Phosphatidylserine transport to the mitochondria is regulated by ubiquitination.

Mitochondrial membrane biogenesis requires the interorganelle transport of phospholipids. Phosphatidylserine (PtdSer) synthesized in the endoplasmic reticulum and related membranes (mitochondria-associated membrane (MAM)) is transported to the mitochondria by unknown gene products and decarboxylated to form phosphatidylethanolamine at the inner membrane by PtdSer decarboxylase 1 (Psd1p). We have designed a screen for strains defective in PtdSer transport (pstA mutants) between the endoplasmic reticulum and Psd1p that relies on isolating ethanolamine auxotrophs in suitable (psd2Delta) genetic backgrounds. Following chemical mutagenesis, we isolated an ethanolamine auxotroph that we designate pstA1-1. Using in vivo and in vitro phospholipid synthesis/transport measurements, we demonstrate that the pstA1-1 mutant is defective in PtdSer transport between the MAM and mitochondria. The gene that complements the growth defect and PtdSer transport defect of the pstA1-1 mutant is MET30, which encodes a substrate recognition subunit of the SCF (suppressor of kinetochore protein 1, cullin, F-box) ubiquitin ligase complex. Reconstitution of different permutations of MAM and mitochondria from wild type and pstA1-1 strains demonstrates that the MET30 gene product affects both organelles. These data provide compelling evidence that interorganelle PtdSer traffic is regulated by ubiquitination.

Mitochondrial membrane assembly requires the import of proteins and lipids. Whereas much is known regarding mitochondrial protein import (1,2), little is known at the molecular level about mitochondrial import of phospholipids (3). Mitochondrial lipid import is thought to be essential for biogenesis of this organelle in all eukaryotic organisms. In addition to importing lipids, mitochondria also export significant quantities of PtdEtn to other organelles. Elucidation of the mechanisms of lipid import into and export out of mitochondria and the gene products that participate in the process remains a fundamental problem of cell biology and biochemistry.
Our approach to understanding mitochondrial lipid traffic has been to focus on aminoglycerophospholipid transport events in the de novo biosynthetic pathway of Saccharomyces cerevisiae. Fig. 1 depicts the synthesis of aminoglycerophospho-lipids in S. cerevisiae. PtdSer 1 is synthesized from serine and CDP-diacylglycerol by PtdSer synthase 1 (Pss1p) in the ER and MAM (4). Subsequently, PtdSer is transported to the site of PtdSer decarboxylase 1 (Psd1p) in the inner mitochondrial membrane, where it is converted to phosphatidylethanolamine (PtdEtn) (5)(6)(7)(8). This routing of PtdSer appears true for all eukaryotes. In yeast, PtdSer can also be decarboxylated at the locus of PtdSer decarboxylase 2 (Psd2p) in the vacuole and Golgi compartments (9,10). We have chosen to name the Ptd-Ser transport pathways PSTA and PSTB to designate traffic to the mitochondria and the Golgi/vacuole, respectively. PtdEtn can be exported from the loci of Psd1p and Psd2p, and we have named these pathways PEEA and PEEB (denoting PtdEtn export), respectively. The exported PtdEtn can be used for membrane synthesis directly, or it can be methylated in the ER by the PtdEtn methyltransferase enzymes, Pem1p and Pem2p, to form phosphatidylcholine (PtdCho) (11). PtdEtn and PtdCho can also be synthesized via the Kennedy pathways from the precursors ethanolamine or choline (3). Both PtdEtn and Ptd-Cho are thought to be essential for S. cerevisiae growth in medium that contains glucose or nonfermentable carbon sources (5,(12)(13)(14).
In S. cerevisiae, Psd1p is the major PtdSer decarboxylase, producing most of the cellular PtdEtn in the absence of an ethanolamine precursor (10). Whereas PtdEtn made by Psd1p appears to be important for mitochondrial function, it is also important for other cellular functions. For example, a Chinese hamster ovary cell mutant defective in intramitochondrial transport of PtdSer has significantly reduced cell surface Ptd-Etn levels, resulting in a defect in contractile ring disassembly during cytokinesis (15,16). PtdEtn is also the donor of ethanolamine phosphate to glycosylphosphatidylinositol anchors, whose synthesis is essential for yeast cell viability (17,18). Furthermore, PtdEtn plays a central role in S. cerevisiae lysosome/vacuole autophagy by covalently conjugating to Apg8p (19). Hence, the question of how PtdSer is transported from the ER to the inner mitochondrial membrane for PtdEtn production is also an issue of major importance for cellular functions other than mitochondrial biogenesis.
Transport of PtdSer from the ER/MAM to mitochondria is thought to occur through close physical association of the membranes of the organelles (6, 20 -24). The exact mechanism of translocation is unknown, but previous studies suggested transport is mediated by MAM and involves proteins (4,21,22).
At present, no protein or gene product has been identified that mediates or regulates ER to mitochondria membrane association and/or transport of PtdSer between these organelles. Hence, the goals of our study were to 1) isolate new aminophospholipid transport mutants in the PSTA pathway, 2) characterize these transport mutants biochemically, and 3) identify the gene that complements transport-defective mutants.
In this paper, we describe the isolation of a new ethanolamine auxotrophic yeast strain, in a psd2⌬ genetic background, with the characteristics of a pstA mutation. Biochemical analyses demonstrate the presence of a defect in the transport of PtdSer to the mitochondria. The pstA defect is complemented by the MET30 gene encoding an essential protein in yeast that has a key role in the ubiquitination of specific substrates.

EXPERIMENTAL PROCEDURES
Chemicals-Media components for yeast and bacterial growth were purchased from Difco, U. S. Biological, Sigma, and Fisher. The radioisotopes, L-[3-3 H]serine and L-[1-14 C]serine, were purchased from Amersham Biosciences. Phosphatidyl[1Ј- 14 C]serine was synthesized from L-[1- 14 C]serine and CDP-diacylglycerol (5). Phospholipid standards and CDP-diacylglycerol were purchased from Avanti Polar Lipids. Thin layer silica gel H plates were obtained from Analtech Corp.
