Nonvesicular phospholipid transfer between

Nonvesicular phospholipid transfer between peroxisomes and the endoplasmic reticulum Sumana Raychaudhuri and William A. Prinz* Laboratory of Cell Biochemistry and Biology, National Institute of Diabetes and Digestive and Kidney Diseases, National Institutes of Health, Bethesda, MD 20892

The endoplasmic reticulum (ER) plays an important role in peroxisome biogenesis; some peroxisomal membrane proteins are inserted into the ER and trafficked to peroxisomes in vesicles. These vesicles could also provide the phospholipids required for the growth of peroxisomal membranes, because peroxisomes lack phospholipid biosynthesis enzymes. To test this, we established a novel assay to monitor phospholipid transfer between the ER and peroxisomes and found that phospholipids are rapidly trafficked between these compartments. This transport is not blocked in mutants with conditional defects in Sec proteins required for vesicular trafficking from the ER or in Pex3p, a protein required for peroxisome membrane biogenesis. ER to peroxisome lipid transport was reconstituted in vitro and does not require cytosolic factors or ATP. Our findings indicate that lipids are directly transferred from the ER to peroxisomes by a nonvesicular pathway and suggest that ER to peroxisome vesicular transport is not required to provide lipids for peroxisomal growth. ER 兩 lipid transport 兩 membrane 兩 organelle biogenesis

P

eroxisomes are found in nearly all eukaryotic cells, where they play a central role in lipid metabolism and are required for the ␤-oxidation of fatty acids and the elimination of the hydrogen peroxide produced by this process (1, 2). Their importance is underscored by the fact that a number of inherited diseases are caused by defects in peroxisome formation (3). The mechanism of peroxisome biogenesis has been controversial. For some time, the prevailing view was that peroxisomes were autonomous organelles that, like mitochondria and chloroplasts, arose exclusively by the growth and division of preexisting peroxisomes (1). This view was consistent with the fact that all peroxisome matrix proteins and most peroxisomal membrane proteins are synthesized in the cytosol on free ribosomes and posttranslationally translocated into peroxisomes. However, in recent years, a number of studies have indicated that the ER plays a critical role in peroxisome membrane biogenesis. These studies showed that some peroxisomal membrane proteins are first inserted into the ER membrane and subsequently trafficked to peroxisomes (4–8). The ER is also required for de novo peroxisome biogenesis, which can occur in both mammalian and yeast cells. Mutants missing one of the genes required for peroxisome biogenesis, termed PEX genes, lack peroxisomes but will form them de novo when expression of the missing PEX gene is restored. In mammalian cells, de novo peroxisome biogenesis also occurs in wild-type cells (7). In contrast, this process could not be detected in wild-type Saccharomyces cerevisiae (9). The mechanisms of ER to peroxisome vesicular transport and de novo peroxisome formation from ER membranes remain poorly understood. Although ⬎30 Pex proteins have been identified, only 2, Pex3p and Pex19p, are specifically required for peroxisome membrane biogenesis in the yeast S. cerevisiae (10). In mammalian cells, the protein Pex16 is also required (11, 12), but S. cerevisiae lacks a Pex16 homolog. The functions of these proteins are not known, but they could sort peroxisomal membrane proteins into domains of the ER that will generate peroxisomal precursor www.pnas.org兾cgi兾doi兾10.1073兾pnas.0808321105

vesicles. The formation of these vesicles may not require the Sec proteins needed for vesicular transport between the ER and the Golgi complex (13, 14). Although the transport of proteins to peroxisomes has been extensively studied, much less is known about how the lipids required for peroxisome membrane biogenesis are trafficked to peroxisomes. The growth and division of peroxisomes requires lipids because peroxisomes lack enzymes that synthesize membrane lipids (15, 16). Therefore, it has been suggested that one important function of ER to peroxisome vesicular transport, in addition to bringing some membrane proteins to peroxisomes, may be to provide peroxisomal membranes with lipids (4, 17). Here, we establish an assay to monitor lipid trafficking between the ER and peroxisomes. This assay is similar to one previously used to assess lipid transfer between the ER and either mitochondria or the Golgi complex (18). Two of the most abundant glycerophospholipids in the cell are phosphatidylethanolamine (PtdEtn) and phosphatidylcholine (PtdCho). These can be synthesized by two separate pathways (Fig. 1A). In one, phosphatidylserine (PtdSer) is decarboxylated to form PtdEtn, which can in turn be methylated to form PtdCho. PtdEtn and PtdCho can also be made by the Kennedy pathway, in which either ethanolamine (Etn) or choline (Cho) is used to make PtdEtn or PtdCho. All of the enzymes in both pathways are in the ER except phosphatidylserine decarboxylase (Psd), which converts PtdSer to PtdEtn. S. cerevisiae has two Psds, one in mitochondria (Psd1p) and the other in the Golgi complex (Psd2p). Thus, the conversion of PtdSer into PtdEtn and PtdCho requires lipid transport out of the ER and back again. The localization of Psd1p and Psd2p has been exploited to investigate glycerophospholipid transport (18). These studies demonstrated that phospholipids are rapidly moved between these compartments by nonvesicular, energy-independent pathways. To study glycerophospholipid transfer between the ER and peroxisomes, we localized Psd to the peroxisomal matrix in a strain missing the endogenous Psds. In these cells, the conversion of PtdSer into PtdEtn and PtdCho indicates that these lipids have been moved between the ER and peroxisomes (Fig. 1B). We find that a nonvesicular transport pathway rapidly moves lipids from the ER to peroxisomes. Our results suggest that ER to peroxisome vesicular transport is not required to supply peroxisomes with lipids. Nonvesicular lipid transfer between the ER and peroxisomes likely provides a mechanism for cells to rapidly regulate the amount and composition of lipids in peroxisomal membranes. Results An Assay to Monitor Lipid Transfer Between Peroxisomes and the ER.

