Cell-Free Reconstitution of Multivesicular Body Formation and Receptor Sorting

Abstract

The number of surface membrane proteins and their residence time on the plasma membrane are critical determinants of cellular responses to cues that can control plasticity, growth and differentiation. After internalization, the ultimate fate of many plasma membrane proteins is dependent on whether they are sorted for internalization into the lumenal vesicles of multivesicular bodies (MVBs), an obligate step prior to lysosomal degradation. To help to elucidate the mechanisms underlying MVB sorting, we have developed a novel cell-free assay that reconstitutes the sorting of a prototypical membrane protein, the epidermal growth factor receptor, with which we have probed some of its molecular requirements. The sorting event measured is dependent on cytosol, ATP, time, temperature and an intact proton gradient. Depletion of Hrs inhibited biochemical and morphological measures of sorting that were rescued by inclusion of recombinant Hrs in the assay. Moreover, depletion of signal-transducing adaptor molecule (STAM), or addition of mutated ATPase-deficient Vps4, also inhibited sorting. This assay reconstitutes the maturation of late endosomes, including the formation of internal vesicles and the sorting of a membrane protein, and allows biochemical investigation of this process.

Endocytosis is required for the uptake of essential nutrients from the extracellular environment, as well as to retrieve proteins and lipids that are added to the plasma membrane during fusion of secretory vesicles (1,2). The endocytic pathway can be separated into numerous stages based on the movement of cargo and the identification of morphologically defined compartments (1,3). Following transport from early to late endosomes, proteins to be degraded in the lysosome are internalized into the lumen of the late endosome via a process of membrane invagination and vesicle fission (1,3,4). This fission reaction results in the formation of an organelle with a limiting membrane and internal vesicles, called a multivesicular body (MVB) (1,3,4). The manner in which proteins are sorted on the limiting membrane of an MVB and marked for internalization into internal vesicles seems to involve attachment of ubiquitin sorting signals, and appears analogous to those that may be used for sorting plasma membrane receptors for internalization. An example of endosomal sorting in the MVB pathway is the downregulation of activated epidermal growth factor receptor (EGFR) (4–6). Upon agonist binding, activated EGFRs are rapidly internalized and a significant portion of these internalized receptors are sorted into the MVB pathway and degraded following delivery to the lumen of the lysosome (4–7). MVB sorting and the subsequent lysosomal degradation of signaling cell surface receptors is therefore a critical mechanism for regulating agonist-induced signaling.

Hrs is a mammalian protein, predominantly localized on endosomes (8,9), that physically interacts with a number of proteins including eps15 (10), SNX-1 (11), TSG101 (12) and SNAP-25 (13), previously implicated in membrane trafficking. Deletion or mutation of Hrs results in an enlarged endosomal phenotype in mouse (8), fly (14) and yeast (15). There is abundant evidence suggesting a role for Hrs in cargo sorting/trafficking at the endosome (14,16–18). Hrs binds to ubiquitinated cargo with its ubiquitin-interacting motif (UIM) domain (16,19,20) that is required for its cargo sorting function, as mutation of that domain blocks sorting of ubiquitinated cargo proteins (16,20). The endosomal sorting function has also been hypothesized to require protein complexes called endosomal sorting complexes required for transports (ESCRTs) (17,18). Hrs has been suggested to recruit the ESCRT I complex to endosomes by initially recruiting TSG101 (yeast Vps23) (17,21). Following ESCRT I binding, ESCRT II and III bind (17,22,23) and although the function of each of the complexes and their individual constituents is not clear (24), these complexes have a role in sorting and then are subsequently dissociated by the action of an AAA ATPase, Vps4 (25). Hrs also binds to signal-transducing adaptor molecule (STAM) (26) (yeast homolog Hse1) forming ESCRT-0 that may have a separate role in cargo sorting from the ESCRT I, II and III complexes (16,27). Thus, deletion of Vps27, Vps23, or Hse1 alone or Vps27/Vps23 results in an enlarged class E compartment (MVB) (15,16,28,29). However, deletion of Vps27 and Hse1 results in the inhibition of formation of internal vesicles (16). These data suggest some functional differences in the ESCRT-0 and I/II/III complexes.