Whole Cell Radiolabeling and Phospholipid Analysis-Yeast strains were grown in synthetic lactate plus ethanolamine medium to midlog phase. Cultures were then washed twice with sterile water and suspended in SL medium minus serine to an A 600 of 0.2 in a volume of 2 ml. Radiolabeling was initiated by adding 10 Ci/ml L-[3-3 H]serine (32 Ci/mmol). Labeling proceeded at 30°C for 4 h with vigorous shaking. Growth and metabolism were arrested by the addition of trichloroacetic acid to a final concentration of 5% (w/v), along with ϳ10 mg of unlabeled carrier cells. Cell pellets were washed twice by centrifugation with ice-cold water. Lipids were extracted as previously described (5,9) and separated by TLC on Silica Gel H plates, using the solvents chloroform, methanol, 2-propanol, 0.25% KCl, triethylamine (30:9:25:6:18, v/v/v/v/ v). Phospholipid standards were visualized by spraying the TLC plates with 0.1% 8-anilino-1-naphthalene sulfonic acid, followed by exposure to UV light. Separated lipid spots were scraped into 0.5 ml of H 2 O with 4.5 ml of ScintiSafe 30% scintillation fluid. (Fisher). Radioactivity of the samples was measured by liquid scintillation spectrometry (Beckman catalog no. LS6500).
In Vitro Radiolabeling and Characterization of Crude Mitochondria-Crude mitochondria were prepared as described (25) from 1-liter cultures grown to early log phase in semisynthetic lactate medium supplemented with uracil (20 mg/liter), L-leucine (100 mg/liter), L-histidine (20 mg/liter), L-lysine (30 mg/liter), and ethanolamine (3 mM) at 30°C. Crude mitochondrial pellets were suspended in buffer B (0.6 M sorbitol, 20 mM K ϩ MES, pH 6.0) and assayed immediately for protein (Bio-Rad protein assay). Intactness of the outer membrane was assessed by measuring cytochrome c oxidase activity (27). Pss1p and Psd1p enzyme activities of the crude mitochondrial preparations were assayed as described previously (5,10). Reconstituted aminophospholipid synthesis and transport were measured using [3-3 H]serine incorporation into lipids as outlined by Achleitner et al. (6) with a few modifications. Briefly, crude mitochondria were diluted in buffer C (0.6 M sorbitol, 20 mM K ϩ HEPES, pH 7.4) to a protein concentration of 1 g/l. The reaction contained 50 l of crude mitochondria, 16.7 l of 6 mM MnCl 2 /buffer C, and 113.3 l of buffer C. Samples were equilibrated to 30°C in a shaking water bath before initiating the reaction by adding 20 Ci (1 Ci/l) of L-[3-3 H]serine (32 Ci/mmol). After 15 min, 4 l of 0.2 M EDTA in buffer C was added (final concentration ϳ4 mM). Reactions were stopped after an additional 75 min by the addition of 1.5 ml of methanol, 1.5 ml of chloroform, and 1.2 ml of 0.2 M KCl. Zero time points were made by adding L-[3-3 H]serine and methanol simultaneously to the above mixture. Lipids were extracted by vortexing the samples, followed by centrifugation to separate organic and aqueous phases. The chloroform phase was washed twice, each time with 2.8 ml of practical upper phase solution (500 ml of methanol plus 450 ml of phosphatebuffered saline, pH 7.4, saturated with 50 ml of chloroform). The chloroform phase was collected and dried under a nitrogen stream. Samples were suspended in chloroform and separated by TLC on silica Gel H plates.
Intramitochondrial Transport of PtdSer-The radiolabeled lipid analog 1-acyl-2-(N-4-nitrobenzo-2-oxa-1,3-diazole)-aminocaproyl-phosphatidyl[1Ј-14 C]serine (NBD-Ptd[1Ј-14 C]Ser) was used to assess transport between the outer and inner mitochondrial membranes. The analog was synthesized as described previously (28). Purified mitochondria were prepared on Nycodenz gradients (25). The transport reactions contained 0.5-8 g of mitochondrial protein and 0.02 Ci of NBD-Ptd[1Ј-14 C]Ser (55 Ci/mol) in a volume of 400 l. The reactions were performed at 30°C for 20 min in a sealed tube containing 2 M KOHimpregnated filter paper for trapping 14 CO 2 . The reactions were terminated by the addition of 0.5 ml of 0.25 M H 2 SO 4 through the gas-tight seal with a syringe and needle. The 14 CO 2 released from the reaction that was trapped on the filter paper was quantified by liquid scintillation spectrometry.
In Vitro Reconstitution of Lipid Traffic between Purified MAM and Mitochondria-Crude mitochondria were isolated from 4.5 liters of culture as described above. Subsequent isolation of purified mitochondria and MAM was performed as described by Achleitner et al. (6). Briefly, crude mitochondria suspended in buffer B were layered on a density gradient composed of 20 -50% sucrose constructed in 1-ml steps differing in 3.3% sucrose increments in buffer B. The gradient was centrifuged at 100,000 ϫ g for 1 h at 4°C using an SW 41 rotor. Mitochondria were harvested from the lower third of the gradient, diluted 3-4-fold in buffer B, and pelleted for 10 min at 12,000 ϫ g at 4°C. MAM was collected from the top of the gradient, diluted 3-4-fold in buffer B, and pelleted for 1 h at 100,000 ϫ g at 4°C. Resulting mitochondrial and MAM pellets were suspended in buffer B and buffer C, respectively, and thoroughly homogenized with a plastic pestle. Freshly isolated mitochondria and MAM were immediately assayed for protein and L-[3-3 H]serine incorporation into aminophospholipids as outlined by Achleitner et al. (6) with a few modifications. Freshly purified mitochondria were also assayed for intactness of the outer membrane by measuring cytochrome c oxidase activity. In a 200-l reaction, MAM and purified mitochondria were reconstituted at final protein concentrations of 0.05 and 0.85 mg/ml, respectively. Radiolabeling was conducted in the presence of 0.2 mM CDP-diacylglycerol. Subsequent lipid analysis was performed as described above.
