Cargo sorting into multivesicular bodies in vitro

0 downloads 0 Views 1MB Size Report
Oct 13, 2009 - with free biotin in vitro by Escherichia coli biotin ligase (BirA) (7). A trypsin ..... yeast genomic DNA into bacterial plasmid pGEX-2T and was.
Cargo sorting into multivesicular bodies in vitro John H. Trana,1, Ching-Jen Chena, Scott Emrb, and Randy Schekmana,2 aDepartment of Molecular and Cell Biology, and Howard Hughes Medical Institute, University of California, Berkeley, CA 94720; and bCornell University, Weill Institute for Cell and Molecular Biology and Department of Molecular Biology and Genetics, Ithaca, NY 14853

Contributed by Randy Schekman, August 19, 2009 (sent for review July 16, 2009)

endosome 兩 multivesicular bodies 兩 reconstitution

M

ultivesicular bodies (MVBs) lie at the intersection of pathways that target biosynthetic and plasma membrane proteins to the lysosome/vacuole. The MVB pathway controls important processes in multicellular organisms such as receptor downregulation, antigen presentation, cytokinesis, neuronal survival, and retroviral budding (1). MVBs are a subset of late endosomes that have undergone internal vesicle budding, allowing membrane proteins entering the MVB pathway to reach the vacuole lumen instead of the vacuole limiting membrane. Carboxypeptidase S (CPS) is a biosynthetic MVB cargo protein synthesized as a type II transmembrane precursor protein (2). It has a short N-terminal tail exposed to the cytosol followed by a transmembrane domain and a C terminus comprising the lumenal, protease catalytic domain. When CPS reaches the endosomal limiting membrane, it is ubiquitylated and included in internal vesicles of the MVB (3). Fusion of the MVB with the vacuole releases the internal vesicles into the vacuole lumen, where CPS is proteolytically cleaved by proteinase A (Pep4p) to release the CPS (mCPS) C-terminal catalytic domain into the vacuole lumen (2). Previous research using yeast genetics has identified a number of the vacuolar protein sorting (VPS) genes necessary for proper sorting of vacuolar proteins. The vps mutants fall into six classes that correlate to distinct steps in the biosynthesis of vacuolar proteins (4). Class E proteins define the MVB pathway. Proteins of this class make up five distinct endosomal sorting complexes required for transport (ESCRT) complexes, including Vps27p, ESCRTs I, II, and III, and Vps4p. Defects in each of the seventeen class E VPS genes prevent internal vesicle formation and thus result in an increase in the endosome surface area, creating an exaggerated compartment called the class E compartment (1). Recent efforts to reconstitute MVB formation in vitro have been reported. In one study, the internalization of a fluid marker into vesicles and sorting of EGFR were observed (5). In another, a component of the ESCRT III complex, Snf7, was found to promote membrane invagination and scission of vesicles internalized into giant unilamellar vesicles (GUVs) (6). In this study, we describe a reconstitution reaction that tracks the internalization of a cargo protein and offers the possibility of a functional biochemical characterization of the entire ESCRT pathway. www.pnas.org兾cgi兾doi兾10.1073兾pnas.0909473106

Results BBT-CPS Properly Sorts to Late Endocytic, Pep4p-Containing Compartments. To establish an in vitro reconstitution assay for MVB

formation, a chimeric version of CPS was designed that allowed us to biochemically tag a cargo protein at the surface of the endosome (Fig. 1A and Fig. S1 A). CPS is a type II transmembrane protein, with a short, 19-aa N-terminal tail exposed to the cytosol, as the precursor transits from the ER via the Golgi to endosomes. The bulk of the 576-aa long precursor is found in the lumen of the compartment through which it transits until it is sorted into internal vesicles at the endosome. CPS exists in two glycosylation states (Fig. 1B) (2) that appear as a doublet on a gel. The chimeric form of CPS, termed BBT-CPS, contained N-terminal tandem biotin acceptor peptides (BAPs) that have been shown to be specifically labeled with free biotin in vitro by Escherichia coli biotin ligase (BirA) (7). A trypsin cleavage site (TCS) followed, containing five arginine (R)/lysine (K) amino acids (RKKRK). The remaining protein consisted of the native CPS N terminus, including K8 for ubiquitylation by the cell, and the transmembrane domain. At the C terminus, a 3⫻HA epitope was included to monitor the integrity of the endosomal membrane during the in vitro incubation. We examined the fate of the tagged CPS in vivo and confirmed proper sorting to the vacuole and endosome (Fig. S1). In Vitro Reconstitution Assay. To take advantage of transiting steps

