Distinct mechanisms enable inward or outward budding from late endosomes/multivesicular bodies

Abstract

Regulating the residence time of membrane proteins on the cell surface can modify their response to extracellular cues and allow for cellular adaptation in response to changing environmental conditions. The fate of membrane proteins that are internalized from the plasma membrane and arrive at the limiting membrane of the late endosome/multivesicular body (MVB) is dictated by whether they remain on the limiting membrane, bud into internal MVB vesicles, or bud outwardly from the membrane. The molecular details underlying the disposition of membrane proteins that transit this pathway and the mechanisms regulating these trafficking events are unclear. We established a cell-free system that reconstitutes budding of membrane protein cargo into internal MVB vesicles and onto vesicles that bud outwardly from the MVB membrane. Both budding reactions are cytosol-dependent and supported by Saccharomyces cerevisiae (yeast) cytosol. We observed that inward and outward budding from the MVB membrane are mechanistically distinct but may be linked, such that inhibition of inward budding triggers a re-routing of cargo from inward to outward budding vesicles, without affecting the number of vesicles that bud outwardly from MVBs.

Introduction

The movement of cell surface membrane proteins from the plasma membrane into the endocytic pathway enables regulation of cellular responses to environmental cues, homeostatic maintenance of intracellular compartments, and tuning the activity of signaling molecules (Maxfield and McGraw, 2004). The late endosome/multivesicular body (MVB) is thought to be the final site for sorting of membrane protein cargo in the endocytic pathway prior to fusion of the MVB with the lysosome and cargo degradation. Membrane proteins that bud inwardly from the limiting membrane into internal vesicles of MVBs can be degraded upon MVB-lysosome fusion, while membrane proteins that bud outwardly from endosomes prior to this step are transported to other cellular compartments and escape degradation (Grant and Donaldson, 2009, Gruenberg, 2001, Maxfield and McGraw, 2004, Thompson et al., 2007, Weigert et al., 2004).

There are multiple routes that allow membrane proteins to escape the cannonical pathway to lysosomal degradation (Fig. 1A) (Gruenberg, 2001, Tanowitz and von Zastrow, 2003, Grant and Donaldson, 2009, Maxfield and McGraw, 2004, Thompson et al., 2007, Weigert et al., 2004). Early in the endocytic pathway, cargo can bud from early endosomes in a Rab4/ Rab5-dependent manner for movement back to the plasma membrane (van der Sluijs et al., 1992, Sonnichsen et al., 2000, McCaffrey et al., 2001), or take a Rab 11-dependent route to the plasma membrane (Grant and Donaldson, 2009, Maxfield and McGraw, 2004, Thompson et al., 2007, Weigert et al., 2004). Alternatively, cargo (e.g. mannose-6 phosphate receptor) may bud from endosomes in a Rab9 and dynamin-dependent manner for movement to the trans-Golgi network (TGN) (Lombardi et al., 1993, Nicoziani et al., 2000, Riederer et al., 1994, Robinson, 1994) or in a retromer-dependent manner from endosomes to the TGN (Arighi et al., 2004). Membrane proteins that reach the MVB can also remain on the limiting MVB membrane and become incorporated into the lysosomal membrane upon MVB-lysosome fusion (Eden et al., 2012, Felder et al., 1990, Futter et al., 1996, Katzmann et al., 2002, Piper and Luzio, 2001, Raymond et al., 1992, Hurley and Hanson, 2010, Babst, 2011, Adell and Teis, 2011, Baumgart et al., 2007). Interestingly, some membrane proteins (e.g. epidermal growth factor receptor, major histocompatibility complex class II, and tetraspanins) can be recycled from the MVB to the plasma membrane (Felder et al., 1990, Rocha and Neefjes, 2008, Cho et al., 2015), however the mechanisms underlying potential outward budding events from the MVB to the plasma membrane are unknown.

Figure 1: Reconstitution of EGFR protein sorting from endosomal membranes using cytosol isolated from HeLa cells, Saccharomyces cerevisiae, and Drosophila melanogaster.

A) Removal of membrane proteins by endocytosis follows a canonical itinerary in which cargo passes through morphologically and biochemically defined organelles. After internalization and transport from early to late endosomes, membrane proteins can be internalized into the lumen of late endosomes/multivesicular bodies (MVB). Receptors that remain on the limiting membrane of the MVB may bud out from the limiting membrane, or can be sorted into internal MVB vesicles that are degraded upon MVB-lysosome fusion. The sorting of proteins into the internal vesicles of the MVB is known to require recognition by the Endosomal Sorting Complexes Required for Transport (ESCRTs). B) Serum-starved HeLa cells can be stimulated to induce internalization of a receptor from the plasma membrane, resulting in movement of ligand-receptor complex into endosomes (Sun et al., 2010, Sirisaengtaksin et al., 2014). Isolation of partially purified endosomes that contain the receptor can be detected by immunoblotting using an intracellular epitope-specific antibody (1). Incubation of these endosomes with trypsin removes the C-terminal epitope of the receptor that protrudes from the plasma membrane, resulting in a loss of signal for that epitope on an immunoblot (2). Incubation of endosomes with ATP and cytosol (isolated from HeLa cells, Saccharomyces cerevisiae, or Drosophila melanogaster), at 37°C results in formation of internal vesicles (3) (Sun et al., 2010) and protection of the C-terminal EGFR epitope from subsequent trypsin cleavage. Incubation of endosomes with ATP and cytosol for 3 hours at 37°C [as in (3)] followed by centrifugation results in separation of MVBs (in pellet) and outwardly budded vesicles (in supernatant). The MVB pellet (4) is subsequently digested with trypsin while supernatant (5) is further centrifuged to concentrate budded vesicles for collection. (C-E) Endosomal membranes (5 μL, Lane 1) and endosomal membranes (5 μL) digested with trypsin to remove the C-terminal epitope of the receptor (Lane 2), as well as partially purified HeLa endosomal membranes (15 μL) that had been incubated in reactions containing ATP and cytosol derived from: HeLa cells (25 μg, C), Saccharomyces cerevisiae (70 μg, D), or Drosophila melanogaster (25 μg, E), at 37°C prior to trypsin treatment (Lane 3) are shown. Data represents the mean +/− S.E.M. (n=3) normalized to membrane control.

Membrane protein disposition in the late endocytic pathway has been elucidated using biochemical, genetic, and cell-free model systems. Studies using the budding yeast, Saccharomyces cerevisiae, allowed isolation of proteins required for the inward budding of vesicles that enable movement of membrane proteins into the vacuole, the yeast degradative organelle (Tran et al., 2009, Hurley and Emr, 2006, Katzmann et al., 2001, Bowers et al., 2004, Schmidt and Teis, 2012, Babst et al., 2002a, Babst, 2011, Bilodeau et al., 2003, Piper and Luzio, 2001, Teis et al., 2010, Raymond et al., 1992). Characterization of vacuolar protein sorting (Vps) genes resulted in identification of the Endosomal Sorting Complexes Required for Transport (ESCRTs) (Piper et al., 1995, Katzmann et al., 2003, Malerod et al., Babst et al., 2002a, Babst et al., 2002b, Babst et al., 2011, Bilodeau et al., 2003, Piper and Luzio, 2001, Teis et al., 2010, Hurley and Emr, 2006, Kostelansky et al., 2006, Katzmann et al., 2001, Bowers et al., 2004, Lemmon and Traub, 2000). The four ESCRT complexes (ESCRT-0, -I, -II, and -III) are recruited to endosomal membranes and enable the sorting and movement of ubiquitinated membrane proteins into internal vesicles (Lemmon and Traub, Futter et al., 1996, Hurley and Emr, 2006, Katzmann et al., 2001, Wemmer et al., 2011, Raiborg and Stenmark, 2009). While advances have been made in understanding the mechanisms of inward budding at MVBs (Sun et al., 2010, Falguieres et al., 2008, Tran et al., 2009), we lack a complete understanding of mechanisms that underlie regulation of ESCRT function, membrane protein movement, and vesicle formation/budding from the limiting MVB membrane.

The purpose of this study was to identify the molecular machinery that may regulate outward budding from the MVB and determine whether the machinery is distinct from the inward budding machinery. We refined a cell-free assay that assesses the inward budding of cargo into internal MVB vesicles (Sirisaengtaksin et al., 2014, Sun et al., 2010) to enable measurement of outwardly budding vesicles from the same donor membranes. Both inward and outward budding events were found to require cytosolic components and are supported by cytosol isolated from Saccharomyces cerevisiae (yeast), allowing genetic manipulation and screening for molecules that influence these budding events (e.g. ESCRTs). We observed that dynamin regulates outward budding from MVBs and that inhibition of ESCRT-0-dependent inward budding triggers a re-routing of cargo into outward budding vesicles, suggesting that while inward and outward budding from endosomal limiting membranes require distinct molecules, these processes may be linked.

Materials and Methods:

Materials—

Antibodies were purchased from the following commercial sources: EGFR C-terminal (Invitrogen; Cat # PA1–1110; 1:200), EGFR N-terminal (E234; Abcam; Cat # ab32198; 1:1000), V5-tag (Invitrogen; Cat # 46–0705; 1:5000), c-Myc (9E10, Santa Cruz Biotechnologies; Cat # sc-50; 1:500), EEA1 (Thermo Fisher; Cat # PA5–17228; 1:500), LAMP1 (H4A3; Developmental Studies Hybridoma Bank; Cat # H4A3; 1:500), Rab11 (Millipore; Cat # 2413S; 1:1000), Rab 7 (Invitrogen; Cat #D95F2; 1:1000), Transferrin Receptor (TfR, Abcam; Cat # ab190640; 1:1000).