Mitochondrial Phospholipid Content-For measuring mitochondrial phospholipids, cells were first grown to log phase in synthetic lactate medium plus ethanolamine at 30°C. The cells were next washed twice with sterile water, suspended in synthetic lactate medium without ethanolamine, and incubated with shaking at 30°C, for 14 -18 h before harvesting for mitochondrial isolation. Crude mitochondria were prepared by the method of Glick and Pon (25). Lipids were extracted from ϳ2 mg of crude mitochondrial protein, in 4 ml of methanol, 4 ml of chloroform, 3.2 ml of 0.2 M KCl. The chloroform phase was washed twice with 7.6 ml of practical upper phase without phosphate solution (500 ml of methanol plus 450 ml of 0.2 M KCl, saturated with 75 ml of chloroform). The organic phase was collected and dried under a nitrogen stream. Phospholipids were separated by two-dimensional TLC on Silica 60 plates. The first solvent system contained chloroform/methanol/ acetic acid (13/5/2 by volume), and the second solvent system contained chloroform/methanol/formic acid (13/5/2 by volume). Lipids were visualized by staining with iodine vapor and scraped into glass tubes. Phosphorus was quantified using the method of Rouser et al. (29).
Mitochondrial Protein Import-Mitochondria were prepared from parental and pstA1-1 strains grown in semisynthetic lactate medium supplemented with 3 mM Etn using the methods of Glick and Pon (25). Protein import was measured using the methods of Rospert and Schatz (30). The Psd1p was expressed with a carboxyl-terminal V5 epitope, from a pYES plasmid (Invitrogen), under T7 promoter regulation. The Su9-DHFR was expressed from a plasmid under SP6 promoter regulation. Both proteins were synthesized in vitro using the appropriate TNT-coupled transcription/translation system (Promega) in the presence of [ 35 S]methionine. The radiolabeled protein mixtures were incubated with isolated mitochondria in either the absence or presence of valinomycin (1 g/ml) in an import buffer containing 0.6 M sorbitol, 50 mM HEPES-KOH, 50 mM KCl, 0.75 mg/ml methionine, 10 mM MgCl 2 , 1 mg/ml bovine serum albumin, 2.5 mM EDTA, and 2 mM KH 2 PO 4 adjusted to pH 7.0. The reactions also contained 2 mM ATP and 10 mM NADH and were conducted for 3 or 10 min at 30°C. The nonimported proteins were removed from the reactions by treatment with 100 g/ml trypsin for 30 min at 0°C. The action of trypsin was arrested with 200 g/ml soybean trypsin inhibitor. Following the trypsin treatment, the mitochondria were harvested by centrifugation, precipitated in 5% trichloroacetic acid, heat-inactivated, and finally resuspended for loading onto electrophoretic gels. The samples were electrophoresed using 10% Tricine polyacrylamide gels (Novex). The gels were processed through 5% boiling trichloroacetic acid, neutralized, and impregnated with 1 M salicylate before drying. The radiolabeled proteins were visualized using a Storm 860 PhosphorImager. Quantification of protein import was performed using ImageQuant 5.2 software from Amersham Biosciences.
Analysis of Met4p Ubiquitination-The MET4 gene modified to contain multiple C-terminal myc epitope tags was expressed from the chromosome under control of its endogenous promoter (31). The modified MET4 gene was expressed in strains PY725 and PY743 constructed by Dr. P. Kaiser (University of California, Irvine). In addition, we constructed the strains JCY474 (MATa ura3⌬0 leu2⌬0 his3⌬0 psd2⌬::KAN r trp1::HIS3 MET4-(18myc)::TRP1) and JCY475 (MATa ura3⌬0 leu2⌬0 his3⌬0 psd2⌬::KAN r trp1::HIS3 met30-P522L MET4-(18myc)::TRP1) from MSY30 and EAL18, respectively. The yeast strains examined were grown in either SC or SL medium containing 5 mM methionine and supplemented with Etn as indicated. The cells were harvested by centrifugation, and 5% boiling trichloroacetic acid was added to the pellet followed by vortex mixing for 5 s. The acid-treated cells were chilled on ice and sedimented by centrifugation. The resultant pellets were washed twice by resuspension in distilled water and centrifugation. The final pellets were rapidly frozen in liquid nitrogen and stored at Ϫ80°C prior to further processing. For processing, the cell pellets were suspended in an extraction buffer consisting of 8 M urea, 200 mM NaCl, 100 mM Tris-HCl (pH 7.5), 10 mM sodium pyrophosphate, 5 mM EDTA, 5 mM EGTA, 50 mM NaF, 0.1 mM orthovanadate, 1 mM phenylmethylsulfonyl fluoride, and an aliquot of Sigma anti-protease mixture (5 l/100 mg, wet cell weight). Glass beads were added to the samples, and the mixture was homogenized in a bead beater for 70 s. Cell debris was removed by centrifugation at 13,000 ϫ g ϫ 15 min at 20°C. The samples were next diluted to 4 M urea before separation on 8% Tris-glycine polyacrylamide gels. The myc epitope-tagged Met4p was detected using the mouse monoclonal antibody 9E10 (COVANCE) and a secondary goat anti-mouse antibody (Bio-Rad). Antibody reactivity was detected using enhanced chemiluninescence. Isoforms of Met4p were quantified using image scanning and NIH Image 1.62 software.