To investigate glycerophospholipid transfer between the ER and peroxisomes, we localized Psd to the peroxisomal matrix in a strain Author contributions: S.R. and W.A.P. designed research; S.R. and W.A.P. performed research; S.R. and W.A.P. analyzed data; and W.A.P. wrote the paper. The authors declare no conflict of interest. *To whom correspondence should be addressed at: Laboratory of Cell Biochemistry and Biology, National Institute of Diabetes and Digestive and Kidney Diseases, National Institutes of Health, 8 Center Drive, Building 8, Room 301, Bethesda, MD 20892. E-mail: [email protected]. This article contains supporting information online at www.pnas.org/cgi/content/full/ 0808321105/DCSupplemental.

PNAS 兩 October 14, 2008 兩 vol. 105 兩 no. 41 兩 15785–15790

CELL BIOLOGY

Communicated by Tom A. Rapoport, Harvard Medical School, Boston, MA, August 22, 2008 (received for review January 21, 2008)

Fig. 1. Scheme for monitoring phospholipid transfer between the ER and peroxisomes. (A) Pathways for producing PtdSer, PtdEtn, and PtdCho in S. cerevisiae. Enzymes are shown in blue. DHS, dihydrosphingosine; DAG, diacylglycerol; LCA, long chain aldehyde. (B) The Pex15-Psd-YFP fusion in peroxisomes can be used to monitor phospholipid transfer between the ER and peroxisomes.

missing the endogenous Psds (Fig. 1B). Escherichia coli Psd was fused to the C terminus of Pex15p, a peroxisomal protein that has a single transmembrane domain. The C terminus of Pex15p is in the peroxisomal matrix, and it has previously been demonstrated that proteins fused to the C terminus of Pex15p are also localized to the matrix (19). We also fused yellow fluorescent protein (YFP) to the C terminus of Psd to allow localization of the fusion. The plasmid encoding the Pex15-Psd-YFP fusion was integrated into the chromosome so that the endogenous PEX15 promoter expressed the fusion. Pex15p is a very low abundance protein, and the fusion is expressed at low levels (19). Indeed, we were unable to detect the YFP portion of the fusion using immunoblotting (data not shown). However, we could detect the fusion using fluorescent microscopy, which showed a punctate distribution (Fig. 2A). We demonstrated that it was in peroxisomes by colocalizing it with cyan fluorescent protein (CFP) with the C-terminal peroxisomal targeting sequence SKL (Fig. 2 A). To confirm that the Pex15-Psd-YFP fusion is in peroxisomes, we used sucrose density gradients to separate membranes from cells expressing the fusion. Because we were unable to detect the fusion with immunoblotting, we expressed the fusion in cells lacking the endogenous Psds and determined the abundance of the fusion protein in the gradient fractions by

measuring Psd activity. Fractions with the highest Psd activity were also enriched with the peroxisomal membrane protein Pex12p but not the ER protein Dpm1p, suggesting that most of the fusion is in peroxisomes (Fig. 2B). Some of the Psd activity was in fractions enriched in the ER protein Dpm1p but not Pex12p, suggesting that a small percentage of the fusion may be in the ER. In addition, some Psd activity was found near the top of the gradients. This activity may be associated with peroxisomes that have been lysed. To use the Pex15-Psd-YFP fusion to measure lipid transfer between the ER and peroxisomes, cells missing the two endogenous Psds (Psd1p and Psd2p) were labeled with [3-3H]serine, which is used by PtdSer synthase in the ER to form [3H]PtdSer (20). We found that, in addition to missing Psd1p and Psd2p, the labeling had to be performed in cells that were also lacking dihydrosphingosine1-phosphate lyase (Dpl1p); otherwise, radiolabel from [3H]serine can be incorporated into PtdEtn by mechanisms other than decarboxylation of PtdSer (Fig. 1 A). When psd1⌬ psd2⌬ dpl1⌬ cells expressing the Pex15-Psd-YFP fusion were labeled with [3H]serine, [3H]PtdSer was formed and converted to PtdEtn at a linear rate (Fig. 3 B and C). In a control experiment, psd1⌬ psd2⌬ dpl1⌬ cells lacking the Pex15-Psd-YFP fusion made [3H]PtdSer at a linear rate but formed virtually no radiolabeled PtdEtn because they were

Fig. 2. The Pex15-Psd-YFP fusion is mostly in peroxisomes. (A) Colocalization of Pex15-Psd-YFP and CFP-SKL in psd1⌬ psd2⌬ dpl1⌬ cells. Three cells are shown. (Scale bar: 1 ␮m.) (B) Membranes from psd1⌬ psd2⌬ dpl1⌬ cells expressing Pex15-Psd-YFP were separated on sucrose density gradients. The Psd activity in each fraction was determined (Upper). Psd activity is indicated by the amount of NBD-PE formed from NBD-PS in 1 h. Each fraction was also immunoblotted to detect the ER membrane protein Dpm1p and the peroxisomal membrane protein Pex12-TAP with Peroxidase-Anti-Peroxidase (Lower). (C) As in B, except that pex3⌬ psd1⌬ psd2⌬ dpl1⌬ cells expressing Pex15-Psd-YFP were used. 15786 兩 www.pnas.org兾cgi兾doi兾10.1073兾pnas.0808321105