Constructs—

The pCMV-AT1R-Myc construct was kindly provided by Dr. Guangwei Du (UTHealth). The pcDNA3.1-hisB-V5-R4-FGFR4_Gly388 construct was kindly provided by Dr. Michael Ittmann (Baylor College of Medicine). The PcDNA6a myc-tagged EGFR_K721A construct was kindly provided by Dr. Mien-Chie Hung (M.D. Anderson). The PCMV-intron myc Rab11 S25N construct and the pcDNA3.1 control vector were purchased from Addgene.

Cell Culture—

HeLa cells (ATCC) were cultured as a monolayer in 10-cm plastic plates in Dulbecco’s Modified Eagle Medium (DMEM, Mediatech) containing 10% Fetal Bovine Serum (FBS, Sigma) under 5% CO₂ at 37°C. Before each experiment, cells were passaged by removing them from the plate using 0.25% trypsin/ethylenediaminetetraacetic acid (EDTA) and seeded onto 10-cm tissue culture plates.

Recombinant proteins—

Hrs and STAM were produced in insect cells as previously described (Tsujimoto et al., 1999, Sirisaengtaksin et al., 2014). Recombinantly produced Dynamin 1 was kindly provided by Dr. Sandra L. Schmid (UT Southwestern, Dallas, TX).

Cytosol preparation—

Mammalian:

HeLa cells were placed on ice, washed with ice-cold PBS (2x with 5 mL), scraped from the plate, and centrifuged (2000 × g for 15 min) at 4 °C. The cell pellet was resuspended in 100 μL of homogenization buffer (HB) (20 mM HEPES pH 7.4, 0.25 M sucrose, 2 mM EGTA, 2 mM EDTA, and 0.1 mM DTT) containing a protease inhibitor cocktail (112 μM PMSF, 3 μM aprotinin, 112 μM leupeptin, 17 μM pepstatin). Cells were sonicated 5 times (5 pulses of 1 second at output control 2) (Branson Sonifier 250, VWR Scientific). The lysate was centrifuged (2000 × g for 10 min) at 4°C, and the supernatant was further centrifuged (100,000 × g for 1 hour) at 4°C. The supernatant was collected and protein concentration was calculated using a Bradford assay (Sirisaengtaksin et al., 2014).

S. cerevisiae:

S. cerevisiae strains were plated on YPD plates (500 mL ddH₂0 containing: 10 g bactopeptone, 5 g yeast extract, 8 g agar, 25 mL 40% dextrose) and incubated for 48 hours at 30°C. YPD media (5 mL) was inoculated with various S. cerevisiae strains and incubated overnight on a shaker at 30 °C. Cultures were transferred into a secondary culture of YPD media (50 mL) and were grown until OD₆₀₀ reached 0.8–1.0. Cells were collected (3000 × g for 3 min) and washed twice, first with 500 μL of ddH₂O followed by 500 μL TP buffer (20 mM Tris, pH 7.9; 0.5 mM EDTA; 10% glycerol; 50 mM NaCl, 112 μM leupeptin, 3 μM aproptinin, 112 μM PMSF, and 17 μM pepstatin). The cells were collected (3000 × g for 3 min) and resuspended into 130 μL of TP buffer. Acid-washed beads (50 μL) were added to the cells and the cells were lysed (1 min vortex-1 min incubation on ice, 5X). Cells were centrifuged (3000 × g for 10 min) and the supernatant was collected. Protein concentration was calculated using a Bradford assay. The supernatant was divided into 70 μg aliquots and stored at −80 °C. For S. cerevisiae deletion strains (Table 1), we inoculated strains in YPD media containing G418 (500 μg/mL).

Table 1:

Saccharomyces cerevisiae strains used in this study.

Name	Genotype	Source or Reference	Effect on Inward Budding	Effect on Outward Budding
vps27Δ	MATa his3Δ1 leu2Δ0 ura3Δ0 met15Δ0 vps27::KANMX6	Dharmacon	Inhibits inward budding by 40+/−3% (Fig. 2b)	Increases outward budding by 47+/−12% (Fig. S6a)
hse1Δ	MATa his3Δ1 leu2Δ0 ura3Δ0 met15Δ0 hse::KANMX6	Dharmacon	Inhibits inward budding by 83+/−1% (Fig. 2b)	Increases outward budding by 38+/−3% (Fig. 4a)
vps23Δ	MATa his3Δ1 leu2Δ0 ura3Δ0 lys2Δ0 vps23::KANMX6	Dharmacon	Inhibits inward budding by 56+−17% (Fig. 2d)	Did not test
vps28Δ	MATa his3Δ1 leu2Δ0 ura3Δ0 lys2Δ0 vps28::KANMX6	Dharmacon	Inhibits inward budding by 56+/−12% (Fig. 2d)	Did not test
vps37Δ	MATa his3Δ1 leu2Δ0 ura3Δ0 met15Δ0 vps37::KANMX6	Dharmacon	Inhibits inward budding by 87+/−9% (Fig. 2d)	Did not test
mvb12Δ	MATa his3Δ1 leu2Δ0 ura3Δ0 met15Δ0 mvb12::KANMX6	Dharmacon	Inhibits inward budding by 89+/−6% (Fig. 2d)	No significant difference (Fig. 4a)
vps36Δ	MATa his3Δ1 leu2Δ0 ura3Δ0 lys2Δ0 vps36::KANMX6	Dharmacon	Inhibits inward budding by 52+/9% (Fig. 2e)	No significant difference (Fig. 4a)
snf8Δ	MATa his3Δ1 leu2Δ0 ura3Δ0 lys2Δ0 snf8::KANMX6	Dharmacon	Inhibits inward budding by 47+/−7% (Fig. 2e)	Did not test
vps25Δ	MATa his3Δ1 leu2Δ0 ura3Δ0 lys2Δ0 vps25::KANMX6	Dharmacon	Inhibits inward budding by 46+/−13% (Fig. 2e)	Did not test
snf7Δ	MATa his3Δ1 leu2Δ0 ura3Δ0 met15Δ0 snf7::KANMX6	Dharmacon	Inhibits inward budding by 44+/−12% (Fig. 2f)	Did not test
did4Δ	MATa his3Δ1 leu2Δ0 ura3Δ0 met15Δ0 did4::KANMX6	Dharmacon	No significant difference (Fig. 2f)	Did not test
vps20Δ	MATa his3Δ1 leu2Δ0 ura3Δ0 lys2Δ0 vps20::KANMX6	Dharmacon	No significant difference (Fig. 2f)	Did not test
vps24Δ	MATa his3Δ1 leu2Δ0 ura3Δ0 lys2Δ0 vps24::KANMX6	Dharmacon	Inhibits inward budding by 51+/−12% (Fig. 2f)	No significant difference (Fig. 4a)
vps1Δ	MATa his3Δ1 leu2Δ0 ura3Δ0 lys2Δ0 vps1::KANMX6	Dharmacon	No significant difference (Fig. 4d)	Inhibits outward budding by 63+/−2% (Fig. 4c)
yMG1	MATa his3Δ1 leu2Δ0 ura3Δ0 met15Δ0 snf7::HPHMX6	This study	Did not test	Did not test
yMG2	MATa his3Δ1 leu2Δ0 ura3Δ0 met15Δ0 did4::HPHMX6	This study	Did not test	Did not test
yMG3	MATa/MATα his3Δ1/ his3Δ1 leu2Δ0/leu2Δ0 ura3Δ0/ura3Δ0 MET15/met15Δ0 LYS2/lys2Δ0 SNF7/snf7::HPHMX6 VPS20/vps20::KANMX6	This study	Inhibits inward budding by 47+/−5% (Fig. 2f)	Did not test
yMG4	MATa/MATα his3Δ1/ his3Δ1 leu2Δ0/leu2Δ0 ura3Δ0/ura3Δ0 MET15/met15Δ0 LYS2/lys2Δ0 DID4/did4::HPHMX6 VPS24/vps24::KANMX6	This study	Inhibits inward budding by 55+/−1% (Fig. 2f)	Did not test

Drosophila melanogaster:

Frozen whole head homogenates of approximately 1000 fly heads were centrifuged (100,000 × g for 60 min) to pellet total membranes and the supernatant was collected. Protein concentration was calculated using a Bradford assay. Supernatants were stored in 25 μg aliquots at −80°C.

Cell Transfection—

Plasmid DNA was prepared (Qiagen), and HeLa cells were transiently transfected using Lipofectamine 2000 transfection reagent according to the manufacturer’s protocol. The constructs used in each transfection are as indicated. Briefly, cells were plated in 6-well plates and grown until they reached 80–90% confluence. In each well of the plate, DNA (3 μg) was added to Opti-MEM reduced serum medium (250 μL), mixed, and incubated for 5 min at room temperature. Lipofectamine 2000 reagent (3.5 μL) was added to Opti-MEM reduced serum medium (250 μL), mixed, and incubated for 5 min at room temperature. The tubes were combined, mixed gently, and incubated for 20 min at room temperature before adding 500 μL to each well. After 48 hours, the cells were used in the cell-free sorting assay described below.

Cell-free reconstitution of inward budding from MVB membranes—

The reconstitution of inward budding was performed as described (Sun et al., 2010, Sirisaengtaksin et al., 2014, Gireud et al., 2015). In experiments where EGFR was the membrane protein cargo, HeLa cells were grown to 75–80% confluence. Before harvesting, cells were serum starved (2 hours at 37°C) and stimulated with EGF (100 ng/mL; 10 min at 37°C or 2 ng/mL; 10 min at 37°C). Endosomal membranes were isolated as previously described (Sun et al., 2010, Sirisaengtaksin et al., 2014) and resuspended in HB buffer (volume dependent on number of reactions; 5 μL for control reactions and 15 μL per experimental reaction), and used for reconstitution reactions. Briefly, after stimulation, cells were collected, lysed through a 30-gauge needle, and serially centrifuged to obtain endosomal membranes.