The sorting event that occurs at the MVB membrane results in selective incorporation of material to be degraded into membrane patches that invaginate and pinch off into the lumen of the MVB. A liposome-based assay has revealed that the late endosomal lipid lysobisphosphatidic acid (LBPA) has the capacity to drive the formation of membrane invaginations and the Bro1/Vps31p/Alix protein can apparently regulate this process (30). To begin to understand which ESCRT complex members and associated proteins are involved in processes underlying MVB sorting, we have established a cell-free assay that measures both cargo sorting and MVB formation. By using antibodies targeted to the lumenal domain of the EGFR along with trypsin digestion, we are able to determine whether this domain of the receptor is protected from digestion as would occur if the receptor was present in a membrane-bound compartment. The localization of the EGFR is analyzed by examining whether it is protected from trypsin cleavage. We have also examined the ultrastructure of the donor and postreaction compartments. Sorting of the EGFR is dependent on cytosolic factors, ATP and temperature, and is regulated by Hrs, STAM and the catalytic activity of VPS4. This assay allows biochemical examination of molecules required for MVB cargo sorting and internal vesicle formation.

Results

An assay for reconstituting cargo sorting of the EGFR into multivesicular bodies

We designed a scheme for measuring cargo sorting into MVBs based in part on protease protection of the EGFR. This had the advantage of directly assaying a membrane protein instead of using membrane-impermeable fluorescent dyes (31), allowing us to directly monitor the fate of an important cargo molecule. The basis for this assay was to use antibodies specific to the cytoplasmic tail region of EGFR (red domain representing amino acids 1198-1210 of the human EGFR, Figure 1). If this domain was taken up into MVBs, it would no longer be accessible to exogenously added trypsin and thus protected from digestion. It would then be detectable in a western blot after incubation with ATP and cytosolic extracts. However, if this tail was still exposed, exogenous trypsin would digest it and it would no longer be detectable (Figure 1).

Figure 1. Scheme for reconstitution of protein sorting into multivesicular bodies.

HeLa cells are starved and pulsed (10 min) with EGF (50 ng/mL) to induce internalization of the EGFR from the plasma membrane. These conditions result in movement of the ligand-receptor complex into endosomes (32). Cells are lysed and partially purified membranes that contain immunoreactive EGFR are isolated. Incubation of these membranes with trypsin removes all immunoreactive EGFR as detected on a western blot using a rabbit polyclonal antibody that recognizes an epitope at the C-terminal end (amino acids 1198-2110) of the human EGFR. Incubation with rat brain cytosol, ATP, at 37°C before trypsin treatment results in protection from subsequent trypsin cleavage. Hypothetical western blotting results are shown for demonstration purposes.