Raychaudhuri and Prinz

unable to decarboxylate PtdSer (Fig. 3B). The small amount of PtdCho that becomes radiolabeled in this strain but is probably not derived from PtdSer: it is most likely due to degradation of [3H]serine by the ‘‘one-carbon’’ pathway (21). The conversion of newly synthesized PtdSer to PtdEtn in psd1⌬ psd2⌬ dpl1⌬ cells expressing the Pex15-Psd-YFP fusion suggests that the PtdSer formed in the ER of this strain is transferred to peroxisomes, where it is decarboxylated to form PtdEtn. However, because some of the fusion protein may be in the ER (Fig. 2B), we wanted to rule out that most or all of the PtdSer to PtdEtn conversion is catalyzed by the small fraction of Pex15-Psd-YFP that is in the ER rather than by the fusion in peroxisomes. If this were the case, then increasing the amount of the fusion in the ER would significantly enhance the rate of PtdSer to PtdEtn conversion. We increased the amount of the Pex15-Psd-YFP fusion in the ER by expressing it in pex3⌬ cells, which lack peroxisomes and localize some normally peroxisomal membrane proteins in the ER (10, 22). Using sucrose density gradients, we confirmed that the Pex15-PsdYFP remains mostly in the ER in pex3⌬ cells (Fig. 2C). When pex3⌬ psd1⌬ psd2⌬ dpl1⌬ cells expressing the Pex15-Psd-YFP fusion were labeled with [3H]serine, they converted [3H]PtdSer to PtdEtn at a slightly slower rate than isogenic PEX3⫹ cells (Fig. 3 C–E) despite having the same amount of total Psd activity (Fig. 3F). This indicates that the specific activity of the fusion in psd1⌬ psd2⌬ dpl1⌬ cells (where most of the fusion is in peroxisomes) is slightly higher than it is in pex3⌬ psd1⌬ psd2⌬ dpl1⌬ cells (where it is mostly in the ER). Therefore, the conversion of PtdSer to PtdEtn in psd1⌬ psd2⌬ dpl1⌬ cells expressing Pex15-Psd-YFP is mostly occurring in peroxisomes and indicates that PtdSer has been transferred from the ER to peroxisomes. PtdSer Transfer Between the ER and Peroxisomes. To estimate the relative rate of PtdSer transfer from the ER to peroxisomes, we calculated the percent of [3H]PtdSer formed that is converted to PtdEtn per minute. In psd1⌬ psd2⌬ dpl1⌬ cells expressing Pex15Psd-YFP, 0.40 ⫾ 0.05% of the [3H]PtdSer is converted per minute, whereas in wild-type cells (which express Psd1p and Psd2p in mitochondria and the Golgi complex), 1.9 ⫾ 0.20% is converted per minute (Fig. 3E). This 5-fold difference could indicate that PtdSer transport from the ER to peroxisomes is slower than transfer to Raychaudhuri and Prinz

mitochondria and the Golgi complex, or it could be explained by the amount of Psd in the strains. We found that wild-type cells had ⬇6 times more total Psd activity than the psd1⌬ psd2⌬ dpl1⌬ cells expressing the Pex15-Psd-YFP fusion (Fig. 3F). Thus, newly synthesized PtdSer is likely transferred from the ER to peroxisomes about as efficiently as it is moved from the ER to mitochondria or the Golgi complex in wild-type cells. As a further indication of the efficiency of PtdSer transfer between ER and peroxisomes, we determined whether expression of Psd in peroxisomes affected the total amount of PtdSer in cells. We measured the relative abundance of the four major glycerophospholipids in cells: PtdCho, PtdEtn, PtdSer, and phosphatidylinositol (PtdIns). The psd1⌬ psd2⌬ dpl1⌬ strain accumulates ⬇3 times more PtdSer than wild-type cells because they cannot convert PtdSer to PtdEtn (Fig. 3G). Approximately 6.5% of the total phospholipid was PtdSer in wild-type cells, whereas 18% of the total is PtdSer in psd1⌬ psd2⌬ dpl1⌬ cells. The amount of PtdSer is significantly reduced to 13% in psd1⌬ psd2⌬ dpl1⌬ cells expressing Pex15-Psd-YFP (Fig. 3G), indicating that Psd in peroxisomes must have access to a substantial portion of the PtdSer in cells. These findings suggest that most of the PtdSer in the cell continuously cycles in and out of peroxisomal membranes. Peroxisome to ER PtdEtn Transfer. If PtdEtn formed in peroxisomes

is transferred back to the ER, it can be used to synthesize PtdCho (Fig. 1B). Thus, when psd1⌬ psd2⌬ dpl1⌬ cells expressing Pex15Psd-YFP are labeled with [3H]serine, radiolabeled PtdCho can be formed from PtdEtn that has been transferred from peroxisomes to the ER. However, we found it difficult to monitor PtdEtn transfer in cells because very little of the PtdEtn derived from PtdSer was converted to PtdCho, even in wild-type cells (Fig. 3A). It is not clear why this is the case, but it should be noted that the labeling was performed in medium that contained Etn, and it may be that PtdEtn formed by the Kennedy pathway, rather than by PtdSer decarboxylation, is preferentially converted to PtdCho. As an alternative approach, we determined whether psd1⌬ psd2⌬ dpl1⌬ cells expressing Pex15-Psd-YFP were able to grow in conditions that required PtdEtn transfer from peroxisomes to the ER. Yeast strains that cannot synthesize PtdCho are not viable in standard growth medium (23). Thus, psd1⌬ psd2⌬ dpl1⌬ cells PNAS 兩 October 14, 2008 兩 vol. 105 兩 no. 41 兩 15787

CELL BIOLOGY

Fig. 3. PtdSer and PtdEtn are rapidly transferred between the ER and peroxisomes. (A–D) The indicated strains were grown in medium with Etn and labeled with [3H]serine. Shown is the amount of radiolabel (cpm) in fractions containing the indicated lipids per OD600. (E) The results in A–D were used to calculate the percent of radiolabeled PtdSer converted to PtdEtn per minute. Strains refer to those used in A–D. (F) The total Psd activity (rate of NBD-PE formation from NBD-PS) in A–D. (G) The total amount of the four major phospholipids in the strains in A–C. (H) Serial dilutions of the strains in A–C were plated on medium that either does or does not contain Etn. The plates were grown for 3 days at 30°C. In A–F, error bars indicate the SEM (n ⫽ 2– 4).