Endosomal membranes (starting material, 5 μL) were either incubated on ice or were trypsin-treated (6 μL of 0.27 μg/μL trypsin; 4°C for 30 minutes). For reactions containing mammalian cytosol, a standard reaction (50 μL) contained 15 μL endosomal membranes, 6 μL ATP regeneration system (2 mM MgATP, 50 μg/mL creatine kinase, 8 mM phosphocreatine and 1 mM DTT of final concentrations), 25 μg of Hela cytosol and HB to a total reaction volume of 50 μL. For the Saccharomyces cerevisiae cytosol reactions, a standard reaction (50 μL) contained 15 μL membranes, 6 μL ATP regeneration system, 70 μg of S. cerevisiae cytosol and HB to a total reaction volume of 50 μL. For the Drosophila melanogaster cytosol reactions, a standard reaction (50 μL) contained 15 μL membranes, 6 μL ATP regeneration system, 25 μg of Drosophila melanogaster (fly) cytosol, and HB to a total reaction volume of 50 μL.

All experimental reactions were incubated for 3 hours at 37°C, followed by trypsin-treatment (6 μl of 0.27 μg/μL trypsin; 30 min at 4°C). Experimental reactions were centrifuged (20,000 × g; 30 min at 4°C) while control reactions remained on ice. Control reactions were resuspended in sample buffer for SDS-PAGE. For experimental reactions, supernatant was aspirated and pellet was resuspended in sample buffer for biochemical examination by SDS-PAGE. Resultant blots were probed with an antibody that recognizes amino acids1190–1210 (C-terminal epitope) of EGFR (1:200 dilution in 5% nonfat milk with PBS, overnight at 4 °C). To examine the dependence of cargo sorting on the presence of ESCRT proteins, yeast cytosol was prepared from strains listed in Table 1 and used in place of parental yeast cytosol. All reactions were normalized to parental controls. For the FGFR4 experiments, transfected cells were serum-starved in media containing cycloheximide (30 μg/mL) for 2 hours and stimulated with bFGF (50 ng/mL) for 5 hours. Following the bFGF stimulation, subsequent experimental conditions were as described for EGFR. The resulting blots were probed with an antibody that recognized the intracellular V5-tag that was fused to the COOH-terminus of the FGFR4 clone used in these studies (1:5000 dilution in 5% nonfat milk with PBS-Tween, overnight at 4°C). For the AT1R experiments, transfected cells were starved in serum-free media (2 hours at 37°C) and stimulated with angiotensin II (1 mg/mL) (30 minutes at 37°C). Following the angiotensin II stimulation, subsequent experimental conditions were as described for EGFR. The blots were probed with an antibody that recognized the intracellular Myc-tag that was fused to the COOH-terminus of the AT1R clone used in these studies (1:500 dilution in 5% nonfat milk with PBS-Tween, overnight at 4°C). For the EGFR_K721A experiments, transfected cells were starved in serum-free media (2 hours at 37°C) and stimulated with EGF (100 ng/mL) (10 minutes at 37°C). Experimental conditions were then performed as described for EGFR experiments. The blots were probed with an antibody that recognized the intracellular Myc-tag that was fused to the COOH-terminus of the EGFR_K721A clone used in these studies (1:500 dilution in 5% nonfat milk with PBS-Tween, overnight at 4°C).

Isolation of budded vesicles from cell-free reactions—

Experimental conditions were performed as in EGFR inward budding experiments described above, including 70 μg of Saccharomyces cerevisiae, ATP, and the 3-hour reaction incubation. Following the 3-hour incubation, experimental reactions were centrifuged (20,000 × g for 30 min at 4°C). The supernatant was collected and further centrifuged (150,000 × g for 1 hour at 4°C). After ultracentrifugation, the resulting pellet was either resuspended in sample buffer for SDS-PAGE, or subjected to Nanosight Tracking Analysis. For the TfR experiments, cells expressing PcDNA3.1 empty vector or Rab11S25N were starved in serum-free media (2 hours at 37°C) and stimulated with holo-Transferrin (Sigma, 200 μg/mL) (20 minutes at 37°C). Following the transferrin stimulation, subsequent experimental conditions were as described for EGFR. The blots were probed with an antibody that recognized the intracellular epitope of the TfR (1:1000 dilution in 5% nonfat milk with PBS-Tween, overnight at 4°C).

Critera for Calculating Reaction Efficiency in Cell-Free Sorting Assay—

All experimental reactions were normalized to starting endosomal controls to obtain the reaction efficiency. To obtain reaction efficiency, the amount of EGFR on starting membranes added to each reaction was compared to the amount of EGFR in reactions that contained membranes, cytosol, ATP and had been subsequently cleaved by trypsin. Typical reaction efficiencies were 25–50% and are shown in the figures and/or their respective figure legends.

NanoSight Tracking Analysis (McCaffrey et al., 2001)—

NTA measurements were performed on membranes isolated from cell-free reactions in which the reaction supernatant had been centrifuged to isolate outwardly budded vesicles using a NanoSight NS300 instrument following the manufacturer’s instructions. NTA is performed by measuring the rate of Brownian motion of particles in a low volume light scattering system (NanoSight Ltd., Amesbury, United Kingdom). Results are presented as mean size of vesicles (x-axis) and concentration of particles per mL of solution (y-axis). Samples were examined in triplicate.

Electron Microscopy (EM)—

Vesicle size was visualized using TEM on membranes isolated from supernatant obtained from MVB sorting reactions. 5 μL of vesicles were placed on glow discharged carbon formvar grids (TedPella) for approximately 5 min. Grids were rinsed 3x with 5 μL of water, using blotting paper to wick away excess liquid between rinses. Finally, grids were rinsed quickly with 50% mixture of NanoW (Nanoprobes) stain, wicked, and then stained for approximately 30 min before wicking excess liquid and allowing grids to dry for at least 30 min prior to imaging. Micrographs were collected on a JOEL 1400 electron microscope operated at 120 kV. Images were obtained using a Gatan ultrascan camera at 120kx magnifications at the image plane.

OptiPrep Gradient—

After HeLa cells were serum-starved and stimulated (100 ng/mL EGF; 20 min at 37°C), post-nuclear supernatant was isolated and loaded on top of a continuous Opti-prep gradient (Sigma, 10–20%) and centrifuged (150,000 × g for 10 hours at 4°C) in a swinging bucket rotor (TLS 55, Beckman). Fractions (200 μL) were collected and diluted in of HB (200 μL) followed by centrifugation (150,000 × g for 1 hour at 4°C). The resulting pellet was resuspended in sample buffer for biochemical examination and examined via SDS-page followed by staining with coomassie to examine protein distribution across the gradient or by immunoblotting. If the fractions were to be used in the cell-free assay, fractions 3 & 4 containing late endosomal membranes were collected and fractions 8 & 9 containing early endosomal membranes were collected and diluted in HB (400 uL) followed by centrifugation (150,000 × g for 1 hour at 4°C). The resulting pellets were resuspended in HB and membranes were used in cell-free reactions or for cryo-electron microscopy.

Cryo-Electron Microscopy and Endosomal Length Measurements—

Endosomal fractions 3 & 4 (containing late endosomal membranes) and endosomal fractions 8 & 9 (containing early endosomal membranes) were isolated as described above and collected. Endosomal Fractions (n = 1 sample/group) were applied to freshly glow-discharged (30 sec) 2/2 Quantifoil on 200 mesh copper grids. After 30 sec, excess buffer was blotted and the sample was immediately plunged into ethane cooled to liquid N2 temperature. Cryo-preserved grids were stored in liquid N2 until use. Cryo-electron microscopy was performed on a FEI Polara G2 equipped with a Gatan K2 Summit direct electron detector. Multiple areas of the grid were chosen at random and 8 × 8 montages were collected at 4700x in low dose/photon counting mode using SerialEM. To quantify the length of endosomes, individual montages were displayed in IMOD and a line along the diameter of each endosome was drawn and stored in a model for each montage. Lengths were extracted for 475 endosomes from each model table, imported into Excel and the data displayed by separating the lengths into 100 nm bins. To remove potential bias, the person collecting the primary data and the person quantifying the length of individual endosomes were both blinded to the sample identities.

Generation of Saccharomyces cerevisiae heterozygous double knock out strains—

Conversion of snf7::KANMX6 and did4::KANMX6 strains:

snf7::KANMX6 (Dharmacon: MATa his3Δ1 leu2Δ0 ura3Δ0 met15Δ0 snf7Δ::KANMX6) and did4::KANMX6 (Dharmacon: MATa his3Δ1 leu2Δ0 ura3Δ0 met15Δ0 did4Δ::KANMX6) strains were transformed as previously described (Gietz and Woods) with SalI-ClaI digested plasmid pAG32 (HPHMX6) (Goldstein and McCusker, 1999). Briefly, cells were grown in 5 mL of YPD and incubated (overnight at 30°C). The next day, cells were re-inoculated in 20 mL of YPD at a starting OD₆₀₀ of 0.2. Once cells had reached a final OD₆₀₀ of 0.8, they were collected (1000 × g for 5 minutes) and washed in 500 μL of 100 mM LiAc. The cell suspension was then collected (1000 × g for 5 minutes) and resuspended in transformation solution (50% PEG, 0.1M LiAc, 10 μg carrier DNA, ddH₂O) and 1 μg of digested pAG32. Cells were incubated for 30 minutes at 30 °C followed by incubation at 42 °C for 20 minutes. Next, cells were collected (3300 × g for 15 seconds), resuspended in 100 μL of ddH₂0 and plated on YPD plates containing 500 μg/mL Hygromycin B. Transformants were selected based on sensitivity to G418 and resistance to Hygromycin B. The resulting strains were yMG1 and yMG2 (see Table 1).