To examine the requirements for trypsin protection of the EGFR, we compared the amount of EGFR present in the initial membranes with that obtained after trypsin treatment of membranes that had been incubated under various conditions. When membranes were not incubated and treated with trypsin, the levels of EGFR were undetectable (Figure 2A, lane 2). This suggested that most, if not all, of the cytoplasmic domain of the EGFR was cleaved from endosomal membranes. Incubation of reactions at 37°C in the absence of ATP did not protect the EGFR from trypsin cleavage (Figure 2A, lane 4), suggesting that energy was required for EGFR protection. In the absence of rat brain cytosol, the EGFR was also not protected from trypsin cleavage (Figure 2A, lane 5), indicating that cytosolic components were required. If the incubations were performed at 4°C, the EGFR was not protected from trypsin digestion (Figure 2A, lane 6), suggesting that this was a temperature-dependent process. However, when membranes were incubated with ATP and rat brain cytosol at 37°C, nearly 20% of the EGFR was protected from trypsin digestion (Figure 2A, lane 3). This suggested that the C-terminal domain of the EGFR was no longer exposed to the added protease presumably because it was no longer accessible. To examine whether the protection from trypsin was due to inaccessibility because of a membrane barrier, we added Triton-X-100 (1%) to the membranes during the trypsin incubation period, and under these conditions the EGFR signal was undetectable (data not shown). As we have used trypsin to digest the EGFR that has not been internalized, we examined whether we are able to detect the ectodomain that should remain in endosomes that have been treated with trypsin. Membranes were incubated without trypsin (Figure 2B, lanes 1 and 3) or with trypsin (Figure 2B, lanes 2 and 4), separated by SDS–PAGE and the resulting blot containing lanes 1 and 2 was probed with an EGFR antibody directed against an extracellular epitope (Millipore Inc), while the blot containing lanes 3 and 4 was probed with an EGFR antibody directed against an intracellular epitope (ABR Inc). The full-length EGFR is detected in the absence of trypsin treatment (Figure 2B, lane 3). After trypsin treatment, a band of approximately 90 KDa is detected using the extracellular EGFR antibody (Figure 2B, lane 2), indicating that this domain remains associated with the endosome membrane after trypsin treatment. To examine whether a proton gradient must be present for the reaction, we incubated membranes with two different protonophores, carbonyl cyanide chlorophehylhydrzone (CCP) and nigericin, and under both conditions the EGFR signal was markedly reduced (10.6 ± 0.7 and 7.0 0.5% of the control reaction, respectively; Figure 2C, ± lanes 4 and 5, n = 3). In addition to protection of the EGFR after the sorting reaction, we examined whether the ESCRT proteins Hrs, STAM and TSG101 were protected from trypsin digestion after the sorting reaction. These proteins are predicted to be excluded from internal vesicles based on proposed models, suggesting that they are used for protein sorting on the MVB limiting membrane (33,34). As expected, Hrs and STAM were not protease protected after the sorting reaction and the control transmembrane protein lysosome-associated membrane protein (LAMP) remains associated with membranes (Figure 2D, lane 3). However, TSG101 was partially protected after the sorting reaction (Figure 2D, lane 3). The reason for the partial protection of TSG101 is unclear, although it has been previously suggested that TSG101 is present on LAMP-positive membranes and stays associated with membranes longer than Hrs (27).

Figure 2. Protease protection of the EGFR is dependent on energy, cytosolic extracts and temperature.

A) HeLa cells were lysed and a 1500 × g supernatant was obtained (see Materials and Methods). Six aliquots of membranes were treated as follows: lane 1, starting material (20% input); lane 2, starting material with no incubation (trypsin treated); lane 3, membranes after incubation for 3h at 37°C with ATP and rat brain cytosol; lane 4, membranes after incubation for 3 h at 37°C with rat brain cytosol and apyrase (to remove endogenous ATP) but lacking exogenous ATP; lane 5, membranes after incubation for 3 h at 37°C with ATP and no rat brain cytosol and lane 6, membranes after incubation for 3 h at 4°C with ATP and rat brain cytosol. After incubation, the reactions in lanes 2–6 were treated with trypsin on ice for 30 min. All proteins were separated by SDS–PAGE, blotted to nitrocellulose, and probed with primary antibodies against the EGFR C-terminal tail and visualized with ECL. A film image is shown that is representative of three such experiments. The amount of protease protected EGFR is approximately 20% of that found on the initial membranes (cf. lanes 1 and 3). B) Starting membranes were incubated without trypsin (lanes 1 and 3) and with trypsin (lanes 2 and 4). The blot containing lanes 1 and 2 was probed with an EGFR antibody directed against an extracellular epitope (Millipore Inc), whereas the blot containing lanes 3 and 4 was probed with an EGFR antibody directed against an intracellular epitope (ABR Inc). After trypsin treatment, a band of approximately 90 KDa is apparent in lane 2. C) Effect of protonophores on the sorting reaction. Western blot of EGFR labeling of reaction with starting membranes (lane 1, 20% loaded, membranes), reactions incubated at 0°C (lane 2), reactions incubated at 37°C for 3 h (lane 3), reactions incubated for 3 h that included CCCP (0.1 mm, lane 4), reactions incubated for 3 h that included nigericin (5 μm, lane 5) or reactions incubated with the vehicle [dimethyl sulphoxide (DMSO), lane 6] control (n = 3). D) Effect of trypsin on the presence of Hrs, STAM, TSG101 and LAMP1 localization after the sorting reaction. Starting membranes (lane 1) were incubated at either 0°C (lane 2) or 37°C (lanes 3 and 4) with cytosol and an ATP regenerating system for 3 h followed either by incubation with trypsin (lane 3) or no trypsin (lane 4). The resulting blots were probed with EGFR (intracellular epitope), Hrs, STAM, TSG101 or LAMP1 (n = 3).