Isolation and Characterization of an Etn Auxotroph, EAL
18 -Yeast strains that have a psd2⌬::KAN r allele depend on Psd1p for the majority of PtdEtn and PtdCho synthesis in the absence of exogenous ethanolamine or choline (see Fig. 1). We reasoned that mutants with defects in PtdSer import into and PtdEtn export from the locus of Psd1p in the inner mitochondrial membrane would require ethanolamine for growth in a psd2⌬::KAN r genetic background. To generate mitochondrial phospholipid transport mutants, we mutagenized a psd2⌬::KAN r strain, MSY30, and selected for ethanolamine auxotrophs on medium containing lactate as a carbon source. Lactate medium was used to avoid selection of respirationdeficient petites. Screening of the first 10,000 colonies yielded eight ethanolamine auxotrophs, of which two were psd1 mutants and one was an apparent pss1 mutant, as determined by genetic complementation and biochemical enzyme assay. One of the remaining strains, EAL18, displayed a definitive requirement for ethanolamine for growth in liquid SL medium at 30°C, as shown in Fig. 2. In the absence of Etn, the EAL18 strain undergoes two doublings and then arrests. In the presence of ethanolamine, the mutant, EAL18, grows similarly to the parental strain, MSY30, although somewhat slower. The growth of the parental strain is essentially the same in the presence and absence of Etn. At 36°C, EAL18 shows very little or no growth in synthetic lactate medium, regardless of ethanolamine supplementation. In addition, EAL18 is not an ethanolamine auxotroph in synthetic glucose medium at 30°C. As shown in Fig. 3, choline supplementation does not rescue the growth defect of EAL18 in synthetic lactate medium at 30°C, indicating a stringent requirement for Etn under these conditions. Tetrad analysis yielded a 2:2 segregation of ethanolamine auxotrophy consistent with a mutation at a single locus. Diploids that are heterozygous for the mutation are not ethanolamine auxotrophs, indicating the mutation is recessive. The above properties of the mutant are similar but not identical to psd1⌬psd2⌬ strains and are consistent with defects in PtdEtn synthesis.
To test whether the ethanolamine auxotrophic mutant had a defect in the formation of aminoglycerophospholipids, cells were labeled with [ 3 H]serine in SL medium. Incorporation of Mutant EAL18, and parental strain MSY30, were initially grown in SL medium containing 3 mM Etn to early log phase. Cells were harvested by centrifugation, washed twice in sterile water, and reinoculated to give an absorbance (A 600 ) of ϳ0.02 in SL medium. Growth at 30°C was monitored by measuring A 600 in SL medium minus (Ⅺ, E) or plus (f, q) 3 mM Etn.
FIG. 1. Synthesis and transport of aminoglycerophospholipids in yeast. In the absence of Etn and choline (Cho), aminoglycerophospholipid synthesis commences with the formation of PtdSer from serine and CDP-diacylglycerol (CDP-DAG) substrates catalyzed by Pss1p. The PtdSer can be transported to the loci of the PtdSer decarboxylases, Psd1p and Psd2p, for the formation of PtdEtn in either mitochondria or the Golgi/vacuole. The PtdEtn can be exported from these organelles to the ER for methylation by Pem1p and Pem2p to form PtdCho. Etn and choline can also function as precursors for the synthesis of PtdEtn and PtdCho in the ER. We propose that specific genes/mutations can regulate the interorganelle PtdSer transport (pstA and pstB) and PtdEtn export (peeA and peeB). Genetic screens for lethal blocks along the pathway caused by pstA, pstB, peeA, and peeB mutations are performed by identifying strains that require Etn for growth. the radiolabel into aminophospholipids was analyzed by thin layer chromatography of the lipid extract. Both the turnover of radiolabeled PtdSer and the appearance of radiolabel in Ptd-Etn serve as a measure of PtdSer transport to and import into the mitochondria. Fig. 4 shows that the mutant strain, EAL18, has a 53% decreased accumulation of radiolabel in PtdEtn and 16% increased labeling in PtdSer relative to the parental strain, MSY30. Radiolabeling of the PtdCho pool also occurs in both the mutant and the parental cells. Since the cells are pregrown in Etn-containing medium, significant pools of Ptd-Etn and phosphoethanolamine accumulate. The resultant Ptd-Etn serves as a substrate for the methyltransferase enzymes (Pemlp and Pem2p) in the synthesis of PtdCho. The [ 3 H]serine used to label the PtdSer and PtdEtn pools can also radiolabel the one-carbon pool that is used for the synthesis of PtdCho. As a result of this one-carbon metabolism, the labeling of the PtdCho pool often appears unusually high (10). The defective PtdEtn metabolism observed in EAL18 was not a consequence of lesions in Psd1p, as revealed by enzyme assays (parental strain ϭ 23.1 nmol/45 min/mg of protein; mutant strain ϭ 20.1 nmol/45 min/mg of protein). In addition, genetic analyses reveal that the EAL18 strain complemented strains with pss1⌬ and psd1⌬ psd2⌬ null alleles. Collectively, these findings suggest that insufficient PtdSer is arriving at the locus of Psd1p in the inner mitochondrial membrane to support cell growth.
Since Psd1p synthesizes most of the cellular PtdEtn in the absence of supplemental ethanolamine, we expected that a defect in PtdSer transport to mitochondria would lead to a deficiency in cellular and mitochondrial pools of PtdEtn. Therefore, we examined the phospholipid composition of crude mitochondria prepared by centrifugation (25). Table I shows that the mitochondrial PtdEtn pool of the outcrossed mutant strain, MSY54, is about 62% of the parental strain. A relative increase in the PtdCho and phosphatidylinositol pools of the mutant accompanies the deficiency in the PtdEtn pool. These data are consistent with the whole cell radiolabeling observations and a defect in PtdSer transport to the Psd1p.