Creation of heterozygous double knock out strains:

To create the snf7/SNF7 vps20/VPS20 and did4/DID4 vps24/VPS24 strains, the yMG1 and yMG2 strains were crossed with the vsp20Δ and vps24Δ strains, respectively. G418^R Hygromycin B^R heterozygous diploid knock out strains were selected for further studies.

Statistical Analysis—

Statistical significance was determined using the Prism 7 software. Prior to analysis, samples were tested for normality using the Shapiro-Wilk test. If the samples followed a normal distribution, then either a Paired t-test or a one-way ANOVA followed by post-hoc analysis (Tukey or Dunnet’s test) were performed. If samples were not normal, then the Kruskal-Wallis analysis was utilized. A p-value of <0.05 was considered statistically significant with an n=>3.

Results:

Saccharomyces cerevisiae (yeast) and Drosophila melanogaster (fly) cytosol support inward budding of the EGFR from mammalian endosomes.

We used a cell-free approach to reconstitute MVB formation and movement of a membrane protein cargo (the Epidermal Growth Factor Receptor, EGFR) from the limiting endosomal membrane into internal vesicles (Fig. 1B) (Sun et al., 2010, Sirisaengtaksin et al., 2014). Both protease protection of an intracellular epitope of the EGFR and the number of internal vesicles (inward budding) increase over time in reactions that are dependent on cytosol, ATP, temperature, and an intact proton gradient (Sun et al., 2010, Sirisaengtaksin et al., 2014). Trypsin digests EGFR that has not been internalized into MVBs and the EGFR, given the presumed topology of the receptor the ectodomain, should remain intact inside endosomes upon trypsin treatment (as shown in Sun et al, 2010). Starting membranes were incubated with or without trypsin (Fig. S1) and the resulting blots were probed with EGFR antibodies directed against either C- or N-terminal EGFR epitopes. The C-terminal (intracellular) epitope is no longer detected upon trypsin treatment (Fig. S1A, lane 2 compared to lane 1). After trypsin treatment, the N-terminal (extracellular) epitope is no longer detected (Fig. S1B, top panel, lane 2 compared to lane 1) and instead a band of approximately 90 kDa corresponding to the ectodomain is detected using the N-terminally directed (extracellular) EGFR antibody (Fig. S1B, bottom panel, lane 2 compared to lane 1), indicating that the ectodomain remains protected and associated with endosomal membranes after trypsin digestion. Donor membranes and cytosol used in these reconstitution reactions had been previously isolated from mammalian cells (Sun et al., 2010, Sirisaengtaksin et al., 2014). To determine whether yeast or fly cytosol could substitute for cytosol obtained from mammalian sources (Fig. 1C-E) (Sun et al., 2010, Sirisaengtaksin et al., 2014), reactions containing mammalian endosomes were incubated with mammalian (Fig. 1C), yeast (Fig. 1D), or fly cytosol (Fig. 1E). The efficiency of the cell-free reactions, as judged by the EGFR protected from protease cleavage after incubation with cytosol, using yeast and fly cytosol was 24±7% and 32±1%, respectively, compared to what we observed with mammalian cytosol, 44±13% (Fig. 1C-E). These data suggest that while quantitative differences exist between the mammalian, yeast, and fly cytosols, all three organisms contain essential proteins required for the inward budding event.

The sorting of EGFR is dependent on ESCRT proteins.

Previous data suggest that ESCRT proteins play a role in cargo selection at the yeast vacuolar membrane (e.g. Bilodeau et al., 2002, Hurley and Hanson, 2010, Schmidt and Teis, 2012) and data in mammalian cells suggests that these proteins perform a function required for lysosomal degradation (Bilodeau et al., 2002, Katzmann et al., 2001, Sirisaengtaksin et al., 2014, Sun et al., 2010). The regulation of inward membrane budding and ESCRT protein function that enables movement of membrane proteins from the limiting endosomal membrane into internal vesicles has not been clarified. Inward budding is dependent on cytosolic components (Fig. 2A), therefore to examine whether ESCRT proteins regulate inward cargo movement in our reconstituted system, cytosol derived from yeast strains deleted of ESCRT proteins (Table 1) was used in cell-free reactions in place of wild-type (parental) cytosol. Reactions containing cytosol derived from ESCRT-0 deficient yeast strains (vps27Δ, hse1Δ) significantly decreased EGFR epitope protease protection (Fig. 2B) compared to cytosol isolated from parental strains. If recombinant mammalian ESCRT-0 homologs (Hrs and STAM, 180nM) were added into reactions that have cytosol isolated from vps27Δ or hse1Δ strains, the inhibition of inward budding was significantly rescued (Fig. 2C). Control reactions containing recombinant mammalian ESCRT-0 homologs (Hrs and STAM, 180nM) had no effect on inward EGFR budding (Fig. 2C). Thus, deletion of individual ESCRT-0 components decreased inward budding of EGFR into internal endosomal vesicles, an effect that was rescued by addition of exogenous orthologous mammalian recombinant proteins.

Figure 2: ESCRT proteins are required for inward budding of EGFR.

Partially purified HeLa endosomal membranes were isolated as described in Fig. 1b. and detected by immunoblotting using an intracellular epitope-specific antibody A) Inward budding is dependent on cytosolic components (lane 2 compared to lane 1) (p<.02) B) Reactions incubated with cytosol isolated from ESCRT-0 deficient yeast strains (vps27Δ, hse1Δ) significantly decrease protease protection of EGFR compared to cytosol isolated from a parental strain (Lane 2 and 3 compared to lane 1) (p<.02; n=4). C) Recombinant human Hrs (180nM) rescues the inhibition of EGFR protease protection (lane 3 compared to lane 2) (n=5). Recombinant human STAM (180nM) rescues the inhibition of EGFR protease protection (lane 7 compared to lane 6) (p<.05; n=4). D) Reactions incubated with cytosol isolated from ESCRT-I deficient yeast strains (vps23Δ, vps28Δ, vps37Δ, mvb12Δ), significantly inhibit the protease protection of EGFR (lanes 2–5) compared to cytosol isolated from a parental strain (lane 1) (p<.02; n=4). E) Reactions incubated with cytosol isolated from ESCRT-II deficient yeast strains (vps25Δ, snf8Δ, vps36Δ), significantly inhibit the protease protection of EGFR (lanes 2–4) compared to cytosol isolated from a parental strain (lane 1) (p<.02; n=5). F) Reactions incubated with cytosol isolated from yeast strains deleted of the ESCRT-III genes (snf7Δ, vps24Δ) or the AAA ATPase VPS4 gene (vps4Δ) significantly inhibit the protease protection of EGFR (lanes 4, 5, 7), compared to cytosol isolated from a parental strain (lane 1) (p<.05). Reactions incubated with cytosol isolated from vps20Δ or did4Δ strains prior to trypsin treatment, did not alter the protease protection of EGFR (lane 3, 6), compared to cytosol isolated from a parental strain (lane 1). Cytosol isolated from the yeast strain deficient of HSE1 was used as a control in these experiments. Reactions incubated with cytosol isolated from ESCRT-III haplo-insufficient yeast strains (yMG3 and yMG4) significantly inhibit the protease protection of EGFR compared to cytosol isolated from a parental strain (lanes 8–9) (p<.05). However, the EGFR protected was not significantly different from the single deletion strains (lane 4–5) (n=6). Data represents the mean +/− S.E.M. normalized to the control. *Denotes P<0.05 (t-test for A, D-F; Anova for B,C).

Deletion of ESCRT-I, ESCRT-II, or ESCRT-III decreases MVB internal vesicle formation and results in impaired MVB biogenesis (Bilodeau et al., 2002, Schmidt and Teis, Hurley and Hanson, 2010). To determine whether ESCRT-I and II regulate inward budding in our assay, cytosol isolated from yeast strains deleted of ESCRT-I or –II genes (Table 1) was used in place of cytosol isolated from the parental strain. Cytosol derived from ESCRT-I deficient yeast strains (vps23Δ, vps28Δ, vps37Δ, or mvb12Δ) or ESCRT-II deficient yeast strains (vps36Δ, snf8Δ, vps25Δ) significantly decreased protease protection of the EGFR epitope (Fig. 2D and Fig. 2E, respectively), suggesting that ESCRT-I and ESCRT-II components regulate inward membrane budding. To determine whether the yeast deletion strains can complement each other, vps25Δ and vps36Δ cytosols were mixed and used in place of cytosol isolated from the parental strain or the individual deletion strains (vps25Δ, vps36Δ). Cytosol derived from the combination of vps25Δ and vps36Δ strains did not significantly decrease protease protection of the EGFR epitope compared to parental cytosol (Fig. S2), suggesting that these yeast strains provide complementary components.