We wondered whether the transport of PtdSer from the ER to peroxisomes requires Pex3p, one of the two proteins needed for peroxisome membrane biogenesis in S. cerevisiae. Cells with a conditional defect in Pex3p (pex3-B5) contain peroxisomes at permissive temperature but have defects in targeting some membrane proteins to peroxisomes after being shifted to 37°C (31). We determined whether the rate of PtdSer to PtdEtn conversion slowed in a pex3-B5 psd1⌬ psd2⌬ dpl1⌬ strain expressing Pex15-Psd-YFP. However, the rate was not affected after a 10- or 60-min shift to 37°C (Fig. 4A). We confirmed that the Pex15-Psd-YFP fusion still colocalized with CFP-SKL after a 1.5-h shift to 37°C (data not shown) and that this treatment did not affect Psd activity (Fig. 4B). Taken together, these results indicate that ER to peroxisome PtdSer transfer does not require Pex3p. Lipid Transfer Between ER and Peroxisomes in Vitro. Our findings

Fig. 4. ER to peroxisome PtdSer transfer does not require Sec proteins or Pex3p. (A) Strains were grown in medium with Etn at 25°C, shifted to 37°C for either 10 or 60 min as indicated, and labeled with [3H]serine. Shown is the percentage of radiolabeled PtdSer converted to PtdEtn per minute. The data used to calculate these rates are shown in supporting information (SI) Fig. S1. (B) Strains were grown in medium with Etn at 25°C and shifted to 37°C for 45 min, and Psd activity (rate of NBD-PE formation from NBD-PS) was determined. Error bars indicate the SEM (n ⫽ 2– 4).

expressing Pex15-Psd-YFP can only grow in medium that lacks Etn and Cho if the PtdEtn formed in peroxisomes is transferred to the ER and converted to PtdCho. We found that this strain was able to grow slowly in medium without Etn and Cho, whereas psd1⌬ psd2⌬ dpl1⌬ cells could not (Fig. 3H). Therefore, some of the PtdEtn formed in peroxisomes may be transferred back to the ER. This transport may not be very efficient because the cells grow poorly in the absence of Etn. Alternatively, the PtdEtn generated in peroxisomes could be degraded to Etn or other metabolites that can be used by the Kennedy pathway to make PtdCho. PtdSer Transfer to Peroxisomes Does Not Require Sec18p, COPII, or Pex3p. To investigate how PtdSer is transferred from the ER to

peroxisomes, we determined the rate of PtdSer to PtdEtn conversion in two strains with conditional defects in Sec proteins needed for vesicular transport from the ER, Sec18p and Sec16p. Sec18p, the yeast homolog of N-ethylmaleimide-sensitive factor, is required for most vesicular transport including vesicular transport from the ER (24, 25), and Sec16p is required for the formation of COPII vesicles on the ER (26–29). We expressed the Pex15-Psd-YFP fusion in psd1⌬ psd2⌬ dpl1⌬ cells that also contained the conditional alleles sec18-1 or sec16-2. After a 10-min shift to nonpermissive temperature, the rate of PtdSer to PtdEtn conversion in these strains did not slow (Fig. 4A). In contrast, vesicular transport from the ER is substantially decreased in these conditions (25, 30). We ruled out that these results could be explained by differences in the Psd activity of these strains (Fig. 4B). Thus, ER to peroxisome PtdSer transport does not require Sec18p or Sec16p. 15788 兩 www.pnas.org兾cgi兾doi兾10.1073兾pnas.0808321105