We determined the time required for maximal protection of the C-terminal fragment of the EGFR from trypsin cleavage. Reactions that included membranes, ATP and rat brain cytosol were incubated at 37°C for 0–4 h prior to trypsin cleavage. The EGFR epitope was protected from trypsin cleavage in a time-dependent manner (Figure 3B). The protection appeared to occur in a step-wise manner with minimal protection after 30-min incubation (Figure 3B, lane 2) and two distinct steps of increased protection at 1–2 h (lanes 3 and 4) and at 3–4 h (Figure 3B, lanes 5 and 6). Maximal protection appeared to occur at 3 h, because the 4-h reaction resulted in a small but significant decrease in amount of EGFR epitope protected. Moreover, incubation for longer periods of time resulted in the formation of increased number of internal vesicles (Figure 3A). We examined whether yeast cytosol could support this reaction, as there are many yeast mutants available that could be used to probe mechanistic aspects of MVB cargo sorting. However, cytosol derived from either wild-type or many different vps mutant strains were not able to support cargo sorting using our assay conditions (data not shown). HeLa cell cytosol can support the reaction although it was less efficient (amount of EGFR protected after 3 h) than rat brain cytosol (data not shown).

Figure 3. An increase in internal vesicles is observed in endosomes from reactions that result in protection of the EGFR from trypsin digestion.

A) Membranes were incubated with ATP and cytosol for 1 and 3 h at 37°C. After incubation, the membranes were collected by centrifugation at 15 000 × g for 30 min. The supernatants were removed and the pellets were fixed and processed for electron microscopy. As a control, the starting membranes were incubated on ice for 3 h, ATP and cytosol were added, and then membranes were collected by centrifugation at 15 000 × g for 30 min (control). The pellet was processed identically to the membranes that were incubated at 37°C. The number of internal vesicles was counted in double membrane-delineated structures >300 nm in diameter. * indicates p ≤ 0.05. Scale bar = 100 nm. B) Western blot of EGFR C-terminal epitope in reactions (n = 3) that included starting membranes incubated with cytosol and ATP regenerating system at 37°C for 0 min (lane 1), 30 min (lane 2), 1 h (lane 3), 2 h (lane 4), 3h(lane 5) and 4 h (lane 6).

The number of internal vesicles in endosomes increases during reactions

Ultrastructural examination of endosomal membranes before and after incubation with ATP and rat brain cytosol revealed morphological differences. While endosome mean diameter was not significantly altered (initial membranes = 453.7 ± 23.7 nm, n = 300; after 3-h reaction = 416.4 ± 29.9 nm, n = 368, p = 0.35), the number of internal vesicles observed inside the endosomes was significantly increased after a 1-h incubation (Figure 3A). After 3 h, a sixfold increase in the number of internal vesicles (compared with control) was observed (Figure 3A). Invagination of membranous vesicles from the limiting endosomal membranes could frequently be observed (Figure 3A, closed arrows). The presence of apparent coats on internal vesicles (Figure 3A, double arrowheads) was observed infrequently (≤1 per 500 internal vesicles observed) and may have been the result of vesicles formed from remaining plasma membrane structures.

