SFH4 (SEC14 homologue 4) [37,38].
Lipid and protein participants in the PSTB pathway for PtdSer transport
The role played by Stt4p in PtdSer transport is not completely clear. The protein is hypothesized to generate a pool of PtdIns4P on the ER that may serve as a recognition site for docking with the C2 domain of Psd2p and perhaps PstB2p and other proteins. One pool of Stt4p has been localized to the plasma membrane where it is tethered by the protein Sfk1p (suppressor of 4-kinase 1) that acts as a high copy suppressor of temperature-sensitive stt4 mutants . A second pool of Stt4p may also be associated with the ER protein Scs2p (suppressor of choline sensitivity 2) that is an integral membrane protein, which also functions as a tethering protein . Transient localization of Stt4p to the ER may serve to create localized regions of PtdIns4P that could function in protein recruitment and the formation of docking complexes between membranes.
Since PstB2p is a member of the PtdIns/PtdCho transfer protein family, it would seem likely that it could function as a soluble carrier of PtdSer between the ER and Golgi . However, the protein failed to transfer PtdSer between liposomal donor and acceptor membranes in vitro. PstB2p is amphitrophic and occurs in both soluble and membrane-bound forms . The mechanism of membrane attachment is unknown. Reconstitution studies demonstrated that PstB2p must be bound to the Golgi membrane for PtdSer transport from the ER to the Golgi to occur [12,41]. Although PstB2p does not act as a soluble PtdSer carrier, it has the properties expected of a protein that can reversibly attach to membranes and form complexes with protein and/or lipids to form docking and transport complexes between membranes.
The Psd2p enzyme contains a catalytic α-subunit and a non-catalytic β-subunit [36,42]. Within the β-subunit is a C2 domain . C2 domains participate in protein–lipid and protein–protein interactions [43,44]. To probe the functions of this C2 domain, it was deleted from the gene, and a psd2-C2Δ (mutation in which the C2 domain of Psd2p is deleted) strain was created, and the effects on PtdSer transport and decarboxylation were examined . Measurement of catalytic activity revealed that the C2 domain of Psd2p is not necessary for enzyme activity. The C2 domain of Psd2p is also not required for correct subcellular localization of the enzyme. However, reconstitution experiments, using either biological membranes or liposomes as donors, reveal that the C2 domain of Psd2p participates in the transport of PtdSer between donor membranes and the Golgi [12,41]. These studies strongly suggest that the β-subunit of Psd2p may co-operate with PstB2p in forming a complex that participates in moving PtdSer between membranes.
In addition to proteins, phospholipids also appear to play a significant role in regulating PtdSer transport in the PSTB pathway. As described above, the phosphatidylinositol 4-kinase, Stt4p, is required for PtdSer transport to Psd2p . The product of the reaction, PtdIns4P, and other polyphosphoinositides can function to mark organelle identity and recruit protein complexes from the soluble compartment to membrane surfaces . The same type of surface signals could, in principle, also be used to recruit protein complexes on one membrane to dock with a second membrane. We are currently testing the hypothesis that an acceptor membrane complex containing Psd2p and PstB2p may recognize both proteins and lipids on a donor membrane to form stable regions of membrane apposition, where intermembrane PtdSer transport can occur. Further evidence for lipid involvement in PtdSer transport comes from reconstitution studies that use liposomes of defined composition as donor compartments for transfer of PtdSer to membranes containing Psd2p . These reconstitution experiments reveal a remarkable preference for donor membranes that contain very high concentrations of PtdSer. Surface dilution of liposomal PtdSer by 50% with PtdCho almost completely abrogates PtdSer transport between the membranes. However, if the PtdSer is diluted with PtdOH (phosphatidic acid), there is no loss of transport efficiency. In addition, if PtdSer is diluted with PtdIns4P, there is only a partial loss of PtdSer transport. These findings suggest that local high concentrations of PtdSer alone, or PtdSer along with PtdOH and PtdIns4P, can be recognized by transport machinery on the acceptor membrane. How such domains of anionic lipid might form in vivo remains uncertain. Protein–protein interactions between donor and acceptor membranes might activate a system for corralling PtdSer and other anionic phospholipids. Alternatively, domains of anionic phospholipids within the donor membrane may trigger recognition and docking by the acceptor membrane and subsequent engagement of the transport machinery.
The combined genetic and biochemical analysis of PtdSer transport in yeast is beginning to provide some new insights into the mechanisms of non-vesicular lipid transport in eukaryotes. The emerging data support models in which donor and acceptor membranes interact through molecular recognition of proteins and lipids in juxtaposed bilayers. These interactions are proposed to form the core structures for recruiting additional molecules that constitute a final lipid and protein assembly competent to move phospholipids between bilayers. We currently know of only a few of the molecular participants. For the PSTA pathway, we must now focus on identification of the molecular components present on the MAM and mitochondria that are responsible for promoting the interactions between the membranes. For the PSTB pathway, additional progress will be made by identifying the full repertoire of proteins and lipids that interact with Psd2p and PstB2p. Molecular dissection of the role of the C2 domain of Psd2 and domains of PstB2p will also provide important insight into how these proteins work in concert.
Non-Vesicular Intracellular Traffic: Biochemical Society Focused Meeting held at Goodenough College, London, U.K., 15–16 December 2005. Organized and edited by S. Cockcroft (University College London, U.K.) and T. Levine (Institute of Ophthalmology, London, U.K.).
Ca2+ and phospholipid binding sequence
PtdEtn export A pathway from mitochondria to the ER
PtdEtn export B pathway from the Golgi to the ER
- PSTA pathway
PtdSer transport A pathway
- PSTB pathway
PtdSer transport B pathway
Skip1/Cullin/F box protein components of ubiquitin ligase
This work was supported by the National Institutes of Health MERIT Award 2R37GM32453.