The EAL18 Strain Has a Defect in PtdSer Transport to the Mitochondria-We next employed an in vitro transport assay, in which a crude mitochondrial preparation containing MAM is [ 3 H]serine labeled in the presence of 0.5 mM MnCl 2 . These conditions provide for an initial pulse of radiolabel incorporation into PtdSer. After 15 min, EDTA is added to a final concentration of 4 mM, allowing the chase of radiolabel into PtdEtn for an additional 75 min. Fig. 5 shows an increase of 27% in PtdSer and a decrease of 63% in PtdEtn labeling in EAL18 relative to the parent. The in vitro labeling results recapitulate the whole cell [ 3 H]serine labeling data. Since Sadenosylmethionine is not added to the in vitro transport assay as a substrate for PtdCho synthesis, we observe almost no radiolabel accumulating in PtdCho. This finding shows that the  decrement in PtdEtn labeling of the mutant is not due to increased PtdEtn turnover to PtdCho. Fig. 6 demonstrates that the very same mitochondrial preparations of the mutant have Psd1p and Pss1p enzyme activities that are similar to the parental preparations. The minor increase in Psdlp activity may constitute some adaptation of the mutant cells to depletion of the cellular PtdEtn pool. Since the decrement in PtdEtn accumulation is not due to aberrant Pss1p or Psd1p enzyme activities, we conclude that the defect lies between these two biosynthetic steps. Taken together, these data demonstrate that the EAL18 mutant is deficient in PtdSer translocation from the MAM to the locus of Psd1p in the inner mitochondria membrane. Therefore, we have designated this mutant as pstA1-1. The pstA1-1 Mutant Shows Normal Intramitochondrial Transport of a PtdSer Analog but Abnormal Density-The defect in lipid traffic in the pstA1-1 mutant could occur between the MAM and the outer mitochondrial membrane or between the outer and inner mitochondrial membranes. Studies by Emoto et al. (16) have identified a mammalian cell mutant defective in PtdSer transport between mitochondrial membranes, but the complementing gene has not yet been identified. In previous studies (28), we developed a method to load the outer mitochondrial membrane with the lipid analog NBD-Ptd[1Ј-14 C]Ser and follow its import into the inner membrane. The data in Fig. 7 show the results of measurements of the transport of the lipid analog between the mitochondrial membranes using highly purified organelles. The rate of import of NBD-Ptd[1Ј-14 C]Ser was identical when compared between the parental and the mutant yeast strains. These findings demonstrate that the lesion in the pstA1-1 mutant does not occur between the mitochondrial membranes but occurs between the MAM and the outer mitochondrial membrane.
A defect in PtdSer transport to the mitochondria might be expected to alter the structure of the organelle. When preparing high purity mitochondria for reconstitution studies using sucrose density gradients, we immediately noticed that mutant strains contained mitochondria of abnormal density. In Fig. 8 the results of a density gradient purification of mitochondria are shown. The mitochondria from parental cells typically sediment as a broad band between 40 and 43% sucrose. In some preparations, the band takes on the appearance of a doublet. In contrast to the parental cells, the pstA1-1 strain shows a doublet of mitochondria between 43 and 47% sucrose. The major lower density band of the pstA1-1 mitochondria corresponds to the higher density band of the parental mitochondria. The higher density band of the pstA1-1 mitochondria is unique. The phospholipid phosphorus/protein ratio of the total purified mitochondrial pool derived from parental cells is 0.56 mol/mg of The crude mitochondria were further processed and subjected to sucrose density gradient centrifugation for the isolation of MAM and purified mitochondria. The density gradient was constructed in 1.0-ml steps of 3.3% increments from 20 -50% sucrose. The MAM and purified mitochondria were resolved by centrifugation at 100,000 ϫ g for 1 h in an SW41 rotor. The centrifuge tubes containing the parent and pstA1-1 strains are shown in the figure, and the location and percentage of the sucrose steps are shown on the right. The gradient shown is one of four prepared during this study. protein, and that of the pstA1-1 cells is 0.46 mol/mg of protein, which reflects the differences in buoyant densities. These data are entirely consistent with the pstA1 strain having a defect in lipid traffic to the mitochondria. The data also clearly demonstrate that mitochondrial populations of at least two different densities can be found in both the parental and mutant cells.
Mitochondria from pstA1-1 Cells Have a Minor Defect in Protein Import-Since our experimental results demonstrated defects in both lipid traffic to the mitochondria and abnormal organelle density, we also examined the import of proteins into purified mitochondria prepared on Nycodenz gradients (25). For these studies, we employed an Su9-DHFR chimeric construct consisting of 69 amino acids from the N terminus of subunit 9 of the mitochondrial ATPase and mouse dihydrofolate reductase (32) that is routinely used to study the import of matrix proteins. In addition, we measured the import and processing of Psd1p. The [ 35 S]methionine-labeled precursor proteins were produced using a coupled transcription/translation reaction (30), and the time-and membrane potential (⌬⌿)dependent import of the proteins was measured as shown in Fig. 9. The precursor for Su9-DHFR is readily imported into the mitochondria and processed to its mature form in both parental and pstA1-1 cells. The rate of Su9-DHFR import into mutant mitochondria is ϳ60% of that found for parental mitochondria. The precursor for Psd1p is also imported into the mitochondria and targeted to the inner membrane of both parental and mutant strains. Consistent with stepwise processing of the mitochondrial targeting sequence and the inner membrane sorting sequence, two precursor forms of Psd1p (designated pI and pII) are identifiable. The data in Fig. 9 show the appearance of the mature ␤ subunit of Psd1p and reveal that the import into mutant mitochondria occurs at 70 -80% of the rate for parental mitochondria. These studies provide evidence that the pstA1-1 mutant has a very mild defect in the rate of protein import into the mitochondria. When comparing the differences in protein and phospholipid import rates into the mitochondria, it is also important to understand that changes in protein import principally affect protein accumulation in the mitochondria, whereas changes in PtdSer import affect the total cellular biosynthesis of PtdEtn and PtdCho.