in the normal trafficking of CPS, we used vps27ts, an MVB mutant previously shown to trap CPS at the endosome and vacuole limiting membranes (3). vps27ts contains a temperature-sensitive allele of the VPS27 gene (8). Vps27p is the ESCRT protein that initiates the capture of ubiquitylated cargoes at the endosome surface. The vps27ts mutant accumulates CPS at the surface of endosomes at 37 °C. Mutant endosomes harboring accumulated CPS chimeric molecules were isolated from vps27ts mutant cells incubated for 2.5 h at 37 °C. The in vitro reconstitution reaction consisted of three steps: (Step 1) in vitro surface biotin labeling of the accumulated BBT-CPS cargo by E. coli BirA; (Step 2) incubation of membranes containing biotinylated chimeric BBT-CPS cargo with rATP (9), cytosol, and recombinant proteins; (Step 3) trypsin cleavage of surface-exposed biotin. This reaction permitted us to selectively biotin label only the CPS cargo molecules with their N termini still exposed at the surface of the endosome and then follow the fate of the chimeric CPS cargo. As depicted in Fig. 1 A, we predicted that a successful reaction conducted with mutant membranes and wild-type cytosol would result in BBT-CPS internalization (step 2) and thus sequestering biotinylated CPS precursor (step 3, bottom), whereas a control incubation with BSA in place of cytosol would retain biotinylated Author contributions: J.H.T. and R.S. designed research; J.H.T. and C.-J.C. performed research; S.E. contributed new reagents/analytic tools; J.H.T., S.E., and R.S. analyzed data; and J.H.T. and R.S. wrote the paper. The authors declare no conflict of interest. 1Present address: Cell Biology Program, Memorial Sloan–Kettering Cancer Center, 430 East

67th Street, RRL 661, New York, NY 10065. E-mail: [email protected]. 2To

whom correspondence should be addressed. E-mail: [email protected].

This article contains supporting information online at www.pnas.org/cgi/content/full/ 0909473106/DCSupplemental.

PNAS 兩 October 13, 2009 兩 vol. 106 兩 no. 41 兩 17395–17400

CELL BIOLOGY

Genetic studies have identified a number of proteins required for the internalization of biosynthetic and endocytic cargo proteins transported to the multivesicular body (MVB). We have developed a cell-free reaction that recapitulates the internalization of a yeast biosynthetic membrane cargo protein, carboxypeptidase S (CPS), into the interior of an endosome. A recombinant form of CPS containing a biotinylation site from an Escherichia coli protein is accumulated in a vps27 yeast mutant blocked in the MVB internalization event. Endosomes isolated from the vps27 mutant are exposed to E. coli biotin ligase, which acts on only those CPS molecules with a cytosolexposed N-terminal domain. Internalization of biotin-tagged CPS is measured by the detection of trypsin-inaccessible, membrane-protected species. Biotinylated CPS internalization requires ATP and functional forms of Vps27p and Vps4p and depends on the availability of an exposed lysine residue critical for CPS ubiquitylation.

Fig. 1. A cell-free system to study MVB formation. (A) The in vitro reconstitution assay consisted of three steps. (Step 1) Biotinylation. A chimeric form of the cargo CPS, termed BBT-CPS and trapped at the endosome surface as described in the text, was labeled specifically by E. coli biotin ligase to distinguish it from already invaginated CPS in vacuoles. (Step 2) Internalization. Biotinylated BBT-CPS was sorted into internal vesicles and localized to the endosome lumen. (Step 3) Proteolysis. Any remaining surface-localized BBT-CPS was treated with trypsin [except in Fig. 1 D and E, in which PK was used], which resulted in removal of the N-terminal tail containing the tandem biotin tags. BAP-biotin acceptor peptide; TCStrypsin cleavage site; HA-3⫻HA epitope. (B) Wild-type (WT) cytosol reconstitution reaction. WT cytosol (lanes 1–3) or BSA (lanes 4 – 6) was incubated with vps27ts membranes containing BBT-CPS under standard reaction conditions. Trypsin was incubated with two duplicate reactions for the WT cytosol (lanes 2–3) and BSA (lanes 5– 6). Biotin signals from proteolyzed samples were divided by unproteolyzed controls (lanes 1 and 4) to calculate percent biotin protection. Unproteolyzed samples contained 3-fold less total protein sample than proteolyzed samples. The calculated percent biotin protection values were 17%, 19%, 3.6%, 3.4% for lanes 2, 3, 5, 6. (C) Quantitation of biotin protection in panel B. Experiments were performed three times, and the relative biotin protection was measured as a percentage of the WT cytosol biotin signal, which was set at 1. The values were plotted as the mean ⫾ standard deviation. (D) WT cytosol (lanes 1–3) or BSA (lanes 4 – 6) was incubated with vps27ts membranes containing BB-CPS in membrane buffer for 1 h at 23 °C. PK was incubated in two duplicate reactions for the WT cytosol (lanes 2–3) and BSA (lanes 5– 6). Triton was added in lanes 3 and 6. Unproteolyzed samples contained the same amount of total protein sample as proteolyzed samples. The calculated percent biotin protection values were 14%, 1.2%, 1.7%, 0.3% for lanes 2, 3, 5, 6. (E) Quantitation of biotin protection in panel B. Experiments were performed three times, and the relative biotin protection was measured as a percentage of WT cytosol biotin signal, which was set at 1. The values were plotted as the mean ⫾ standard deviation.