suggest that the transport of lipids from the ER to peroxisomes either requires a novel vesicular pathway or, more likely, is nonvesicular. To better understand lipid transfer between the ER and peroxisomes, we reconstituted this transport in vitro using purified microsomes containing [3H]PtdSer and peroxisomes containing Pex15-Psd-YFP. To obtain microsomes with [3H]PtdSer, total cellular membranes from a psd1⌬ psd2⌬ dpl1⌬ strain were labeled with [3H]serine. The membranes were pelleted and washed to remove unused [3H]serine. Peroxisomes were obtained by fractionating membranes from psd1⌬ psd2⌬ dpl1⌬ cells expressing Pex15Psd-YFP on sucrose gradients as described in Fig. 2B. The fractions enriched in the peroxisomal membrane protein Pex12p (fractions 13–16) were pooled. When the labeled microsomes and peroxisomes were incubated together for 1 h at 30°C, 2.5% of the [3H]PtdSer was converted to [3H]PtdEtn (Fig. 5A), indicating that the PtdSer had been transferred from the microsomes to the peroxisomes. No [3H]PtdEtn was formed in control reactions lacking peroxisomes. We calculated that the rate of conversion of [3H]PtdSer to PtdEth is 0.04% per minute, ⬇10 times slower than in intact cells. The formation of PtdEtn from PtdSer was not substantially enhanced by addition of cytosol and did not require ATP, suggesting that PtdSer transfer is nonvesicular. Interestingly, transport was enhanced ⬇2-fold by the addition of 5 mM ATP (Fig. 5A). We confirmed that peroxisomes remain intact during our assay. Treating peroxisomes from psd1⌬ psd2⌬ dpl1⌬ cells expressing Pex15-Psd-YFP with Pronase did not decrease the amount of Psd activity, indicating that the Psd in the peroxisomal lumen was not exposed to the cytosol (Fig. 5B). As a control, we showed that Psd-YFP expressed in the cytosol is almost completely digested by the same amount of Pronase (Fig. 5C). It should be noted that Psd undergoes an autocatalytic cleavage (32), and only the C-terminal fragment of Psd fused to GFP can be detected by immunoblotting with antibody that recognizes GFP. As an additional control, we showed that Pronase also degrades the peroxisomal membrane protein Pex12-TAP (Fig. 5D). Taken together, these findings show that PtdSer is transferred from the microsomes to peroxisomes in vitro by a mechanism that is nonvesicular. Discussion We show that PtdSer is transferred between the ER and peroxisomes. This has been demonstrated in an artificially engineered strain in which the endogenous Psds have been replaced by peroxisomally localized Psd but remains to be shown in wild-type cells. However, it seems likely that PtdSer and other phospholipids are transferred from the ER to peroxisomes in wild-type cells as well. Two findings suggest that this transport is rapid and efficient. First, it is well established that newly synthesized PtdSer is rapidly moved out of the ER to the endogenous Psds in wild-type cells, and our findings suggest that PtdSer is moved at a similar rate between the ER and peroxisomes. Second, we find that expression of Pex15Psd-YFP in psd1⌬ psd2⌬ dpl1⌬ significantly reduces the total Raychaudhuri and Prinz

transfer. Indeed, it has been suggested that sites of contact between the ER and a number of organelles may be important for nonvesicular lipid transfer (35, 36). Interestingly, PtdSer transfer to peroxisomes did not require ATP but was stimulated by ATP. This has also been found for PtdSer transport from ER to mitochondria in permeabilized mammalian cells (37). How ATP stimulates transport remains to be determined, but it may be that it is needed to maintain close contacts between membranes to facilitate transfer. Now that we have established a method to measure lipid transfer to peroxisomes, we can begin to understand the role of lipid trafficking in peroxisome biogenesis and to identify proteins required for phospholipid transport between the ER and peroxisomes.

Fig. 5. Phospholipid transfer between ER and peroxisomes in vitro does not does require ATP or cytosol. (A) Radiolabeled microsomes containing [3H]PtdSer were mixed with peroxisomes isolated from sucrose density gradients and incubated for 1 h at 30°C. The percent of PtdEtn formed from PtdSer was calculated by subtracting the amount of radiolabeled PtdEtn in control reactions containing microsomes alone. In some reactions, 5 mM ATP or 1 mg of total cytosolic proteins was included. (B) Peroxisomes were isolated from sucrose density gradients and incubated with or without 100 ␮g of Pronase. They were then used for a PtdSer transport assay as in A. (C) A lysate from cells expressing Psd-YFP in the cytosol were incubated with the indicated amount of Pronase and then immunoblotted with Anti-GFP antibodies. Note that Psd undergoes an autocatalytic cleavage and only the C-terminal fragment fused to GFP can be detected. (D) The same membranes used in A were used for immunoblotting to detect the peroxisomal membrane protein Pex12-TAP. In A and B, error bars indicate the SEM (n ⫽ 3).

amount of PtdSer in this strain, indicating that much of the PtdSer in cells is moved to peroxisomes. This finding also suggests that phospholipid transfer between the ER and peroxisomes is bidirectional, because only a very small fraction of the total PtdSer in cells is in peroxisomes (because peroxisomes are not abundant in glucose grown cells). Our findings demonstrate that a nonvesicular transport pathway moves phospholipids between the ER and peroxisomes. This is consistent with earlier studies suggesting that nonvesicular lipid transfer to peroxisomes may occur in the yeast Yarrowia lipolytica and plants (33, 34). Although lipids must also reach peroxisomes by vesicular transport, it is not yet possible to determine what fraction of the lipid needed for peroxisome membrane biogenesis reaches peroxisomes by vesicular and nonvesicular pathways. The nonvesicular pathway could be efficient enough to provide the lipids needed for the growth and division of peroxisomes. If this is the case, then the primary function of ER to peroxisome vesicular transport may not be to provide lipids to peroxisomes but to bring some integral membranes proteins to peroxisomes. These proteins could be too hydrophobic to be inserted into peroxisomal membranes posttranslationally. Nonvesicular lipid transport between the ER and peroxisomes probably also allows cells to rapidly regulate the amount and composition of lipids in peroxisomal membranes. The mechanism of lipid transport between ER and peroxisomes is not yet known but is likely similar to the nonvesicular phospholipid transport pathway that operates between ER and mitochondria (18). Lipid transfer between these organelles is thought to require the close apposition of their membranes. Close contacts between the ER and peroxisome might also be required for lipid Raychaudhuri and Prinz