We also observed that protease protection of the EGFR is dependent on Vps4 as well as the ESCRT-III components Snf7 and Vps24 (Fig. 2F, lanes 3, 4, 6). However, cytosol derived from yeast strains deficient in the ESCRT-III genes, VPS20 and DID4, did not significantly impair protease protection of the EGFR (Fig. 2F, lanes 2, 5). Haplo-insufficient yeast strains of the two major ESCRT-III complexes (Snf7/Vps20 (yMG3) and Vps24/DID4 (yMG4)) were generated to determine whether the components of these complexes might have overlapping or redundant roles (Table 1). Cytosol derived from yMG3 and yMG4 yeast strains significantly decreased protease protection of EGFR compared to cytosol isolated from parental strains (Fig. 2F, lanes 7–8 compared to lane 1). However, the effect of yMG3 (45±3%) compared to the snf7Δ strains (46±4%), or yMG4 (34±3%) compared to the vps24Δ single deletion strains (44±3%), on protease protection of the EGFR epitope was not significantly different (Fig. 2F, Lane 7 compared to lane 3, and lane 8 compared to lane 4). The effect of cytosol isolated from ESCRT-III double deletion strains appears additive, suggesting that Vps20 and Did4 are not required for inward budding and Snf7 and Vps24 are sufficient to support inward budding of the EGFR.

FGFR4 and AT1R are sorted into MVBs in an ESCRT-dependent manner.

Endocytosis is required to tune the signaling responses of various types of membrane proteins (e.g. ion channels, receptors, and transporters) to extracellular stimuli. To determine whether we could measure the MVB budding of other types of membrane proteins that may transit the late endocytic pathway, we examined the MVB trafficking of the Fibroblast Growth Factor Receptor 4 (FGFR4, Fig. S3 A, B) and the G-protein coupled Angiotensin Type 1 receptor (AT1R, Fig. S3C,D). The intracellular epitopes of both FGFR4 and AT1R were protected from protease cleavage following cell-free reactions suggesting that they budded inwardly into MVBs (Fig. S3A and Fig. S3C). Cytosol derived from yeast strains lacking HSE1 significantly decreased endosomal inward budding of both FGFR4 and AT1R compared to reactions containing parental yeast cytosol (Fig. S3B and Fig. S3D). Thus, multiple types of membrane proteins transit the MVB pathway in an ESCRT-0-dependent manner and can be examined using the cell-free approach we describe herein.

Isolation of vesicles that bud outwardly from the MVB compartment.

Membrane protein cargo that is not sorted into regions of the endosomal membrane that invaginate and bud inwardly to form internal MVB vesicles may remain on the limiting endosomal membrane for incorporation into the lysosomal membrane, or could potentially bud outwardly on vesicles that would enable transport to other cellular compartments, a mechanism that could limit cargo accumulation on the MVB limiting membrane (Hurley and Hanson, 2010, Babst, 2011, Adell and Teis, 2011, Baumgart et al., 2007), and ensure consistent MVB size (Hurley and Hanson, 2010, Babst, 2011). However, these hypothesized transport vesicles have never been isolated and therefore requirements for this budding reaction are unknown. We have isolated vesicles that are liberated from the endosomal membrane during our cell-free reactions (Fig. 3). We characterized this vesicle population using light scattering and Brownian motion analysis (Nanosight Tracking Analysis, NTA), electron microscopy, and immunoblotting. NTA analysis revealed that vesicles recovered from cell-free reaction supernatant were found in a major peak with a hydrodynamic diameter of 118±13.6 nm (Fig. 3A), similar in size to what has been previously reported using electron microscopy by Felder et al. (Felder et al, 1990). When the isolated vesicles were visualized using electron microscopy they were approximately 100 nm in size (Fig. 3B). The outward budding of vesicles in our reactions is dependent on cytosolic factors (Fig. 3C, lane 2 compared to lane 1). In addition, the isolated vesicles were immunoreactive for EGFR (Fig. 3C, lane 1). The percentage of total EGFR recovered in the outwardly budded vesicles we isolated is 6.5±1.5% (Fig. 3C) compared to 36±6% of total EGFR that is protected from protease cleavage (inwardly budded) (Fig. 2A). The intracellular epitope of the EGFR was cleaved from isolated outwardly budded vesicles by trypsin incubation (Fig. 3D, lane 2 compared to lane 1), confirming that the tail domain of the EGFR is present on the outside of isolated vesicles as would be expected from vesicles that have budded outwardly from the endosomal limiting membrane.

Figure 3: Budding of vesicles from endosomal membranes.

Outwardly budded vesicles were isolated as described in Fig. 1B. A) Nanosight tracking analysis revealed the vesicles were 118+/−13.6 nm in size. B) Vesicles visualized using electron microscopy were approximately 100 nm in size. Scale bar = 100 nm. C) The amount of EGFR found in the budded vesicles is 6.5+/−1.5% of the total added into the reactions (c, lane 1). Outward budding is dependent on cytosolic components (lane 2 compared to lane 1) (p<.05; n=3). D) The intracellular epitope of the EGFR was cleaved from isolated outwardly budded vesicles by trypsin incubation. E) EGFR-containing outwardly budded vesicles were isolated from HeLa cells expressing a dominant negative Rab11 (Rab11S25N) construct containing a C-terminal myc epitope tag or a control vector (n=3). F) Vesicles derived from EGFR_K721A expressing HeLa cells showed that 5+/−1.3% of total EGFR_K721A is found in the budded vesicles (lane 1) and 19+/−4.8% of total EGFR _K721A is protected from trypsin digestion (lane 2) (n=3). Data represents the mean +/− S.E.M. normalized to the control. *denotes p < 0.05).

We further characterized the isolated vesicles, by examining whether they contained additional cargo proteins (e.g. Transferrin Receptor (TfR)) or endosomal markers (LAMP1 and Rab11) (Fig. S4A). GTPases have been implicated in multiple membrane budding events (Nicoziani and Van Deurs, 2000; Stenmark et al, 2009; Traub et al, 2010; Grant et al, 2009; Maxfield and McGraw, 2004; Thompson et al, 2007; Weigert et al, 2004; Kobayashi et al, 2013; McCaffrey and Bucci, 2001; Yamashiro and Maxfield, 1984; Hopkins and Trowbridge, 1994; Ghosh and Maxfield, 1995). The GTPase Rab11 regulates outward budding from early/sorting endosomes (van Dam and Stoorvogel, 2002, Kobayashi and Fukuda, 2013), thus we considered the possibility that the outwardly budded vesicles containing EGFR we had isolated were simply budded Rab11 vesicles. However, neither Rab11 nor Transferrin Receptor (Fig. S4A) (TfR, which is known to traffic in Rab11 compartments) were detected on vesicles isolated from our reaction supernatant. It is possible that membrane-bound Rab11 may disassociate from outwardly budding vesicles under our assay conditions resulting in undetectable levels on isolated vesicles. We therefore examined whether dominant-negative Rab11 (S25N) expression (Fig. 3E and Fig. S4B) affects outward budded vesicles containing EGFR from MVBs. To confirm that expression of Rab11S25N inhibited Rab11 function, we examined the outward budding of TfR-containing vesicles on endosomes isolated from cells transfected with Rab11S25N or a control pcDNA 3.1 empty vector. TfR was present on starting membranes when cells were starved and stimulated with EGF (Fig. S4A), but we only observed TfR on outwardly budded vesicles that had been isolated from cells that were serum starved and stimulated with transferrin (Fig. S4B.), suggesting that the protein-sorting event we are measuring is cargo-dependent and occurs on the limiting membrane of endosomes. TfR recycles from early/sorting endosomes to the plasma membrane in a Rab11-dependent manner (Ullrich et al., 1996, Ren et al., 1998) and, as expected, expression of Rab11S25N significantly inhibited TfR levels on vesicles budded from Rab11S25N endosomes compared to vesicles budded from endosomes isolated from cells transfected with a control vector (Fig. S4B). However, the percentage of total EGFR recovered in the outwardly budded vesicles we isolated from cells expressing Rab11S25N is 12.5±1.7% (Fig. 3E) and is comparable to the percentage of total EGFR recovered in the outwardly budded vesicles isolated from control cells in the absence of Rab11S25N (Fig. 3E; 12.6±1.8%), suggesting that Rab11 is not required for outward budding of EGFR-positive vesicles from MVBs. These data suggest that endosomal budding of EGFR-containing vesicles is not dependent on Rab11 and that budding of EGFR- and TfR-containing endosomal vesicles utilize distinct mechanisms.

To determine whether the outwardly budded vesicles we have characterized might play a role in membrane protein recycling or plasma membrane targeting we examined whether an EGFR mutant lacking kinase activity, EGFR_K721A, is found in outwardly budded MVB vesicles isolated from our assay. EGFR_K721A is internalized from the plasma membrane and is transported to the limiting membranes of MVBs prior to budding outwardly into vesicles, some of which reach the plasma membrane (Felder et al., 1990). We found that 5±1.3% of total EGFR _K721A is recovered in the isolated vesicles (Fig. 3F) and 19±4.8% of total EGFR _K721A is protected from trypsin digestion (inwardly budded) (Fig. 3F). These data suggest that some of the outwardly budded vesicles isolated from these cell-free reactions may be targeted to the plasma membrane.

Distinct molecular machineries regulate outward vesicle budding and inward vesicle budding.