CPS precursor in a trypsin-accessible form (step 3, top). An incubation conducted in these two conditions produced approximately 3-fold more protease protected CPS than in the control (Fig. 1 B and C). The reaction was performed with cytosol from a wild-type strain containing all ESCRT complexes and deficient for two major vacuolar proteases (Pep4p and Prb1p) and a proteasome component (Pre1p). To assess the integrity of membranes used in the reactions, proteinase K (PK) was incubated with membranes containing BB-CPS cargo. We used PK because trypsin failed to cleave the lumenal side of BB-CPS and thus could not be used to assess membrane integrity. Similar to the trypsin reaction (Fig. 1 B and C), WT cytosol elicited BB-CPS cargo protection when PK was used to assess internalization (compare Fig. 1 B and D). When reactions 17396 兩 www.pnas.org兾cgi兾doi兾10.1073兾pnas.0909473106

were incubated with Triton, both the residual biotin signal and the HA signal were cleaved by PK, indicating the BB-CPS protection observed in the reactions incubated only with WT cytosol was due to internalization (Fig. 1D, lanes 5–6). Finally, in the absence of E. coli BirA in step 1 (Fig. 1D, lane 7), no detectable biotin signal was observed, indicating that the BAP is specifically labeled by BirA from E. coli and not by the yeast biotin ligase. We tested 15-min, 30-min, and 1-h incubations using the BB-CPS cargo in vps27ts membranes for step 2 (internalization step) and found that the reaction efficiency was similar for reactions performed at 30 min (36.8 ⫾ 6.1% protease protection; n ⫽ 9) and 1 h (29.7 ⫾ 15.9% protease protection; n ⫽ 8) and lower at 15 min (22.2 ⫾ 5.9% protease protection; n ⫽ 5). The higher variability at 1 h might be attributed to the loss of membrane integrity after Tran et al.

prolonged incubations at 23 °C. Therefore, we chose 30 min for step 2 to achieve a reliable signal above the background. MVB Formation Is ATP and Temperature-Dependent. Vps4p is the only

nucleotide-binding enzyme known to be directly involved in MVB formation. Mutations in the ATP binding domain of Vps4p result in MVB mutant phenotypes; thus MVB formation should be ATP-dependent (10). Our standard reaction contained ATP and an ATP regeneration system (rATP) (Fig. 2A, lanes 1–4). When rATP was left out of the reaction and replaced by apyrase, an enzyme that hydrolyzes ATP, the efficiency of BBT-CPS internalization was reduced approximately 2-fold (Fig. 2A, lanes 5–8). To test the effect of temperature, complete reactions as described in Fig. 2 A, in lanes 1–4, were incubated on ice (4 °C) (lanes 9–12). The incubation at 4 °C almost completely abolished the BBT-CPS trypsin protection, indicating that MVB formation is a temperature-dependent event. Although a GTP-binding protein had not been implicated in MVB formation, we tested this possibility indirectly using a nonhydrolyzable GTP analog that often inhibits reactions requiring GTP hydrolysis. We found that GTP␥s had no effect on the reaction. MVB Formation Is ESCRT-Dependent. MVB sorting and internalization require cytosolic ESCRT complexes that recognize ubiquitylated cargo as well as the phosphatidylinositol-3-phosphate [PI (3)P] phospholipid enriched in endosomal membranes (1). We tested the requirement for two ESCRT proteins in our cell-free reaction. Vps27p is the initial protein in the pathway to recognize Tran et al.

Fig. 3. The role of ESCRTs in MVB formation. (A) WT cytosol was incubated with vps27ts membranes at 30 °C (lanes 1–3) and 4 °C (lanes 4 – 6). vps27⌬ cytosol was incubated with vps27ts membranes at 30 °C (lanes 7–9) and 4 °C (lanes 10 –12). Biotin signals from duplicate proteolyzed samples (lanes 2–3; 5– 6; 8 –9; 11–12) were divided by their respective unproteolyzed controls (lanes 1, 4, 7, 10), and the ratios were averaged to yield percent biotin protection. Unproteolyzed samples contained 3-fold less total protein sample than proteolyzed samples. The calculated percent biotin protection values were 21%, 27%, 0.2%, 0.7%, 9%, 7%, 2.4%, 2.5% for lanes 2, 3, 5, 6, 8, 9, 11, 12. (B) In a parallel experiment, compared to WT cytosol incubated with vps27ts membranes at 30 °C (lanes 1–3), vps4⌬ cytosol was incubated with vps27ts membranes at 30 °C (lanes 4 – 6) and 4 °C (lanes 7–9), and vps27⌬/vps4⌬ cytosol was incubated with vps27ts membranes at 30 °C (lanes 10 –12) and 4 °C (lanes 13–15). Biotin signals from duplicate proteolyzed samples (lanes 2–3; 5– 6; 8 –9; 11–12; 14 –15) were divided by their respective unproteolyzed controls (lanes 1, 4, 7, 10, 13), and the ratios were averaged to yield percent biotin protection. Unproteolyzed samples contained 3-fold less total protein sample than proteolyzed samples. The calculated percent biotin protection values were 20%, 17%, 7.9%, 7.3%, 0.5%, 1%, 7.9%, 4.5%, 3.1%, 2.1% for lanes 2, 3, 5, 6, 8, 9,11, 12, 14, 15. (C) Quantitation of biotin protection. Experiments were performed at least three times, except the 4 °C reactions for the vps27⌬/vps4⌬ and vps4⌬ cytosol, which were performed once. Relative biotin protection was measured as a percentage of WT cytosol at 30 °C, which was set at 1. The values were plotted as the mean ⫾ standard deviation.