Fluorescence Microscopy. Strains were grown in synthetic complete medium lacking histidine and uracil, and imaged live at room temperature by using an Olympus BX61 microscope, a UPlanApo ⫻100/1.35 lens, a QImaging Retiga EX camera, and the program IVision (version 4.0.5). Sucrose Density Gradients. Membranes were separated by sucrose density gradient centrifugation as described in ref. 4. Growing cells were lysed in a MiniBeadBeater-8 (BioSpec) in 20 mM Tris (pH 7.4), 20 mM EDTA, 10% sucrose, and protease inhibitors. After spinning twice at 500 ⫻ g for 10 min to remove debris, 4 mg of the lysate was loaded on a 20 – 40% sucrose gradient and centrifuged at 100,000 ⫻ g for 16 h at 4°C in a Beckman SW41 rotor. Fractions of 800 ␮l were collected from the top, and one set was used for the Psd assay with NBD-PS as substrate as described in the next section. The other part was precipitated with 10% trichloroacetic acid and used for immunoblotting to detect the ER membrane protein Dpm1p (Invitrogen) and the peroxisomal membrane protein Pex12-TAP with Peroxidase-Anti-Peroxidase (Sigma). Psd Assay. Lysates were prepared as described in the previous section. 1-Oleoyl2-[6-[(7-nitro-2–1,3-benzoxadiazol-4-yl)amino]hexanoyl]-sn-glycero-3-phosphoL-serine (NBD-PS) (Avanti Polar Lipids) was dried under nitrogen and solubilized in assay buffer [20 mM Tris (pH 7.4), 20 mM EDTA] to a final concentration of 25 mM. Cell lysate or fractions from gradients were added to the NBD-PS, and the final volume of the assay was maintained at 1 ml with assay buffer. The assay mixture was incubated at 37°C for the indicated times. Lipids were extracted by using 2:1 methanol:chloroform and analyzed by thin-layer chromatography, using the following solvent system: methanol:choloroform:acetone:water:acetic acid 10:50:20:5:15. Formation of NBD-PE was quantitatively detected by using a CAMAG TLC scanner and winCATS software. Metabolic Labeling with [3H]serine. Growing cells were labeled with L-[3(American Radiolabeled Chemicals) as described in ref. 39. Briefly, 100-ml cultures were grown to an OD600 of ⬇0.6 – 0.8, and then 1 mM S-adenosyl methionine, 2 mM methionine, and 10 ␮g/ml cerulenin (from a 5 mg/ml stock in DMSO) were added to the medium. After 2–5 min, 200 ␮Ci of [3H]serine (American Radiolabeled Chemicals) was added, and 25-ml samples were taken every 10 min and mixed with an equal volume of ice-cold H2O. The cells were then washed 3H]serine

PNAS 兩 October 14, 2008 兩 vol. 105 兩 no. 41 兩 15789

CELL BIOLOGY

Methods Strain, Plasmids, and Growth Conditions. The following yeast strains were used: BY4741 (MATa his3⌬1 leu2⌬ met15⌬ ura3⌬), WPY702 (BY4741 psd1::kanMX4 psd2::kanMX4 dpl1::kanMX4), EBY23 (sec18⫺1 WPY702), SRY71 (sec16-2 WPY702), SRY69 (pex3::LEU2 WPY702), SRY62 (pex3-B5:LEU2 WPY702), SRY75 (PEX12-TAP:HIS3MX6 WPY702), and SRY76 (PEX12-TAP:HIS3MX6 SRY69). To make SRY62, pex3-B5 was amplified by PCR from a plasmid (a gift from S. Gould, The Johns Hopkins University) and subcloned into YIplac128 (38) between the XbaI and KpnI sites. The resulting plasmid was used as template for PCR, and the product was used for linear transformation of WPY702. Transformants were selected on medium without leucine and screened for those that lacked peroxisomes at nonpermissive temperature. The plasmid pWP1297 encodes Pex15-Psd-YFP. It was made in YIplac211 (38) and has the following elements: the MET17 promoter, the coding sequence for Pex15, the coding sequence for E. coli Psd, the coding sequence for YFP, and the CYC1 terminator. It was integrated into PEX15 by cutting with BstEII and using the linear plasmid to transform strains. Transformants were selected on medium without uracil and verified by PCR. The plasmid pWP1209 is a HIS3/CEN plasmid that encodes CFP-SKL under the MET17 promoter. All strains were grown in synthetic complete medium (0.67% yeast nitrogen base and 2% glucose) with amino acids and 5 mM Etn.

once with H2O, and lipids were extracted (40). The major phospholipids were separated by HPLC (41) and counted by liquid scintillation counting. Quantification of Total Glycerophospholipids. Lipids were extracted as described in ref. 40 after cells were lysed in a Mini-BeadBeater-8 (BioSpec). Total lipids were dried under N2, resuspended in CHCl3, and injected onto a Zorbax RX-Sil 250 ⫻ 4.6-mm (5 ␮m) column (Agilent Technologies). Glycerophospholipids were separated basically as described in ref. 42. Solution A was CHCl3, methanol, and ammonium hydroxide (80:19:1); solution B was CHCl3, methanol, and ammonium hydroxide (60:39:1); and solution C was CHCl3, methanol, H2O, and ammonium hydroxide (60:34:5:1). Solution A was set at 100% for 3 min, and then solution B was increased to 100% by 16 min and held at 100% for 5 min. Solution C was then increased to 100% over 9 min and held at 100% for 2 min. The column was reequilibrated with solution A for 10 min before each injection. Lipids were detected with a PL-ELS 2100 evaporative light scattering detector (Polymer Laboratories) with an evaporator temperature of 105°C, a nebulizer temperature of 50°C, and a gas (N2) flow rate of 2 liters/min. The peaks containing the four major glycerolphospholipids were identified by comparison with known standards (Avanti Polar Lipids). Calibrations for mass amounts were based on external standards.