Hrs is required for protection of the EGFR C-terminal tail from trypsin cleavage

To examine the dependence of the EGFR trypsin protection on proteins thought to be required for endosomal sorting, we depleted Hrs from both cytosol and membranes. In the absence of Hrs, the C-terminus of the EGFR was not protected from trypsin cleavage (Figure 4A, lane 2). This suggested that Hrs was required for protection of the EGFR from trypsin. As might be expected, depletion of Hrs resulted in a partial co-depletion of STAM (16.5 ± 5% remaining in the cytosol, n = 3). However, when recombinant Hrs (that contained no detectable STAM protein by western analysis) was added into reactions that had previously been depleted of Hrs, the Hrs bound to membranes (Figure 4A, lane 3), and approximately 60% (compared with the control reaction, lane 1) of the EGFR was protected from trypsin cleavage (Figure 4A, lane 3). Ultrastructural examination of membranes that had been depleted of Hrs revealed that they contained significantly fewer internal vesicles than MVBs from control reactions (control = 7.96 ± 1.08, n = 286; Hrs depleted = 1.25 ± 0.06, p ≤ 0.01, n = 314; Figure 4B), although the internal vesicle size was significantly larger (control reactions = 53.7 ± 3.2 nm, n = 301; Hrs depleted = 62.5 ± 3.1 nm, n = 335). The mean endosomal diameter was not significantly different when Hrs was depleted (control = 483.5 ± 37.6 nm, n = 199; Hrs depleted = 459.4 ± 29.3 nm, n = 179, p = 0.61). When recombinant Hrs was added into Hrs-depleted reactions, the number of internal vesicles was significantly increased compared with the Hrs-depleted endosomes (control = 4.5 ± 0.3, n = 218, p ≤ 0.01; Figure 4B), although the mean endosomal diameter was not significantly altered. Thus, Hrs depletion significantly decreased the number of internal vesicles while increasing their size. These results suggest that Hrs is required for MVB maturation and may play a role in determining the size of internal MVB vesicles (34).

Figure 4. Hrs is required for protection of EGFR from trypsin cleavage and formation of internal endosomal vesicles.

A) HeLa cells were lysed and a 1500 × g supernatant was obtained (see Materials and Methods). Three aliquots of membranes were treated as follows: lane 1 (control), membrane proteins after incubation for 3 h at 37°C with ATP and rat brain cytosol were separated by SDS–PAGE and probed with anti-C-terminal EGFR and Hrs antibodies; lane 2 (Hrs depleted), Hrs was immunodepleted from the rat brain cytosol and starting membranes. In additional, the Hrs domain required for membrane association was added and the reaction was incubated as in lane 1. Lane 3 (Hrs restored): exogenous Hrs (180 nM) was added to reactions that had been depleted of Hrs as in lane 2. Western blots for both EGFR and Hrs are displayed above and quantitation of the band intensity (NIH image) is shown below. B) Ultrastructural examination of initial membranes (0), membranes from reactions that had been depleted of Hrs (depleted, as indicated) and membranes depleted of Hrs that were subsequently incubated with exogenous Hrs (restored, as indicated). Scale bar indicates 500 nm. * indicates p ≤ 0.05, ** indicates p = 0.2. Bar in inset is 100 nm.