ubiquitylated cargo on the endosome (1). At the end of the process, Vps4p removes the ESCRT proteins from the endosome surface for subsequent rounds of invagination (10). We replaced VPS27 and VPS4 by integration of KanMx4 and NatMx4 into their chromosomal loci in a strain deficient for the proteases Pep4p, Prb1p, and Pre1p. The deletions of VPS27 and VPS4 were confirmed by PCR amplification, DNA sequencing, and immunoblot using antibodies against Vps27p and Vps4p, respectively. Reconstitution assays were performed using either wild-type cytosol or cytosol from the mutant strains. Each mutant cytosol resulted in an ⬇2-fold reduction in efficiency of BBT-CPS trypsin protection relative to wild-type, indicating that the MVB formation was impaired (Fig. 3). Incubation with vps27⌬ vps4⌬ double mutant cytosol also resulted in a reduction in the efficiency, similar to either single-mutant cytosol. All assays incubated on ice showed marked reduction in internalization of BBT-CPS biotin. These data confirmed the essential role of Vps27p and Vps4p in MVB formation and the relevance of the in vitro biochemical assay to in vivo conditions. PNAS 兩 October 13, 2009 兩 vol. 106 兩 no. 41 兩 17397

CELL BIOLOGY

Fig. 2. The role of ATP in MVB formation. (A) vps27ts membranes were incubated with cytosol from a wild-type strain (WT cytosol) under standard reaction conditions (lanes 1– 4), or at 4 °C (lanes 9 –12), or in the absence of ATP and presence of 30 units/mL apyrase (lanes 5– 8). Biotin signals from triplicate protease treated lanes (lanes 2– 4; 6 – 8; 10 –12) were divided by their respective unproteolyzed controls (lanes 1, 5, 9), and the ratios were averaged to yield percent biotin protection. Unproteolyzed samples contained 3-fold less total protein sample than proteolyzed samples. The calculated percent biotin protection values were 19%, 21%, 28%, 8.5%, 6.2%, 8.5%, 1.6%, 1.6%, 1.6% for lanes 2– 4; 6 – 8; 10 –12. (B) Quantitation of biotin protection. Experiments were performed at least five times, and the relative biotin protection was measured as a fraction of WT cytosol with ATP at 30 °C, which was set at 1. The values were plotted as the mean ⫾ standard deviation.

Fig. 4. The role of Vps27p in MVB formation. (A) To test the effects of rVps27p on MVB formation, 500 nM rVps27p (lanes 1–3) or 500 nM BSA (lanes 4 – 6) was incubated with vps27ts membranes under standard reaction conditions. Unproteolyzed samples in lanes 1 and 4 contained 3-fold less total protein sample than proteolyzed samples in lanes 2–3 and 5– 6. The calculated percent biotin protection values were 17%, 16%, 7.7%, 7.7% for lanes 2, 3, 5, 6. (B) rVps27p (500 nM) was incubated with vps27⌬ membranes at 30 °C (lanes 1– 4) or at 4 °C (lanes 5– 8). Unproteolyzed samples in lanes 1 and 5 contained 3-fold less total protein sample than proteolyzed samples in lanes 2– 4 and 6 – 8. The calculated percent biotin protection values were 18%, 16%, 16%, 2%, 1.2%, 1.8% for lanes 2, 3, 4, 6, 7, 8. (C) Membranes extracted from a vps27ts strain and washed in 75 mM KCl or 500 mM KCl, were analyzed for presence of Vps23p, Snf7, Vps4p, and HA by immunoblotting. Arrows indicate residual proteins after high-salt wash. (D) vps27ts membranes from 75 mM KCl (low salt) and 500 mM KCl (high salt) washes were incubated with 500 nM rVps27p or BSA under standard reaction conditions. The results of at least four experiments are shown. Percent biotin protection was measured as a percentage of rVps27p incubated with membranes washed in low salt, which was set at 1. The values were plotted as the mean ⫾ standard deviation.