In Vitro Transport Assay. A psd1⌬ psd2⌬ dpl1⌬ strain was lysed in 50 mM Tris (pH 8.0) and 10% sucrose by using a bead-beater. The lysates were precleared two times by spinning at 500 ⫻ g for 10 min. Ten milligrams of lysate was used for labeling with 100 ␮Ci of [3H]serine at 37°C for 30 min. Microsomes were then pelleted at 100,000 ⫻ g for 30 min, washed by pelleting twice to remove free [3H]serine, and resuspended in 20 mM Tris (pH 7.4), 20 mM EDTA, and 10% sucrose. Peroxisomes were isolated from psd1⌬ psd2⌬ dpl1⌬ cells expressing Pex15-Psd-YFP by fractionation on sucrose density gradients as described above. The fractions containing peroxisomes (fractions 13–16) were pooled, and the

1. Purdue PE, Lazarow PB (2001) Peroxisome biogenesis. Annu Rev Cell Dev Biol 17:701– 752. 2. Platta HW, Erdmann R (2007) Peroxisomal dynamics. Trends Cell Biol 17:474 – 484. 3. Steinberg SJ, et al. (2006) Peroxisome biogenesis disorders. Biochim Biophys Acta 1763:1733–1748. 4. Hoepfner D, Schildknegt D, Braakman I, Philippsen P, Tabak HF (2005) Contribution of the endoplasmic reticulum to peroxisome formation. Cell 122:85–95. 5. Kragt A, Voorn-Brouwer T, van den Berg M, Distel B (2005) Endoplasmic reticulumdirected Pex3p routes to peroxisomes and restores peroxisome formation in a Saccharomyces cerevisiae pex3Delta strain. J Biol Chem 280:34350 –34357. 6. Tam YY, Fagarasanu A, Fagarasanu M, Rachubinski RA (2005) Pex3p initiates the formation of a preperoxisomal compartment from a subdomain of the endoplasmic reticulum in Saccharomyces cerevisiae. J Biol Chem 280:34933–34939. 7. Kim PK, Mullen RT, Schumann U, Lippincott-Schwartz J (2006) The origin and maintenance of mammalian peroxisomes involves a de novo PEX16-dependent pathway from the ER. J Cell Biol 173:521–532. 8. Karnik SK, Trelease RN (2007) Arabidopsis peroxin 16 trafficks through the ER and an intermediate compartment to pre-existing peroxisomes via overlapping molecular targeting signals. J Exp Bot 58:1677–1693. 9. Motley AM, Hettema EH (2007) Yeast peroxisomes multiply by growth and division. J Cell Biol 178:399 – 410. 10. Hettema EH, Girzalsky W, van Den Berg M, Erdmann R, Distel B (2000) Saccharomyces cerevisiae pex3p and pex19p are required for proper localization and stability of peroxisomal membrane proteins. EMBO J 19:223–233. 11. Honsho M, et al. (1998) Mutation in PEX16 is causal in the peroxisome-deficient Zellweger syndrome of complementation group D. Am J Hum Genet 63:1622–1630. 12. South ST, Gould SJ (1999) Peroxisome synthesis in the absence of preexisting peroxisomes. J Cell Biol 144:255–266. 13. South ST, Baumgart E, Gould SJ (2001) Inactivation of the endoplasmic reticulum protein translocation factor, Sec61p, or its homolog, Ssh1p, does not affect peroxisome biogenesis. Proc Natl Acad Sci USA 98:12027–12031. 14. South ST, Sacksteder KA, Li X, Liu Y, Gould SJ (2000) Inhibitors of COPI and COPII do not block PEX3-mediated peroxisome synthesis. J Cell Biol 149:1345–1360. 15. Bishop WR, Bell RM (1988) Assembly of phospholipids into cellular membranes: Biosynthesis, transmembrane movement and intracellular translocation. Annu Rev Cell Biol 4:579 – 610. 16. Zinser E, et al. (1991) Phospholipid synthesis and lipid composition of subcellular membranes in the unicellular eukaryote Saccharomyces cerevisiae. J Bacteriol 173:2026 –2034. 17. Schekman R (2005) Peroxisomes: Another branch of the secretory pathway? Cell 122:1–2. 18. Choi JY, Wu WI, Voelker DR (2005) Phosphatidylserine decarboxylases as genetic and biochemical tools for studying phospholipid traffic. Anal Biochem 347:165–175. 19. Elgersma Y, et al. (1997) Overexpression of Pex15p, a phosphorylated peroxisomal integral membrane protein required for peroxisome assembly in S. cerevisiae, causes proliferation of the endoplasmic reticulum membrane. EMBO J 16:7326 –7341. 20. Achleitner G, et al. (1999) Association between the endoplasmic reticulum and mitochondria of yeast facilitates interorganelle transport of phospholipids through membrane contact. Eur J Biochem 264:545–553. 21. Atkinson K, Fogel S, Henry SA (1980) Yeast mutant defective in phosphatidylserine synthesis. J Biol Chem 255:6653– 6661. 22. Shimozawa N, et al. (2000) Identification of PEX3 as the gene mutated in a Zellweger syndrome patient lacking peroxisomal remnant structures. Hum Mol Genet 9:1995– 1999.