A Genomic Library Screen Identifies MET30 as the Gene That Complements the pstA1-1 Strain-The pstA1-1 strain, EAL18, was outcrossed twice to yield the pstA1-1 strain, MSY57, which was utilized to screen a multicopy vector (pSEY18) yeast genomic library for clones that could complement the ethanolamine auxotrophy of MSY57. Approximately 22,000 uracil prototrophic pSEY18 library transformants were screened for ethanolamine prototrophy. Plasmids from 45 uracil/ethanolamine prototrophs were rescued into Escherichia coli. Purified plasmids were retransformed into pstA1-1 strains. Of the 45 retransformed plasmids, only one conferred robust growth on selective medium, whereas other library clones yielded intermediate, poor, or no growth on selective medium. The library clone that gave the best complementation of pstA1-1 ethanolamine auxotrophy contained five open reading frames: PCL7, DGF10, NEO1, SYG1, and MET30. Subsequent subcloning of the open reading frames into single copy yeast vectors, YCP50 and/or YCplac33, demonstrated that MET30 conferred the ethanolamine prototrophy to pstA1-1 mutant strains. Fig. 10 shows that a single copy plasmid, YC-plac33, containing the MET30 gene, can rescue the growth defect of a pstA1-1 mutant in SL medium devoid of ethanolamine. The MET30 gene encodes a substrate specificity subunit of SCF ubiquitin ligase (33). Major motifs found within Met30p include an F box for interaction with the Skp1p protein subunit and WD40 domains for interaction with substrates (34,35).
To demonstrate that MET30 can complement the PtdSer transport defect observed in our biochemical assays, we employed the in vitro transport assay with crude mitochondria using a pstA1-1 strain carrying MET30 on a low copy plasmid. Fig. 11 shows that MET30 on plasmid YCplac33 can restore the accumulation of radiolabel into PtdEtn compared with the parental and pstA1-1 strain harboring empty vector. Taken together, the in vivo and in vitro data show that wild type MET30 can complement both the ethanolamine auxotrophy and biochemical transport-deficient phenotype of pstA1-1. (1 g/ml) was added to eliminate membrane potential (⌬⌿). For PSD1-V5, pI is the nascent translation product, pII is the processing intermediate lacking the mitochondrial targeting sequence, and ␤ is the mature ␤ subunit of the enzyme. For Su9-DHFR, p denotes the precursor form, and m denotes the mature form. The percentage of import was calculated by PhosphorImager analysis. The levels of mature ␤ subunit for PSD1-V5 and mature Su9-DHFR from wild type cells after a 10-min reaction were set at 100% import. These data are from a representative one of three experiments.
FIG . 10. The MET30 gene complements the ethanolamine auxotrophy of pstA1-1. Liquid cultures were grown to early log phase in synthetic medium containing 3 mM ethanolamine. Cells were harvested by centrifugation, washed twice with sterile water, and suspended to give an A 600 of 0.02 in synthetic lactate medium without ethanolamine. The cultures were incubated with shaking at 30°C, and the A 600 was monitored over time. Ⅺ, parent strain, MSY30, plus empty vector, YCplac33; E, mutant strain, EAL18 plus empty vector, YCplac33; q, mutant strain, EAL18 plus MET30 on low copy vector YCplac33.
We also tested whether the defect in cell growth and lipid traffic we observe with our strains is reproduced with an independently isolated met30 allele. We obtained a met30-2 allele (36) and through gene eviction combined it with the psd2⌬::KAN r allele. The new strain, MSY77, grew on glucose at 30°C but failed to grow on glucose at 36°C. Supplementation of the strain with Etn repaired the growth defect at 36°C. We also isolated crude mitochondria from MSY77 grown on semisynthetic lactate medium supplemented with Etn that had been shifted to the nonpermissive temperature overnight and performed reconstitution studies. When compared with its parental strain, MSY77 harboring the mutant allele exhibited an accumulation of radiolabel in PtdSer and a defect in the transport-dependent formation of PtdEtn as shown in Fig. 12. These findings reinforce our conclusions about the function of MET30 in regulating PtdSer traffic to the mitochondria. We also examined the mitochondrial density in the MSY77 strain, and the results are shown in Fig. 13. Similar to the findings with pstA1-1 strains, the MSY77 strain with the met30-2 allele accumulates a dense mitochondrial band. The strain also lacks the low density band of mitochondria found in the wild type cells. Thus, the met30-2 allele in conjunction with psd2⌬ produces Etn auxotrophy, defective PtdSer transport, and abnormal mitochondrial density similar to that observed with our met30 allele.
To identify the structural mutation of our met30 allele, we recovered the allele from a pstA1-1 strain using the gapped plasmid technique (37). We mapped the mutation within the met30 open reading frame by creating mutant/wild-type chimeras and assaying for complementation of pstA1-1 ethanolamine auxotrophy. After delineating the mutation to a 510-base pair region, the allele was sequenced on both strands. A single nucleotide change, which results in a substitution of leucine for proline at amino acid position 522, was identified. This mutation precedes the fifth WD40 motif of Met30p that participates in substrate recognition. It is likely that this mutation interferes with the three-dimensional orientation of the WD40 domains and alters substrate recognition.
The pstA1-1 Strain Is Defective in Protein Ubiquitination in Vivo-The WD40 domains are predicted to direct the interaction of Met30p with multiple substrates. The best characterized substrate for Met30p-directed ubiquitin ligase is Met4p, a transcriptional regulator of genes in methionine biosynthesis (31,38). We examined the ubiquitination of Met4p in response to methionine supplementation in our pstA1-1 strain and a met30-6ts strain and corresponding wild type strains. A brief summary of methionine regulation of Met4p is outlined in Fig.  14A. In the presence of methionine, Met30p interacts with Met4p and the SCF complex to promote ubiquitination of Met4p (Met4p-Ubi). The Met4p-Ubi cannot be recruited to multiple MET gene promoters and essentially becomes inactive. Under some growth conditions, the Met4p-Ubi appears stable, but other growth conditions promote degradation (31,38). In our experiments, ubiquitination of Met4p was detected using a myc epitope-tagged version of Met4p. Methionine addition produced two prominent forms of Met4p-Ubi in wild type cells, as shown in Fig. 14B. In glucose medium, the parental strain for pstA1-1 shows the presence of Met4p-Ubi (bands A and B) and almost no detectable Met4p. The pstA1-1 mutant also shows Met4p-Ubi but in addition reveals the presence of nonubiquitinated Met4p (bands D and E). The wild type control cells for the met30-6ts strain also produce Met4p-Ubi and essentially no FIG. 13. The met30-2 mutant has mitochondria of abnormal density. Crude mitochondria were prepared from wild type and mutant (MSY77) strains grown in semisynthetic lactate medium plus Etn at 36°C overnight. The mitochondria were subjected to sucrose density gradient centrifugation at 100,000 ϫ g for 1 h in an SW41 rotor. The centrifuge tubes containing the parental and met30-2 strains are shown, and the location and percentage of the sucrose steps are shown on the right.