Recombinant Vps27p (rVps27p) Promotes MVB Formation in vps27 Mutant Strains. To further investigate the role of Vps27 protein in

MVB formation, we tested if recombinant Vps27 (rVps27p) could facilitate MVB formation in membranes isolated from vps27ts mutant strains. Recombinant Vps27 (rVps27p) was cloned from yeast genomic DNA into bacterial plasmid pGEX-2T and was expressed and purified using E. coli BL21(DE3). The purified protein was cleaved by thrombin to remove the GST tag. rVps27p was added to the in vitro reconstitution assay using membranes from the vps27ts strain under standard reaction conditions but without wild-type cytosol. BBT-CPS trypsin protection was approximately 2-fold higher than the protection observed in a reaction with BSA alone (Fig. 4A). To further confirm this finding, we created a vps27⌬ mutant and used the membranes from this strain to test the effect of rVps27p, with results similar to those seen with vps27ts mutant membranes (Fig. 4B). The ability of rVps27p to replace cytosol in this reaction suggested that other ESCRT proteins may remain associated with the membranes at levels sufficient to sustain CPS internalization. To test this hypothesis, we isolated membranes from the vps27ts strain and subjected them to 500 mM KCl salt washes, which were then used in the in vitro reconstitution assay. We postulated that high-salt (500 mM) washes would remove ESCRT proteins from the membranes and thus diminish the effect of rVps27p. After the high-salt washes, the membranes were probed for the presence of representative subunits of three ESCRT complexes: Vps23p (ESCRT I), Snf7p (ESCRT III), and Vps4p (Fig. 4C). As a protein control, BBT CPS-HA was found to be present at approximately similar levels in all lanes. These Vps proteins persisted in membranes washed in membrane buffer containing 75 mM KCl, but not in a buffer containing 500 mM KCl (Fig. 4C). However, trace 17398 兩 www.pnas.org兾cgi兾doi兾10.1073兾pnas.0909473106

amounts of Snf7p and Vps4p were observed associated with the membranes even after the high-salt wash. When high-salt washed membranes were used in the in vitro reconstitution reaction in the presence of rVps27p and rATP under standard reaction conditions, the positive effect of rVps27p on MVB formation observed in the low-salt washed membrane was decreased (Fig. 4D). The residual reaction could be from the ESCRT proteins remaining after a high-salt wash (Fig. 4C). In addition, the low level of BBT-CPS trypsin protection produced with high-salt washed membranes incubated with rVPS27 was comparable to the protection observed in the presence of BSA alone. We conclude that recombinant Vps27p is active in our reconstitution, although it is not likely to be sufficient for CPS internalization. A Lysine Implicated in Cargo Ubiquitylation Is Important in CPS Internalization. In general, cargo trafficking through the MVB

pathway are ubiquitylated and targeted to the endosomal surface, where they interact with the proteins of Vps27p, ESCRT I, and II complexes via a ubiquitin moiety. Previous genetic studies of MVB formation have implicated ubiquitylation as a major regulatory step in the pathway; however, the role of ubiquitylation was not evaluated in the reconstituted MVB reactions published recently (5, 6). We used BBT-CPS that contains the entire native amino acid sequence, including the lysine 8 (K8) residue that is the target for ubiquitylation. To test for the importance of K8 in cargo sorting, we mutated K8 to R in BBT-CPS, because the K8R mutation had been to shown cause sorting deficiencies in vivo (3). The BBT K8 CPS was designated WT CPS, and the arginine mutant was designated K8R CPS (Fig. 5A). Upon incubation of WT CPS- or K8R CPS-containing memTran et al.

branes and WT cytosol under normal reaction conditions, WT CPS underwent internalization with approximately 2-fold higher efficiency than K8R CPS (Fig. 5 B and C). Addition of 500 nM rVps27p elicited a similar differential effect: The WT CPS was sorted into internal vesicles with 2-fold greater efficiency than the K8R CPS. In comparison, a reaction consisting of WT CPS containing membranes and 500 nM rVps27p on ice was approximately 17-fold lower in internalization efficiency. We infer that this residue is important as a ubiquitin target, although this remains to be documented in the context of our cell-free reaction. Discussion MVB formation is a complex, well-coordinated process that involves multiple protein complexes recruited to the endosome membrane. A transmembrane cargo protein that is ubiquitylated on at least one lysine residue (monoubiquitylation) activates a cascade of ESCRT protein machinery, the first three of which have domains that recognize ubiquitin and endosomal membranes (Vps27, ESCRT I, and ESCRT II) (1). Vps27p is the first protein in the pathway to recognize ubiquitylated cargo on the endosome, resulting in sequential recruitment of ESCRTs I, II, and III. The ESCRT III complex is likely the downstream complex, because it does not assemble on the endosome without Tran et al.

PNAS 兩 October 13, 2009 兩 vol. 106 兩 no. 41 兩 17399

CELL BIOLOGY

Fig. 5. The role of CPS K8 on MVB formation. (A) Diagram of BBT-CPS constructs. K8 represents the wild-type form of CPS that is ubiquitylated by the cell, while R8 is the K8R mutant unable to be ubiquitylated. An ‘‘X’’ was placed over the R8 to indicate loss of ability to be ubiquitylated. (B) To test the effects of K8R on MVB formation, WT cytosol was incubated with vps27ts membranes containing BBT WT CPS (lanes 1–3) and BBT K8R CPS (lanes 4 – 6) under standard reaction conditions. rVps27p at 500 nM was incubated with vps27ts membranes containing BBT WT CPS (lanes 7–9) and BBT K8R CPS (lanes 10 –12) under standard reaction conditions. Unproteolyzed samples in lanes 1, 4, 7, 10 contained 3-fold less total protein sample than proteolyzed samples in lanes 2–3, 5– 6, 8 –9, and 11–12. The calculated percent biotin protection values were 21%, 17%, 9.8%, 9.2%, 20%, 21%, 11%, 11% for lanes 2, 3, 5, 6, 8, 9,11, 12. (C) Quantitation of biotin protection. Experiments were performed at least four times, and the relative biotin protection was measured as a fraction of WT cytosol or rVps27p incubated with WT CPS, which was set at 1. The values were plotted as the mean ⫾ standard deviation.