While ESCRT proteins are clearly involved in inward MVB cargo movement, we examined whether ESCRTs may play a role in outward vesicle budding. Cytosol derived from a yeast strain lacking a representative protein of each ESCRT complex: ESCRT-0 (hse1Δ), ESCRT-I (mvb12Δ), ESCRT-II (vps36Δ), or ESCRT-III (vps24Δ) was used in place of cytosol isolated from a parental yeast strain in our reconstitution assay. Interestingly, EGFR immunoreactivity was significantly increased on outwardly budded vesicles isolated from hse1Δ reaction supernatant compared to cytosol isolated from the parental strain (Fig. 4A, lane 2 compared to lane 1). EGFR immunoreactivity was not altered in outwardly budded vesicles that were isolated from mvb12Δ, vps36Δ, or vps24Δ reaction supernatant compared to cytosol isolated from the parental strain (Fig. 4A, lane 3–5 compared to lane 1). Protease cleavage of the EGFR was inhibited in reactions containing hse1Δ, mvb12Δ, vps36Δ, or vps24Δ deficient cytosol compared with reactions containing cytosol isolated from parental strains (Fig. 4B, lane 2–5 compared to lane 1). To determine whether loss of HSE1 resulted in fewer outwardly budded vesicles, we used NTA to measure mean vesicle size and concentration (particles/mL). Our data suggests that there is no significant difference in mean vesicle size between vesicles that were isolated from hse1Δ reaction supernatant compared to cytosol isolated from the parental strain (155±15.4 nm compared to 145±8.6 nm, respectively) or in the overall number of outwardly budding vesicles (Fig. S5), suggesting that the increase we observe in the amount of EGFR incorporated into outwardly budding vesicles is due to a re-routing of EGFR from inward to outward budding vesicles, not to an increase in the number of vesicles that bud outwardly from MVBs. We also found that deletion of the other ESCRT-0 component, Vps27, increased EGFR-positive outward budding vesicles, similar to Hse1 (Fig. S6A, lane 2 and 3 compared to lane 1). Protease cleavage of the EGFR was inhibited in reactions containing hse1Δ or vps27Δ deficient cytosol compared with reactions containing cytosol isolated from parental strains (Fig. S6B, lane 2–3 compared to lane 1) suggesting that inward budding was inhibited. Thus, while ESCRT-0 is not required for EGFR-positive outward vesicle budding, inhibition of inward budding by deleting either ESCRT-0 component re-routes EGFR from inward to outward budding vesicles and suggests that inward and outward budding may be linked processes.

Figure 4. Distinct molecular mechanisms regulate inward versus outward vesicle budding.

Outwardly budded vesicles were isolated as in Figure 1B. A) Vesicles isolated from hse1Δ reaction supernatant significantly increase EGFR immunoreactivity compared to cytosol isolated from a parental strain (lane 2, compared to lane 1) (p<0.04; n=5). Vesicles isolated from mvb12Δ, vps36Δ, or vps24Δ reaction supernatant did not alter EGFR immunoreactivity compared to cytosol isolated from a parental strain (lane 3–5, compared to lane 1) (n=6). B) Reactions incubated with cytosol isolated from ESCRT deficient yeast strains (hse1Δ, mvb12Δ, vps36Δ, vps24Δ), significantly inhibit the protease protection of EGFR (lanes 2–5) compared to cytosol isolated from a parental strain (lane 1) (p<0.01; n=5). C) Vesicles isolated from vps1Δ reaction supernatant significantly decrease EGFR immunoreactivity compared to cytosol isolated from a parental strain (lane 2 compared to lane 1) (p<0.05). Recombinant human dynamin1 (1μM) partially rescues the inhibition of EGFR immunoreactivity (lane 3 compared to lane 1) (p=0.051; n=4). D) Reactions incubated with cytosol isolated from vps1Δ strains or vps1Δ strains plus dynamin1 did not alter protease protection of EGFR compared to cytosol isolated from a parental strain (lane 2 and 3 compared to lane 1) (n=3). Data represents the mean +/− S.E.M. normalized to the control. *denotes p < 0.05).

To identify cytosolic factors that may play a role in outward budding, we screened a limited number of yeast genes by isolating cytosol from yeast deletion strains and examined their ability to support outward budding. Interestingly, cytosol derived from a yeast strain lacking a dynamin-like yeast ortholog, VPS1 (vps1Δ), inhibited the amount of EGFR-immunoreactivity in isolated vesicles compared to cytosol isolated from the parental strain (Fig. 4C, lane 2 compared to lane 1). Cytosol from the vps1Δ strain had no significant effect on the protease protection of the EGFR (Fig. 4D) suggesting that inward budding was not altered in the absence of Vps1. Rescue experiments in which we added recombinant mammalian Dynamin1 (1 μM) to reactions containing Vps1-deficient cytosol, partially rescued the inhibition of EGFR-immunoreactivity observed on outward budded vesicles (Fig. 4D, lane 3 compared to lane 2). These data suggest that dynamin is required for outward budding of EGFR-containing vesicles, but not inward budding of EGFR into internal MVB vesicles.

Budding from isolated early and late endosomal compartments.

While early endosomes are thought to mature into late endosomes, it is not clear when endosomes are competent to sort membrane proteins into membrane structures that bud from their limiting membranes. To determine whether early and late endosomal populations can bud EGFR-containing vesicles, we separated endosomal populations using Optiprep gradients (Fig. 5A) and used them as donor membranes in our cell-free reactions. Coomassie staining of the resultant gradient fractions revealed the protein distribution across the gradient (Fig. S7) and immunoblotting allowed examination of markers for endosomal compartments (EEA1 for early endosomes, LAMP1/Rab7 for late endosomes, Rab11 for recycling endosomes) as well as EGFR and TfR (Fig. 5B). Mixed endosomal membranes (containing sorting and early markers, fractions 8 and 9) and late endosomal membranes (fractions 3 and 4) were collected and the protease (trypsin) resistance of EGFR for early versus late endosomal fractions was assessed. We observed that late endosomal membranes contain a greater amount of EGFR that is protected from trypsin cleavage compared to mixed endosomes (Fig. 5C), suggesting that more EGFR is incorporated into intraluminal vesicles as EGFR proceeds from early to late endosomes. Ultrastructural examination of mixed and late endosomal membrane fractions was performed to determine the size distribution of endosomes for each fraction (Fig. 5D) and average endosomal diameter in each fraction (Fig. 5E). Our results suggest that the distribution of fractions 3&4 are skewed to the right (Fig. 5D) and have a significantly increased endosomal diameter (296nm) compared to fractions 8&9 (291nm) (Fig. 5E, lane 2 compared to lane 1). Endosomes in the range of 200–500nm (indicative of MVB size) in both fractions were then analyzed for presence or absence of internal vesicles with a diameter of 50 ± 10nm (Fig. 5F). In fractions 8&9, 20% of endosomes contained internal vesicles whereas 41.8% of endosomes in fractions 3&4 contained internal vesicles (Fig. 5F). A representative endosome is shown in Fig. 5G for fractions 8&9, and in Fig. 5H for fractions 3&4. Together, these data suggest that late endosomal fractions have larger endosomal diameters and are enriched in MVBs. Mixed and late endosomal fractions were incubated in separate reactions containing ATP and yeast cytosol. We observed that inward and outward budding of EGFR occurred from both endosomal populations (Fig. 6A and Fig. 6B). Interestingly, reactions containing cytosol isolated from the hse1Δ yeast strain produced significantly increased EGFR immunoreactivity in supernatant from reactions containing late endosomal membranes compared with cytosol isolated from the parental yeast strain, suggesting that ESCRT-0 deletion triggers a re-routing of cargo into outward budding vesicles from late endosomes. In contrast, there was no increase in EGFR-immunoreactivity in supernatant from reactions containing mixed endosomal reactions and incubated with Hse1-deficient cytosol compared to the parental yeast cytosol, suggesting that outward budding from the early endosome is not altered by ESCRT-0 protein deletion (Fig. 6A, lane 2 compared to lane 1). The protease protection of the EGFR was significantly decreased in reactions containing either mixed or late endosomal membranes and incubated with Hse1-deficient cytosol (Fig. 6B, lane 2 compared to lane 1) suggesting that both membrane populations we isolated can invaginate and protect EGFR in an ESCRT-dependent manner. Reactions containing cytosol isolated from the vps1Δ yeast strain produced significantly decreased EGFR immunoreactivity in supernatant from reactions containing either mixed or late endosomal membranes, suggesting that Vps1/dynamin is required for outward budding of EGFR from both membrane populations (Fig. 6A, lane 3 compared to lane 1). Vps1-deficient cytosol had no significant effect on protease protection of EGFR in either population (Fig. 6B, lane 3 compared to lane 1) suggesting that Vps1 is not required for internal vesicle formation of MVBs. In summary, both inward and outward vesicle budding can occur from the mixed and late endosomal populations and Dynamin plays a role in outward budding of EGFR-positive vesicles from both endosomal populations. The ESCRT-0 component Hse1 reduces inward budding of EGFR-positive vesicles and enhances the outward movement of EGFR suggesting that Hse-1 plays a role in the inward vs outward routing of EGFR in the late endosome.

Figure 5. EGFR buds outwardly from late endosomal membranes in a dynamin dependent manner.

Postnuclear supernatant was loaded onto a continuous 10–20% overnight opti-prep gradient. A) Fractions were collected and the refractive index was measured. B) Fractions were immunoblotted for endosomal markers (EEA1 for early endosomes, LAMP1/Rab7 for late endosomes, RAB11 for recycling endosomes, EGFR, and TfR). C) Trypsin resistance of early and late endosomal fractions. Top: Early endosomes. Left lane: EGFR signal without trypsin digestion Right lane: EGFR signal with trypsin digestion. Bottom: Late endosomes. Left lane: EGFR signal without trypsin digestion Right lane: EGFR signal with trypsin digestion. Late endosomal membranes contained more EGFR protected from trypsin cleavage compared to early endosomal membranes. D-H) Endosomes were visualized using cryo-electron microscopy and endosomal diameter was measured using IMOD. Endosomal distribution across fractions (D) and average endosome size (E) was determined. Fractions 3&4 contain significantly larger endosomes (p<0.0001; n=475). F) Endosomes in the range of 200–500nm (indicative of MVB size) in both fractions were then analyzed for presence or absence of internal vesicles with a diameter of 50 ± 10nm. In fractions 8&9, 20% of endosomes contained internal vesicles whereas 41.8% of endosomes in fractions 3&4 contained internal vesicles. Representative images captured using cryo-electron microscopy for fractions 8&9 (G) and fractions 3&4 (H) are shown. Arrowheads represent internal vesicles. Scale bar = 100 nm

Figure 6. EGFR buds outwardly from late endosomal membranes in a dynamin dependent manner.