STAM depletion inhibits internal vesicle formation and is required for protection of the EGFR C-terminal tail from trypsin cleavage

The ESCRT-0 complex of Hrs-STAM is thought to be an initiating step for vesicle formation and as we observed a requirement for Hrs in the internalization of EGFR, we examined whether STAM was also required for internalization in our assay. In the absence of STAM, the C-terminus of the EGFR was not protected from trypsin cleavage (Figure 5A, lane 2). This suggested that STAM was required for protection of the EGFR from trypsin. Ultrastructural examination of membranes that had been depleted of STAM revealed that they contained significantly fewer internal vesicles than MVBs from control reactions (control = 7.11 ± 1.63, n = 164; STAM depleted = 2.50 ± 0.09, n = 1779, p < 0.01; Figure 4B), the vesicle size was significantly smaller (control reactions = 49.3 ± 5.0 nm, n = 863; STAM depleted = 33.4 ± 2.2 nm, n = 1849, p ≤ 0.01) and the mean endosomal diameter was significantly larger when STAM was depleted (control = 543.9 ± 41.6 nm, n = 653; STAM depleted = 698.8 ± 64.2 nm, n = 711, p ≤ 0.05). Thus, STAM depletetion significantly decreased the number of internal vesicles and decreased their size. These results suggest that STAM is required for MVB maturation and may play a role in determining the size of internal MVB vesicles.

Figure 5. STAM depletion inhibits protection of EGFR from trypsin cleavage and formation of internal endosomal vesicles.

A) HeLa cell membrane proteins after incubation for 3h at 37°C with ATP and rat brain cytosol were separated by SDS–PAGE and probed with anti-C-terminal EGFR antibodies (top). Anti-STAM antibodies were used to probe membranes and rat brain cytosol that were used in reactions before (lane 1) and after (lane 2) immunodepletion of STAM. Bar graph depicts quantification of the western blot data. B) Ultrastructural examination of initial membranes (top, STAM), membranes from reactions that had been depleted of STAM (bottom). * indicates p ≤ 0.05. Scale bar indicates 500 nm.

Vps4E236Q inhibits internal endosomal vesicle formation and trypsin protection of the EGFR C-terminal tail in a concentration-dependent manner

To examine the role of Vps4 in trypsin protection of the EGFR C-terminal tail, various concentrations of a dominant-negative Vps4 protein lacking ATPase activity were added into the endosomal EGFR sorting reactions. Vps4E236Q inhibited the protection of the EGFR C-terminal tail in a concentration-dependent manner and at the highest concentration the EGFR was only protected minimally (Figure 6, lanes 1–6). Ultrastructural examination of membranes that had been incubated with Vps4E236Q (10.8 μM) revealed that they contained significantly fewer internal vesicles than MVBs from control reactions (control = 11.20 ± 0.65, n = 226; Vps4E236Q = 3.37 ± 0.08, n = 746, p ≤ 0.01; p Figure 6B), although the size was not altered (control = 50.9 ± 2.2 nm, n = 330; Vps4E236Q = 48.2 ± 4.8 nm, n = 266) nor was the mean endosomal diameter (control = 496.8 ± 26.6 nm, n = 344; Vps4E236Q = 463.5 ± 37.2 nm, n = 1748, p = 0.70). Thus, exogenous Vps4E236Q decreased the number of internal vesicles without significantly altering the size of either the endosomes or the internal vesicles, suggesting that the ATPase activity of Vps4 is linked to the process of MVB vesicle formation.

Figure 6. Vps4E236Q inhibits protection of EGFR from trypsin cleavage and formation of internal endosomal vesicles.

A) HeLa cells were lysed and a 1500 × g supernatant was obtained (see Materials and Methods). Six aliquots of membranes were treated as follows: lane 1 (control), membrane proteins after incubation for 3 h at 37°C with ATP and rat brain cytosol were separated by SDS–PAGE and probed with anti-C-terminal EGFR antibodies; lanes 2–6, exogenous Vps4E236Q in increasing concentrations was added to reactions as described for lane 1. B) Ultrastructural examination of initial membranes (0), membranes from reactions that had been incubated with Vps4E236Q (3h + Vps4E236Q, as indicated) and membranes that were incubated without exogenous ps4E236Q (3 h control, as indicated). * indicates p ≤ 0.05. Scale bar indicates 500 nm.