ESCRT-II, and ESCRT-III components do not have any ubiquitin binding motifs. Moreover, ESCRT III has been shown to form internal vesicles in GUVs and to form spiral rings when overexpressed at the plasma membrane (6, 11). Therefore, ESCRT III has been postulated to have a more mechanistic role in membrane deformation. Finally, the AAA ATPase Vps4p couples its ATP hydrolysis activity to removal of ESCRT complexes from the membrane surface for recycling. Even though the MVB pathway has been studied extensively, the mechanism of cargo segregation and internal vesicle formation and fission have not been fully elucidated. A recent report by Falguie`res and Gruenberg (5) documented reconstitution of MVB formation in semi-intact cells by monitoring the uptake of a fluorescent dye and EGFR sorting into internalized vesicles. To further understand the MVB pathway, we report establishment of a biochemical in vitro reconstitution assay monitoring a native cargo protein trafficking in the MVB pathway. We created a chimeric form of CPS, BBT-CPS that can be labeled with tandem biotin tags at its N terminus, and expressed it at a nonpermissive temperature using a derivatized vps27ts strain. Previous studies of the vps27ts strain showed that, upon shifting to a nonpermissive temperature, the mutation results in an enlarged defective endosome that traps cargo and exhibits aberrant cargo protease maturation and sorting phenotypes (8). Importantly, a shift back to permissive temperature restores the distribution of cargo to the vacuole concomitant to the disappearance of the enlarged endosome. These findings indicate that the endosome defects in the vps27ts strain are reversible and thus suitable for use in a cell-free reaction. In our assay, in vitro biotinylation, which marks BBT-CPS at the endosome surface, was completely susceptible to trypsin. Cytosol extracted from a wild-type strain and incubated with vps27 mutant membranes elicited trypsin protection of BBT-CPS-biotinylated cargo, indicative of sorting of this cargo into internal vesicles. The internalization reaction was performed for 30 min, as that incubation time ensured that the reaction was carried to near completion with the least variability. Cytosols from vps27⌬, vps4⌬, and vps27⌬ vps4⌬ strains elicited reduced BBT-CPS trypsin protection relative to cytosol from a wild-type strain, demonstrating the importance of Vps27p and Vps4p in the pathway. The loss of Vps27p and Vps4p resulted in similar reduction of internalization efficiency. However, the defects in vps27⌬ and vps4⌬ cytosols may be a reflection of two different steps in MVB formation. Whereas loss of Vps27p reflects a direct inability to sort CPS into internal vesicles because it recognizes ubiquitylated cargo, loss of Vps4p may reflect inability to carry out multiple rounds of the MVB sorting reaction because it is required for removal of ESCRTs from the endosome surface. Addition of recombinant Vps27 (rVps27p) to membranes in the presence of vps27 mutant membranes resulted in an increase in protection. BBT-CPS protection in the in vitro assay with wild-type cytosol was suppressed at 4 °C and inhibited by addition of apyrase, indicating a physiological and ATP-dependent reaction, but not by addition of a nonhydrolyzable GTP analog, GTP␥s, indicating GTP-hydrolysis independence. The residual internalization reactions in the presence of apyrase and BSA (Figs. 1 and 2) might be a reflection of one round of transport taking place without ATP or without cytosol, due to bound and active pools of ESCRTs on the isolated endosomes (Fig. 5) (6). CPS normally is ubiquitylated at K8 on its short cytosolic N-terminal tail. Mutation of this residue to arginine abrogates ubiquitylation and sorting into internal vesicles, indicating that ubiquitylation is critical for cargo to be recognized by Vps27p and to be sorted properly (3). We confirmed this observation in vitro, as the K8R mutant is less active in the internalization reaction. Clearly, more work is essential to validate the role of the ESCRT proteins in the internalization reaction we report here. The availability of recombinant proteins and complexes that may represent the full pathway should allow reconstitution of each stage of the MVB sorting reaction (cargo sorting, membrane deformation, and