15790 兩 www.pnas.org兾cgi兾doi兾10.1073兾pnas.0808321105

labeled microsomes were added. After 1 h at 30°C, the reaction was stopped by the addition of 6 ml of methanol:CHCl3 (2:1), and lipid extraction was performed. The PtdSer and PtdEtn were separated by HPLC (41) and counted by liquid scintillation counting. The amount of PtdEtn formed from PtdSer was calculated by subtracting the amount of radiolabeled PtdEtn in control reactions containing microsomes alone. Protease Protection Assay. Peroxisomal membranes from psd1⌬ psd2⌬ dpl1⌬ cells expressing Pex15-Psd-YFP were partially purified from sucrose density gradients as described above. The fraction containing peroxisomes (fractions 13–16) were pooled. To 0.5 ml of the pooled fractions, 100 ␮g of Pronase (Fluka) was added, and samples were incubated at 37°C for 30 min. The reaction was stopped by the addition of protease inhibitors and placed on ice for 10 min. Psd activity was determined by adding 10 ␮M [3H]PtdSer and Triton X-100 to 0.1%, and the reactions were incubated for 1 h at 30°C. The amount of radiolabeled PtdSer and PtdEtn was determined as described above. The [3H]PtdSer was synthesized by incubating ⬇10 mg of microsomes with 1 mCi of [3H]serine in 20 mM Tris (pH 8.0) and 0.6 mM MnCl2 for 30 min at 37°C. The [3H]PtdSer was then purified by HPLC as described above. For control reactions, ⬇1 mg of lysate from psd1⌬ psd2⌬ dpl1⌬ cells expressing Psd-YFP was lysed in a Mini-BeadBeater-8 (BioSpec) in 20 mM Tris (pH 7.4), 20 mM EDTA, and 10% sucrose. After a spinning twice at 500 ⫻ g for 10 min to remove debris, ⬇1 mg of lysate was incubated with different amounts of Pronase and incubated at 37°C for 30 min. The samples were then immunoblotted with antibodies against GFP. ACKNOWLEDGMENTS. We thank S. Gould for plasmids and advice, and O. Cohen-Fix, T. Schulz, and J. Hanover for critical reading of the manuscript. This work was supported by the intramural research program of the National Institute of Diabetes and Digestive and Kidney Diseases.

23. Choi JY, Martin WE, Murphy RC, Voelker DR (2004) Phosphatidylcholine and Nmethylated phospholipids are nonessential in Saccharomyces cerevisiae. J Biol Chem 279:42321– 42330. 24. Wilson DW, et al. (1989) A fusion protein required for vesicle-mediated transport in both mammalian cells and yeast. Nature 339:355–359. 25. Graham TR, Emr SD (1991) Compartmental organization of Golgi-specific protein modification and vacuolar protein sorting events defined in a yeast sec18 (NSF) mutant. J Cell Biol 114:207–218. 26. Nakano A, Muramatsu M (1989) A novel GTP-binding protein, Sar1p, is involved in transport from the endoplasmic reticulum to the Golgi apparatus. J Cell Biol 109:2677– 2691. 27. Kaiser CA, Schekman R (1990) Distinct sets of SEC genes govern transport vesicle formation and fusion early in the secretory pathway. Cell 61:723–733. 28. Gimeno RE, Espenshade P, Kaiser CA (1995) SED4 encodes a yeast endoplasmic reticulum protein that binds Sec16p and participates in vesicle formation. J Cell Biol 131:325–338. 29. Supek F, Madden DT, Hamamoto S, Orci L, Schekman R (2002) Sec16p potentiates the action of COPII proteins to bud transport vesicles. J Cell Biol 158:1029 –1038. 30. Espenshade P, Gimeno RE, Holzmacher E, Teung P, Kaiser CA (1995) Yeast SEC16 gene encodes a multidomain vesicle coat protein that interacts with Sec23p. J Cell Biol 131:311–324. 31. Fang Y, Morrell JC, Jones JM, Gould SJ (2004) PEX3 functions as a PEX19 docking factor in the import of class I peroxisomal membrane proteins. J Cell Biol 164:863– 875. 32. Li QX, Dowhan W (1988) Structural characterization of Escherichia coli phosphatidylserine decarboxylase. J Biol Chem 263:11516 –11522. 33. Chapman KD, Trelease RN (1991) Acquisition of membrane lipids by differentiating glyoxysomes: Role of lipid bodies. J Cell Biol 115:995–1007. 34. Titorenko VI, Eitzen GA, Rachubinski RA (1996) Mutations in the PAY5 gene of the yeast Yarrowia lipolytica cause the accumulation of multiple subpopulations of peroxisomes. J Biol Chem 271:20307–20314. 35. Voelker DR (2005) Bridging gaps in phospholipid transport. Trends Biochem Sci 30:396 – 404. 36. Levine T, Loewen C (2006) Inter-organelle membrane contact sites: Through a glass, darkly. Curr Opin Cell Biol 18:371–378. 37. Voelker DR (1989) Phosphatidylserine translocation to the mitochondrion is an ATPdependent process in permeabilized animal cells. Proc Natl Acad Sci USA 86:9921– 9925. 38. Gietz RD, Sugino A (1988) New yeast-Escherichia coli shuttle vectors constructed with in vitro mutagenized yeast genes lacking six-base pair restriction sites. Gene 74:527– 534. 39. Raychaudhuri S, Im YJ, Hurley JH, Prinz WA (2006) Nonvesicular sterol movement from plasma membrane to ER requires oxysterol-binding protein-related proteins and phosphoinositides. J Cell Biol 173:107–119. 40. Parks LW, Bottema CD, Rodriguez RJ, Lewis TA (1985) Yeast sterols: Yeast mutants as tools for the study of sterol metabolism. Methods Enzymol 111:333–346. 41. Wang F, Zhao Y, Wang P (2003) Separation and determination of phospholipids in biological samples by high-performance liquid chromatography. J Chromatogr Sci 41:142–144. 42. Stith BJ, et al. (2000) Quantification of major classes of Xenopus phospholipids by high performance liquid chromatography with evaporative light scattering detection. J Lipid Res 41:1448 –1454.

Raychaudhuri and Prinz