Post-nuclear supernatant was loaded onto a continuous 10–20% overnight opti-prep gradient (See Fig 5A). A) Early endosomes (fractions 8 and 9) incubated with cytosol isolated from hse1Δ strains did not alter EGFR immunoreactivity on vesicles isolated from reaction supernatant compared to cytosol isolated from a parental strain (lane 2 compared to lane 1). Late endosomes (fractions 3 and 4) incubated with cytosol isolated from hse1Δ strains significantly increase EGFR immunoreactivity on vesicles isolated from reaction supernatant compared to cytosol isolated from a parental strain (lane 2 compared to lane 1) (p<.02; n=7). Vesicles isolated from reaction supernatant obtained from either early or late endosomes and incubated with cytosol isolated from vps1Δ strains significantly decrease EGFR immunoreactivity compared to cytosol isolated from a parental strain (lane 3, compared to lane 1) (p<.05; n=7) B) Early and late endosomes incubated with cytosol isolated from hse1Δ strains significantly decreased protease protection of EGFR compared to compared to cytosol isolated from a parental strain (lane 2 compared to 1) (p<.01; n=7). Early and late endosomes incubated with cytosol isolated from vps1Δ strains did not alter protease protection of EGFR compared to cytosol isolated from a parental strain (lane 3 compared to lane 1). Data represents the mean +/− S.E.M. (n=7) normalized to the control. blue *denotes p < 0.05 for late endosomes, red *denotes p < 0.05 for early endosomes.

Discussion:

The number of signaling membrane proteins on the cell surface, and the time they spend in an activated state, are critical determinants for cellular responses to extracellular cues that can regulate homeostasis, plasticity, growth, and differentiation (Maxfield and McGraw, 2004; Gruenberg and Stenmark, 2004; Katzmann, 2002; Tanowitz and von Zastrow, 2003; Donaldson and Dutta, 2016; Grant and Donaldson, 2009). Internalization and movement through the endocytic pathway is required to tune the signaling responses of various membrane proteins. While outward budding from the late endosome has been suggested (Felder et al., 1990), this is the first characterization and isolation of vesicles that bud outwardly from MVBs. The ability to isolate outwardly budding vesicles has enabled their characterization and allowed examination of the machinery regulating the outward budding event. Moreover, comparison of the inward and outward budding from the same donor membranes provides an appreciation of the differences in molecular machinery required for these unique budding events. Interestingly, while differences in the mechanisms of the budding events were evident, we observed that the budding processes appear to be linked such that inhibition of inward budding triggers a re-routing of cargo from inward to outward budding vesicles.

Mechanisms of inward budding at the MVB

Since cytosolic components are required for both inward and outward budding events, and cytosol isolated from yeast and fly support these budding events, we used genetic approaches to interrogate the cytosol for molecules involved in the regulation of MVB budding. We found that ESCRT complex components are required for the protease protection of a cargo protein (EGFR), confirming that these molecules are involved in inward MVB budding in our reconstituted system. Cytosol isolated from yeast strains deleted of either of the ESCRT-0 proteins impaired EGFR budding into internal vesicles, an effect that was rescued by inclusion of the mammalian orthologous ESCRT proteins in cell-free reactions.

Similar to ESCRT-0 components, deletion of ESCRT-I impaired EGFR budding into internal MVB vesicles. These data are consistent with previous studies reporting that ESCRT-I is required for membrane protein movement into internal vesicles (Bache et al, 2001; Doyotte et al, 2005). However, the role of ESCRT-II in membrane protein movement into internal vesicles has been controversial. Previous studies suggest that ESCRT-II is dispensable for inward budding of cargo including the EGFR and major histocompatibility complex class I (MHC-I) (Malerod et al., 2007, Bowers et al., 2006). Others have found that ESCRT-II is required for EGFR and ferroportin degradation, and not required for MHC-I degradation (Langelier et al., 2006, Williams and Urbe, 2007). To address the controversy, we measured inward budding of the EGFR and used cytosol lacking ESCRT-II proteins, allowing a direct and unambiguous examination of the role of these molecules. Our results suggest that all components of the ESCRT-I and ESCRT-II complexes are required for inward budding of the EGFR into internal MVB vesicles

Deletion of only two ESCRT-III genes, SNF7 and VPS24, inhibited the movement of EGFR from the endosomal membrane into internal vesicles. These data are consistent with previous studies reporting that deletion of SNF7 or VPS24 inhibits degradation of EGFR (Bache et al., 2006, Shim et al., 2006). Somewhat surprisingly, deletion of the ESCRT-III genes VPS20 and DID4 did not inhibit the protease protection of the EGFR. To determine whether components of the two major ESCRT-III complexes (Vps20/Snf7 and Vps24/Did4) may have redundant or overlapping roles (Babst et al., 2002a), we generated haplo-insufficient yeast strains to examine EGFR protease protection. The haplo-insufficient yeast strains (Vps20/Snf7 or Did4/Vps24) did not inhibit protease protection of EGFR significantly more than cytosol from single deletion (snf7Δ or vps24Δ) strains, suggesting that the ESCRT-III components Vps20 and Did4 are not required for the inward budding of EGFR-containing vesicles. The ESCRT-III complex has been suggested to drive membrane fission events that allow internal vesicle formation within MVBs, viral budding, and cytokinesis (Hurley and Hanson, 2010, McDonald and Martin-Serrano, 2009, Carlton and Martin-Serrano, 2007). Vps20 binds to ESCRT-II and recruits Snf7, which in turn recruits the remaining ESCRT-III components for MVB budding (Adell and Teis, 2011, Hurley and Hanson, 2010). In cytokinesis and viral budding, Vps20 is dispensable and Snf7 binds to an ESCRT-associated protein, Bro1, directly activating Snf7 and the remaining ESCRT-III components (Martin-Serrano et al., 2003, Carlton and Martin-Serrano, 2007, Wemmer et al., 2011). Our results suggest that Vps20 is not required for EGFR budding into internal vesicles of MVBs, consistent with the hypothesis that other proteins (e.g. Bro1) may activate Snf7 and is similar to the mechanisms described for cytokinesis and viral budding. After formation of the Vps20/Snf7 complex, the Vps24/Did4 complex has been hypothesized to act as a cap for Snf7 that is required for membrane scission (Hurley and Hanson, 2010, Adell and Teis, 2011) and Vps4 recruitment (Williams and Urbe, 2007, Wollert et al., 2009). Our results are consistent with the hypothesis that Did4 is not required for inward budding at the MVB. We have previously shown that deletion of Vps4 or expression of a dominant negative form of Vps4 inhibits inward budding (Sun et al., 2010). Vps4 binds other ESCRT-III proteins (e.g. Snf7) (Adell et al., 2014), thus other ESCRT III proteins may recruit Vps4 to the endosomal membrane in the absence of Did4 in our reconstituted system.

Characterization of outwardly budded vesicles from MVBs

Transport from MVBs to other cellular compartments, including the Golgi and plasma membrane, presumably requires an outward vesicle budding event to cluster cargo and allow for targeted movement (Felder et al, 1990; Nicoziani and van Deurs, 2000). When characterizing vesicles that bud outwardly from MVBs during cell-free reactions, we used parental yeast cytosol (that does not contain detectable EGFR) in place of mammalian cytosol, to enable definitive measurement of the amount of EGFR that is transported from the endosome. We found that EGFR_K721A, an EGFR mutant lacking kinase activity, is present on the outwardly budding vesicles, suggesting that at least some proportion of the outwardly budding vesicles we isolate are targeted to the plasma membrane. Interestingly, Felder (Felder et al., 1990) found 25% of the EGFR_K721A on internal endosomal vesicles, 42% on recycling vesicles, and 33% on the limiting endosomal membrane (Felder et al., 1990). We found a similar percentage of EGFR_K721A present on internal vesicles, however we observed a fairly low percentage of EGFR_K721A on outwardly budded vesicles, perhaps due to a low recovery of outwardly budded vesicles in our biochemical assay.

Mechanisms of outward budding from MVBs

By reconstituting both inward and outward budding from the same endosomal membranes we were able to compare these budding events. As we observed for inward budding, EGFR-containing outward budding vesicles from MVBs are also dependent on cytosolic components. To identify the specific cytosolic components that are required for this budding event we took an unbiased approach to screen for genes that are required for outward budding of the EGFR. We observed that cytosol isolated from a dynamin-deficient yeast strain (vps1Δ) inhibited EGFR-containing outward budding vesicles in our cell-free assay, an effect that could be partially rescued by mammalian dynamin 1. The partial rescue may be a result of using the mammalian protein to rescue the yeast deletion or that other dynamin isoforms are required to fully rescue the inhibition of outward budding vesicles containing EGFR. Furthermore, dynamin 1 and 2 have been speculated to have overlapping functions in endocytosis (Ferguson et al., 2009, Liu et al., 2008). Dynamin is localized on late endosomal membranes and regulates the late endosome-Golgi recycling of the mannose-6-phosphate receptor (Nicoziani et al., 2000). Thus, a subset of the vesicles that we have isolated may be transported to the trans-Golgi network.

Small GTPases are also required for various vesicle budding steps in the secretory and endocytic pathways (Baker et al., 1990, Ruohola et al., 1988, Beckers and Balch, 1989, Melancon et al., 1987, Tooze et al., 1990). In this regard, outward budding from early endosomes is dependent on the GTPases Rab4, Rab5, and Rab11 (van Dam and Stoorvogel, 2002, Kobayashi and Fukuda, 2013). Rab11 regulates a slow recycling pathway from sorting endosomes (Grant and Donaldson, 2009, Ullrich et al., 1996, Ren et al., 1998). However, our results suggest that EGFR-containing outward budding vesicles from MVB membranes are not dependent on Rab11.