MVB vesicle fission) and permit detailed mechanistic analysis. Our MVB reconstitution assay and the assay developed by the Gruenberg lab complement each other. Although the Gruenberg lab demonstrated EGFR sorting, the group focused primarily on the ability of a fluid phase marker to be engulfed by internal vesicles and on internal vesicle formation. By tracking a membrane protein, our assay follows the fate of the cargo from recognition to incorporation into an internalized vesicle. The use of two different approaches, one in which internal vesicle formation is observed by fluorescent dyes, and one in which membrane cargoes are monitored, can facilitate further understanding of the connection between sorting and vesicle formation. Materials and Methods Strains and Plasmids. JTY1, derived from RPY2 (8), [Mat␣ vps27–123 (ts) doa4⌬::KanMx4 pep4::NatMx4 pUb-pRS425 leu2–3,112 ura 3,52 his4 –519 gal2] was the source of the membranes used in the reactions, except where noted. DOA4 and PEP4 were deleted with KanMx4 and NatMx4 cassettes, respectively (12, 13). Ubiquitin was supplied to wild-type levels in the doa4⌬ background with plasmid Ub-pRS425, created by PCR amplification of Ub and the CUP1 promoter from plasmid YEP96 (14). JTY1 was transformed with the BBT WT CPS or the BBT K8R CPS plasmids. DDY1810 (MATa gal2 leu2⌬ ura3–52 trp1⌬ prb1–1122 pep4 –3 pre1– 451) from the lab of David Drubin was the source of wild-type (WT) cytosol. VPS27 and VPS4 were deleted in DDY1810 with NatMx4 and KanMx4 cassettes, respectively, as sources for the corresponding ESCRT deletion cytosols. JTY2 (DDY1810 doa4⌬::KanMx4 vps27::NatMx4 pUb-pRS425) was the source for vps27⌬ membranes. Membrane Preparation. vps27ts strains were grown at 23 °C in synthetic medium supplemented with required amino acids for plasmid maintenance and transferred to 37 °C at mid-log phase for 2.5 h to accumulate CPS at the endosome surface. JTY2 was grown similarly except the shift to 37 °C was omitted. Cells were converted to spheroplasts as described (15) with the following exceptions. All incubations were performed at 37 °C. All buffers omitted 10 mM NaN3 and NaF to avoid inhibiting energy-dependent reactions. Spheroplast buffer contained 50 mM HEPES, pH 7.4, 1.2 M sorbitol, 1⫻ YPD, and 1 mM DTT and spheroplasts were stored in membrane buffer [25 mM HEPES, pH 6.8, 75 mM KCl, 5 mM Mg(OAc)2, 1 mM DTT, and protease inhibitors] at -80 °C. An equal volume of membrane buffer was added to spheroplasts, followed by homogenizing five times using a Dounce Homogenizer (Bellco Biotechnology). The lysate was centrifuged for 500 ⫻ g for 5 min, 5,000 ⫻ g for 5 min and then at 12,000 ⫻ g for 10 min. The membrane pellet was resuspended in membrane buffer at ⬇4,000 OD600 cell equivalents/mL, corresponding to 20 –30 OD280 units/mL when diluted in 2% SDS for quantitation, and stored as aliquots at -80 °C. For high-salt washes, 500 mM KCl was added just before the last centrifugation step.

15 min, and 100,000 g for 1 h to sediment membranes. The supernatant aliquots were stored at ⫺80°C. Protein concentration measured by Bradford assay (BioRad Protein Assay kit) ranged from 10 –20 mg/ml. Cargo. The chimeric CPS cargo, BBT-CPS, contained two biotin acceptor peptides (BAPs) that were fused to the N terminus of CPS. The sequence of each BAP, GLNDIFEAQKHIEW was optimized for E. coli biotin ligase, BirA (7). The CPS1 gene was amplified from plasmid pGO45 (17) using the following primers, with tandem BAPs and a trypsin cleavage site (RKKRK) included on the forward primer: 5⬘-GATCTCTAGAATGGGTTTGAATGATATTTTTGAAGCTCAAAAAATTGAATGGCATGAAGGTGGT (the underlined stretch duplicated) AGAAAAAAAAGAAAA (the italicized stretch corresponds to the trypsin cleavage site) GGTGGTATCGCCTTACCAGT-3⬘ and 5⬘-CATGGGATCCAGCGTATTCGTTAACATTAACGAT-3⬘. Following PCR amplification, BBT-CPS was inserted into vector pMet25-HA between the Met-25 promoter and a 3⫻HA tag to generate plasmid pMet25-BBT-CPS-HA. The BB-CPS construct used for the PK digest in Fig. 1D was created similarly, except with no trypsin cleavage site. Reconstitution of Multivesicular Body Formation. The reconstitution was performed in three-steps, referred to as standard reaction conditions in the text. Step 1. Biotinylation. E. coli BirA at 7 ␮g was incubated with 0.03 OD280 units of vps27ts cell membranes in 40 ␮L at 23 °C for 20 min in membrane buffer with 1 mM ATP. Membrane amounts from other strains were adjusted to give equivalent HA immunoblot signals. BirA was titrated to optimize biotinylation, and all reactions were pooled in one tube to ensure the same level of biotinylation for subsequent internalization reactions. Membranes were centrifuged at 12,000 ⫻ g for 12 min to remove the BirA, resuspended in 75 ␮L membrane buffer per reaction, and aliquoted for each step 2 condition. Step 2. Internalization Reaction. Biotinylated CPS in the endosomal membrane fraction was incubated in membrane buffer with cytosol from DDY1810 strains (WT or ESCRT mutant) or recombinant protein and an ATP regeneration system (rATP) (GTP was used only in the early studies using BB-CPS). Reactions in 100 ␮L volume per internalization condition were incubated at 30 °C for 30 min unless indicated otherwise. Following the reactions, the membranes were centrifuged for 12 min at 12,000 ⫻ g to remove cytosolic and soluble components, and resuspended in membrane buffer [with 1 mM Mg(OAc)2] per reaction and aliquoted for each replicate in step 3.