Met4p. In contrast, the met30-6ts strain contains significant Met4p-Ubi at 25°C but also accumulates significant amounts of nonubiquitinated Met4p. Moreover, the met30-6ts strain displays markedly reduced Met4p-Ubi at 37°C and accumulates very high levels of nonubiquitinated Met4p. In addition to studies in glucose medium, we analyzed Met30p function in pstA1-1 cells in lactate medium, since these are the conditions that produce the lethal phenotype. In lactate medium containing methionine, we demonstrated that the parental strain for pstA1-1 contains stable levels of Met4p-Ubi and very low levels of Met4p. In contrast, the pstA1-1 strain not only contains Met4p-Ubi but also accumulates relatively high levels of Met4p. Quantification of the immunoreactive Met4p reveals that the ratio of Met4p/Met4p-Ubi is 30 -40-fold higher in the pstA1-1 strain than its corresponding parental strain. From these data, we conclude that the pstA1-1 strain exhibits a defect in Met4p ubiquitination. The in vivo defect in ubiquitination is more pronounced with growth of the pstA1-1 strain in lactate medium than glucose medium. In parallel with the above studies, we performed genetic experiments in which we constructed a psd2⌬ met4⌬ strain. If Met4p functions downstream of Met30p as a positive regulator of lipid traffic, then psd2⌬ met4⌬ mutations should reproduce the phenotype of the pstA1-1 strain. However, the psd2⌬ met4⌬ strain we constructed had no detectable phenotype. These data make it unlikely that Met4p functions as a positive regulator of lipid transport. These data also indicate that Met4p-Ubi cannot function as a positive regulator of lipid transport. These results raise the possibility that novel substrates for Met30p could be ubiquitinated and function to regulate lipid traffic.
The pstA1-1 Transport Defect Is Associated with Mitochondria and MAM-A previous study has demonstrated that Ptd-Ser can be transported from yeast MAM to mitochondria in a reconstituted system containing both organelles (6). We have applied this reconstituted assay to determine whether the pstA1-1 transport defect is associated with the donor MAM fraction or the acceptor mitochondria. MAM and mitochondria from a pstA1-1 strain and the wild type parental strain were separated on sucrose density gradients. Individual reconstitution reactions were prepared consisting of wild type MAM in combination with either wild type or mutant mitochondria and mutant MAM in combination with either wild type or mutant mitochondria. As described above, the synthesis and transport of PtdSer was followed using a [ 3 H]serine precursor. The turnover of PtdSer and the appearance of radiolabel in PtdEtn serves as a measure of PtdSer transport from the MAM to the mitochondria. To simplify the comparison of conditions, the data are expressed as the ratio of PtdEtn/PtdSer as shown in Fig. 15. These data reveal that the transport defect observed in the intact cell and the crude mitochondrial preparations is also reproduced in the purified MAM and mitochondria reconstitution assay. Wild type cells convert ϳ30% of their nascent Ptd-Ser to PtdEtn, whereas the mutant cells convert only about 10%. The wild type MAM cannot overcome the defect in a reaction containing mutant mitochondria. In addition, wild type mitochondria cannot overcome the defect in a reaction containing mutant MAM. These results demonstrate that the transport defect of pstA1-1 cells is not exclusively associated with mitochondria or MAM but appears to reside in both organelles. From these data, we conclude that Met30p has a regulatory effect on transport constituents present in both mitochondria and MAM.

DISCUSSION
In this paper, we describe the isolation of a new Etn auxotroph with the characteristics expected for a mutant with a defect in PtdSer transport from the MAM to the mitochondria. This study is part of a broad genetic approach outlined in Fig.  1 that we are applying to uncover molecules involved in intermembrane lipid traffic. The new mutant, designated pstA1-1, is the first that we have discovered in the PSTA pathway, that we propose regulates aminoglycerophospholipid movement to the mitochondria.
The pstA1-1 mutant was selected for Etn auxotrophy on a lactate carbon source. This selection was biased with respect to a respiratory carbon source so as to identify strains that had an absolute requirement for fully functional mitochondria. The Etn requirement of the pstA1-1 strain cannot be replaced by choline supplementation. Despite the ability of choline to enable synthesis of PtdCho via the Kennedy pathway, this lipid cannot replace PtdEtn. This latter finding reflects both the need for critical levels of PtdEtn in the mitochondria, under respiratory conditions, and the essential role of PtdEtn in cells that has only recently been recognized (13,14). Although the pstA1-1 strain is an Etn auxotroph, the level and rate of PtdEtn synthesis is only modestly depressed in these cells under nutrient-restrictive conditions. Greater reduction of PtdEtn levels and synthesis may simply not be tolerated under the conditions we have employed in this genetic screen. We suspect that the Etn supplementation of cells allows sufficient PtdEtn to be made by the Kennedy pathway to spare enough PtdEtn made by Psd1p to be retained and used by the mitochondria. In this regard, we highlight the observation that an independently isolated mutant, met30-2 (36), that is allelic to the pstA1-1 mutation shows a similar phenotype when it is combined with a psd2⌬ mutation such that it is an Etn auxotroph on glucose at 36°C.