A relationship between inward MVB budding and recycling has been suggested because deletion of ESCRT proteins increases EGFR recycling to the plasma membrane (Babst et al., 2000). However, the compartment from which recycling originated (e.g. early or late endosomes) remained unresolved (Babst et al., 2000). In our reconstituted system, we found that deletion of ESCRT-0 components re-routes EGFR from inward to outward budded vesicles from the limiting membrane of the MVB, but not from early endosomes. Our results identify the MVB as an endosomal compartment from which EGFR may originate for plasma membrane recycling.

We found that deletion of ESCRT-0, but not the other ESCRT components, enhances the budding of EGFR-positive vesicles by re-routing cargo from inward to outward budding vesicles. In this regard, Vps27 acts as a scaffold to recruit Hse1 to form ESCRT-0 and in turn, sequester cargo into clusters (Babst et al, 2002; Bache et al, 2003; Hurley and Emr, 2006). Thus, our data suggest that binding to ESCRT-0 plays a role in cargo routing. In the absence of ESCRT-0 components, cargo is re-routed to bud outwardly for movement to other compartments suggesting that the ESCRT-0 complex plays a crucial role the itinerary of membrane protein cargo at the late endosome. The increase in EGFR observed on outward budding vesicles that occurs when ESCRT-0 components are removed provides insight into the role ESCRT proteins play in cargo routing.

ESCRT assembly and cargo sorting may provide a clue to the link between inward and outward budding from MVBs

Competing models have been offered to explain how ESCRT complexes may enable cargo sorting at the endosomal membrane. The conveyor belt model (Hurley and Emr, 2006) suggests that cargo molecules are handed off sequentially from one ESCRT complex to the next in a linear fashion. The concentric ring model (Nickerson et al., 2007) suggests that multiple cargoes are clustered beneath an ESCRT super complex. Our data are in favor of the concentric ring model. If membrane cargo is clustered as suggested by the concentric ring model and inward budding is inhibited, the cargo clusters would begin to accumulate on MVB membranes, leading to an increase in endosome size. Therefore, to limit cargo accumulation on the MVB limiting membrane and ensure consistent MVB size, cargo clusters are transported to other cellular compartments, leading to an increase in cargo being transported in outward budding vesicles from MVBs, consistent with our results. The linkage between inward and outward budding may be due in part, to the localization of cargo clusters and the machinery regulating inward and outward budding in microdomains on the MVB membrane. The ESCRT-0 component, Hrs, is found clustered in areas that contained EGFR (Sachse et al., 2002, Tsujimoto et al., 1999). Interestingly, clathrin, a coat proteins that regulates membrane budding events and we observed on outwardly budding vesicles is found in clusters adjacent to Hrs clusters on endosomal membranes (Sachse et al., 2002). Thus, ESCRT components, required for inward budding, and clathrin clusters, perhaps required for outward budding, are localized in adjacent membrane domains that may correspond those undergoing inward and outward budding. This close spatial arrangement of inward and outward budding machinery may play a role in the linkage of these budding events.

Conclusions:

MVB budding events play a role in diverse processes such as viral budding (Carlton and Martin-Serrano, 2007, Garrus et al., 2001, Pornillos et al., 2003) MHC I peptide display (Delamarre et al., 2005, Blum et al., 2013), and receptor trafficking (Sirisaengtaksin et al., 2014, Whittle et al., 2016, Tomas et al., 2014). Disruptions in MVB trafficking may result in disease (Grandal et al., 2007, Wang et al., 2008, Staub et al., 1997), suggesting that understanding the mechanisms underlying these budding events will provide insight into disease. We have measured both inward and outward budding events from the same donor endosomal membranes and find both steps require soluble components. In our reconstituted system we assessed the role of ESCRT complex members in inward endosomal budding and identified dynamin as a molecule involved in outward MVB budding. Although we found that inward and outward budding require distinct cytosolic factors, we observed a linkage between inward and outward budding events suggesting that trafficking at the MVB membrane may be subject to regulation that may alter the degradation/recycling of membrane proteins.

Supplementary Material

1

Supplemental Figure 1: C-terminal domain of the EGFR is degraded upon trypsin treatment. Partially purified HeLa endosomal membranes were isolated as described in Fig. 1B. Starting (donor) membranes were incubated without (lane 1) or with (lane 2) trypsin. Immunoblots were probed with EGFR antibodies directed against an C-terminal (intracellular) epitope (A) or an N-terminal (extracellular) epitope (B). Full length EGFR is detected using a C-terminally directed EGFR antibody without trypsin treatment of endosomal membranes (A, lane 1 compared to lane 2). After trypsin treatment, a band of approximately 90 kDa is detectable with the N-terminally directed EGFR antibody (B, lane 2 compared to lane 1).

2

Supplemental Figure 2: Complementation of EGFR inward budding using cytosol isolated from multiple ESCRT-II deletion strains. Partially purified HeLa endosomal membranes were isolated as described in Fig. 1B. Reactions incubated with cytosol isolated from either of the ESCRT-II deficient yeast strains (vps25Δ and vps36Δ), significantly inhibited the protease protection of EGFR (lanes 2-4) compared to cytosol isolated from a parental strain (lane 1) (p<.02; n=4). Reactions incubated with cytosols isolated from both vps25Δ and vps36Δ strains did not significantly inhibit the protease protection of EGFR compared to cytosol isolated from a parental strain. Data represents the mean +/− S.E.M. (n=4) normalized to the control. *denotes p < 0.05.

3

Supplemental Figure 3: Multiple types of membrane protein cargo are sorted at the MVB in an ESCRT-dependent manner. HeLa cells were transfected with FGFR4 or AT1R constructs containing a C-terminal V5 or myc epitope tag, respectively. A) FGFR4 containing endosomes incubated with mammalian cytosol (25 μg) result in the protease protection of the FGFR4 (n=3). B) Endosomes incubated with yeast cytosol (70 μg) isolated from hse1Δ strains significantly decrease protease protection of FGFR4 (lane 2) compared to cytosol isolated from a parental strain (lane 1) (p=.034; n=3). C) AT1R containing endosomes incubated with mammalian cytosol (25 μg) result in the protease protection of the AT1R (n=3) D) Endosomes incubated with yeast cytosol (70 μg) isolated from hse1Δ strains significantly decrease protease protection of AT1R (lane 2) compared to cytosol isolated from a parental strain (lane 1) (p=.023; n=3)). Data represents the mean +/− S.E.M. normalized to the control. *denotes p < 0.05 (t-test).

4

Supplemental Figure 4: Localization of Transferrin Receptor and Rab11. Outwardly budded vesicles were isolated as in Figure 1B. A) Endosomal membranes (5 μL, Lane 1), endosomal membranes (5 μL) digested with trypsin to remove the C-terminal epitope of the receptor (Lane 2), and EGFR-containing outwardly budded vesicles (Lane 3) were immunoblotted for EGFR, TfR, LAMP1, and Rab11. B) TfR-containing outwardly budded vesicles were isolated from HeLa cells transfected with a vector control (PcDNA 3.1) or a dominant negative Rab11 (Rab11S25N) construct. Expression of Rab11S25N greatly inhibited TfR levels on isolated vesicles. Data represent the mean +/− S.E.M. (n=3) normalized to the control. *denotes p < 0.02 (t-test).

5

Supplemental Figure 5: Size and Concentration of outwardly budded vesicles. Outwardly budded vesicles were isolated as described in Fig. 1B. A) Nanosight tracking analysis revealed the vesicles isolated from reactions containing parental cytosol were 145±8.6 nm in size. B) Nanosight tracking analysis revealed the vesicles isolated from hse1Δ reactions were 155±15.4 nm in size. C) Mean vesicle size did not change in vesicles isolated from reactions containing hse1Δ cytosol compared to vesicles isolated from reactions containing parental cytosol. D) The number of vesicles did not change in vesicles isolated from reactions containing hse1Δ cytosol compared to vesicles isolated from reactions containing parental cytosol.

6

Supplemental Figure 6: Deletion of ESCRT-0 enhances outward budding of EGFR-positive vesicles. Outwardly budded vesicles were isolated as in Figure 1B. A) Vesicles isolated from hse1Δ reaction supernatant or vps27Δ reaction supernatant significantly increase EGFR immunoreactivity compared to cytosol isolated from a parental strain (lane 2 and 3, compared to lane 1) (p<0.05; n=>3). B) Reactions incubated with cytosol isolated from hse1Δ strains or from vps27Δ strains significantly inhibit protease protection of EGFR compared to cytosol isolated from parental strains (lane 2 and 3 respectively, compared to 1) (p<.05; n=>3).

7

Supplemental Figure 7: Protein Distribution across Gradient. Post-nuclear supernatant was loaded onto a continuous 10-20% overnight opti-prep gradient. Fractions were collected and examined via SDS-Page followed by coomassie staining.

Highlights:

• We reconstituted inward and outward budding events from late endosome/MVB membranes.

• Inward and outward budding events from MVBs require cytosolic proteins.

• Inward and outward budding events from MVBs are mechanistically distinct.

• Dynamin is necessary for efficient outward budding.

List of Abbreviations:

ATP

Adenosine Tri-Phosphate

ESCRT

Endosomal sorting complexes required for transport

EGFR

Epidermal Growth Factor Receptor

MVB

Multivesicular Body

NTA

Nanosight Tracking Analysis