Cytosol Preparation. Cytosol preparation was performed as described with the following modifications (16). Strains were grown at 30°C in YPD, harvested at 2.0 OD600/ml, resuspended in membrane buffer and ground in a mortar using liquid nitrogen. The lysate was centrifuged at 3,000 g for 10 min, 20,000 g for

Step 3. Protease Protection Reaction. To remove any biotin signal on remaining surface-trapped CPS, 0.2 mg/mL trypsin or PK was incubated with membrane fractions from step 2 for 30 min at 4 °C in 30 ␮L reaction volume. Trypsin cleavage reactions were terminated with soybean trypsin inhibitor (0.4 mg/mL), incubated for 5 min at 4 °C, and then sample buffer was added (60 ␮L total). PK cleavage reactions were terminated with 60 ␮L 95 °C SDS sample buffer (90 ␮L total) and incubation at ⬎95 °C for 5 min. Aliquots of 5–10 ␮L were loaded onto 8% polyacryamide gels for SDS-PAGE and detected by incubation of PVDF or nitrocellulose membranes with antibodies against biotin (US Biological) or HA (Covance). Quantification was performed by addition of LI-COR secondary antibodies and scanning using the Odyssey Infrared Imaging System (LI-COR Biosciences). Protein purification is described in SI Materials and Methods.

1. Saksena S, Sun J, Chu T, Emr SD (2007) ESCRTing proteins in the endocytic pathway. Trends Biochem Sci 32:561–573. 2. Spormann DO, Heim J, Wolf DH (1992) Biogenesis of the yeast vacuole (lysosome). The precursor forms of the soluble hydrolase carboxypeptidase yscS are associated with the vacuolar membrane. J Biol Chem 267:8021– 8029. 3. Katzmann DJ, Babst M, Emr SD (2001) Ubiquitin-dependent sorting into the multivesicular body pathway requires the function of a conserved endosomal protein sorting complex, ESCRT-I. Cell 106:145–155. 4. Raymond CK, Howald-Stevenson I, Vater CA, Stevens TH (1992) Morphological classification of the yeast vacuolar protein sorting mutants: Evidence for a prevacuolar compartment in class E vps mutants. Mol Biol Cell 3:1389 –1402. 5. Falguieres T, et al. (2008) In vitro budding of intralumenal vesicles into late endosomes is regulated by Alix and Tsg101. Mol Biol Cell 19:4942– 4955. 6. Wollert T, Wunder C, Lippincott-Schwartz J, Hurley JH (2009) Membrane scission by the ESCRT-III complex. Nature 458:172–177. 7. Beckett D, Kovaleva E, Schatz PJ (1999) A minimal peptide substrate in biotin holoenzyme synthetase-catalyzed biotinylation. Protein Sci 8:921–929. 8. Piper RC, Cooper AA, Yang H, Stevens TH (1995) VPS27 controls vacuolar and endocytic traffic through a prevacuolar compartment in Saccharomyces cerevisiae. J Cell Biol 131:603– 617.

9. Baker D, Hicke L, Rexach M, Schleyer M, Schekman R (1988) Reconstitution of SEC gene product-dependent intercompartmental protein transport. Cell 54:335–344. 10. Babst M, Wendland B, Estepa EJ, Emr SD (1998) The Vps4p AAA ATPase regulates membrane association of a Vps protein complex required for normal endosome function. EMBO J 17:2982–2993. 11. Hanson PI, Roth R, Lin Y, Heuser JE (2008) Plasma membrane deformation by circular arrays of ESCRT-III protein filaments. J Cell Biol 180:389 – 402. 12. Brachmann CB, et al. (1998) Designer deletion strains derived from Saccharomyces cerevisiae S288C: A useful set of strains and plasmids for PCR-mediated gene disruption and other applications. Yeast 14:115–132. 13. Goldstein AL, McCusker JH (1999) Three new dominant drug resistance cassettes for gene disruption in Saccharomyces cerevisiae. Yeast 15:1541–1553. 14. Swaminathan S, Amerik AY, Hochstrasser M (1999) The Doa4 deubiquitinating enzyme is required for ubiquitin homeostasis in yeast. Mol Biol Cell 10:2583–2594. 15. Sipos G, Fuller RS (2002) Separation of Golgi and endosomal compartments. Methods Enzymol 351:351–365. 16. Shimoni Y, Schekman R (2002) Vesicle budding from endoplasmic reticulum. Methods Enzymol 351:258 –278. 17. Odorizzi G, Babst M, Emr SD (1998) Fab1p PtdIns(3)P 5-kinase function essential for protein sorting in the multivesicular body. Cell 95:847– 858.

17400 兩 www.pnas.org兾cgi兾doi兾10.1073兾pnas.0909473106

Tran et al.