Examination of the PtdSer traffic between organelles iso-lated from the pstA1-1 strain gives results that reflect the defects in lipid synthesis and composition of intact cells. In cell-free assays that monitor PtdSer synthesis in the MAM and its transport to the mitochondria, the pstA1-1 strain shows a 27% accumulation of PtdSer and greater than 63% decrement in the decarboxylation of this lipid. Although the defect in lipid traffic is not complete, it is sufficient to produce a growth phenotype in the intact cell and a highly reproducible biochemical phenotype in vitro. This lipid transport defect is also reproduced in cells harboring the met30-2 allele. The partitioning properties of (N-4-nitrobenzo-2-oxa-1,3 diazole)aminoacyl lipids containing short acyl chains in the sn-2-position allows them to be used to rapidly load the outer mitochondrial membrane and examine lipid traffic to the inner mitochondrial membrane (28). Our findings with NBD-Ptd[1Ј-14 C]Ser demonstrate that the defect in the mutant is not between the mitochondrial membranes but occurs between the MAM and the outer mitochondrial membrane. Consistent with the assignment of the transport lesion to events between the MAM and mitochondria are the results obtained from density gradient centrifugation that reveal that a significant fraction of the mitochondria from pstA1-1 cells has abnormally high density. The defect in PtdSer transport to the mitochondria caused the density of the organelle to increase. This increase in density indicates that the mitochondria fail to fully compensate for the PtdSer transport defect by increasing the import or synthesis of other phospholipids. The net effect is an overall change in the phospholipid/protein ratio of the mitochondria isolated from the mutant cells. This same defect is also observed in the met30-2 strain. These data further indicate that both the parental cells and mutant cells each appear to have mitochondria of two distinctly different densities. It is also noteworthy that the pstA1-1 strain not only has a unique high density mitochondrial band but also lacks the low density band associated with the parental phenotype.
Although the mechanism of lipid transfer between the MAM and mitochondria is not known, several studies have suggested that proteins on the surface of one or both membranes participate in the process (6,21). Some evidence has also been provided to suggest that proteins with fusogenic properties may participate in the lipid transfer reaction, but none of these components has been clearly identified (39,40). Both biochemical and morphological experiments have implicated zones of close physical association between the MAM and mitochondria as likely elements involved in the intermembrane transfer of PtdSer (4, 6, 20 -24).
The gene that complements the pstA1-1 defects in growth and lipid traffic is MET30. The MET30 gene encodes a protein with an F box and five WD40 domains that dictate substrate specificity for a subset of SCF family ubiquitin ligases (33,34). The F box is the site of interaction of Met30p with the Skp1p component of the ligase complex (35). The WD40 domains are involved in protein-protein interaction. The pstA1-1 mutation (P522L) resides in the carboxyl terminus of the molecule in a region between the fourth and fifth WD40 domain. We anticipate that this mutation will alter the recognition of one or more specific substrates and their ubiquitination. All of the current data are consistent with regulation of PtdSer transport to the mitochondria by the process of ubiquitination. However, the level at which ubiquitin ligase regulates lipid traffic is still unknown. Ubiquitin modification is now recognized to regulate cellular processes at multiple levels that include proteolysis (41), transcriptional activation (42), transcriptional inactivation (31,38), and signaling (43,44). Several targets for SCF-Met30p have been identified, including Met4p (31, 36), the VP16 transcriptional activation domain (42), and Swe1p (45). Ubiquitination leads to inactivation of the transcription factor Met4p (31,36) and activation of a heterologous VP16 transcriptional activation domain (42). The Swe1p is an inhibitory kinase that is degraded after ubiquitination (45).
Since Met4p is the best characterized substrate for Met30pdirected ubiquitin ligase, we also directly tested whether our pstA1-1 mutant was defective in Met4p modification. In experiments shown in Fig. 14, we demonstrated that Met4p ubiquitination in the pstA1-1 mutant was significantly reduced in vivo. From these data, we infer that Met30p will also be defective in recognizing and ubiquitinating other substrates as well. We also designed a specific genetic test to determine whether Met4p could act as a positive regulator of lipid traffic downstream of Met30p. In this experiment, we generated a psd2⌬ met4⌬ strain and tested for Etn auxotrophy. Since the strain was not an Etn auxotroph, we conclude that Met4p does not act as a positive modulator of PtdSer transport. In addition, these results also eliminate a requirement for Met4p-Ubi as a positive regulator of lipid transport.
Ubiquitination is now also recognized to play an important role in membrane traffic as a signaling motif for protein sorting, endocytosis, and viral budding (46). Certainly, a provocative model for Met30p function in lipid traffic entails modification of membrane targets for the assembly of macromolecular docking complexes that facilitate apposition of donor and acceptor membranes (e.g. MAM and mitochondria) and intermembrane lipid transport. In addition to our study, another line of investigation has linked protein ubiquitination to mitochondrial maintenance and function. Using a genetic suppressor screen, Fisk and Yaffe (47) identified the ubiquitin ligase Rsp5p as an essential regulator of mitochondrial inheritance and morphology. Thus, more than one ubiquitin ligase system appears to play a role in maintaining the structure and function of mitochondria. The Rsp5p has also been implicated in endocytosis (48), and protein sorting at the trans-Golgi network (49), and its mammalian homolog, Nedd4, participates in viral budding (50).
Our findings also indicate that the action of Met30p in regulating PtdSer transport to mitochondria is not restricted to this organelle. The reconstituted PtdSer transport system used in these studies enabled us to critically test the ultimate site of effect of Met30p. Using permutations of admixtures of wild type and pstA1-1-derived MAM and mitochondria, we mapped the lesion to both organelles. We interpret this to mean that Met30p causes distinct biochemical changes in both MAM and mitochondria, simultaneously. Hypothetically, one simple means to achieve this effect would be direct ubiquitination of critical proteins on both organelles. Clearly, more experimentation will now be necessary to determine if the effect of Met30p is direct or indirect.
In summary, we have isolated a mutant defective in PtdSer traffic between the MAM and the mitochondria. The mutation alters the PtdEtn content, the total phospholipid content, and the density of the organelle. The effects of the mutation are manifested in both the MAM and mitochondria and are reversed by the MET30 gene, which encodes a component of SCF ubiquitin ligase. These findings now identify a novel mechanism for regulating interorganelle phospholipid transport.