Discussion

We have taken advantage of the well-known itinerary of the EGFR and classical protease protection experiments, with the addition of ultrastructural measures, to devise a cell-free method to examine the sorting of a membrane protein into the endosomal lumen. The sorting reaction is time, temperature, cytosol and energy dependent. EGFR sorting required proteins (Hrs, STAM and the ATPase activity of Vps4) that have been previously implicated in late endosomal sorting. These data suggest that this cell-free assay can be used to examine the mechanistic aspects of this critical sorting step that is used to separate proteins to be recycled from those that are degraded.

While the biochemical measure we utilized to examine cargo movement (protection of a C-terminal EGFR epitope) suggests sorting into membrane-impermeable compartments, the ultrastructural data we report describe the formation of internal endosomal vesicles. Vesicle formation and cargo sorting may be two independent events (35,36). The role of cargo proteins in this process may be passive, wherein vesicles form whether or not there is cargo for them to sort (36), or active, such that cargo triggers vesicle formation (35–37). The role of stimulation of cargo internalization on endosome formation in whole cells is difficult to separate from signaling processes that are triggered by activation of the cargo. In isolated endosomes, this elevator or escalator mechanism of MVB vesicle formation may be examined. We found that the EGFR epitope was protected from trypsin digestion in what appeared to be a step-wise manner, perhaps suggesting that several rounds of sorting and internalization can occur during the reaction period. The number of internal vesicles formed also increased with longer reaction times. The number of vesicles observed at 0 h in double-membrane compartments > 300 nm diameter was between 0 and 1. This may indicate that we are selecting for light endosomal membranes for our reactions, likely the result of the centrifugation steps used to isolate the endosomal membranes. It is not clear whether the rate of vesicle formation would be equivalent when starting with endosomes containing fewer or greater number of internal vesicles; however, the present data suggest that the presence/sorting of cargo and internal MVB vesicle formation are parallel processes.

The dependence of EGFR protease protection on Hrs, STAM and Vps4 was expected to be based on the hypothesized role of these proteins in recruiting machinery (Hrs), sorting cargo (STAM) and ESCRT complex disassembly (Vps4) involved in MVB sorting. Depletion of Hrs was rescued by full-length recombinant Hrs that binds to endosomal membranes in a manner dependent on the coiled-coil domains (32). While immunodepletion of Hrs from the required cytosol also resulted in a partial depletion of STAM and may have resulted in co-depletion of other cytosolic proteins, it is noteworthy that the addition of recombinant Hrs, which contains no detectable STAM protein, can restore the reaction to approximately 60% of its predepletion efficiency. Hrs and STAM depletion as well as Vps4E236Q incubation significantly decreased the number of internal endosomal vesicles observed after the reaction in agreement with previous studies in whole cells (38–40).

ESCRT complex-mediated cargo sorting has been linked to vesicle formation (33,34,44,45), although the mechanistic details of this linkage are uncertain. Multiple models exist describing the roles of ESCRT complexes in cargo sorting; one model proposes that cargo molecules are handed off consecutively from one ESCRT complex to another, while another suggests that multiple cargoes are clustered to an assembly of multiple ESCRT complexes. While there is evidence in support of both models (33,34), we are intrigued by the concentric ring model and its testable predictions for vesicle formation. The ESCRT-0 complex has been proposed to form the center of an assembly of concentric rings of ESCRT complexes (34). This model predicts that ESCRT-mediated membrane sorting is directly linked to regulation of membrane invagination and internal vesicle formation. Internal vesicle size is predicted to be determined by the surface area occupied by the ESCRT-0/I/II core that is surrounded by the ESCRT III ring that would contract to drive vesicle formation. Thus, decreased ESCRT-0 reduces internal vesicle size, while increased internal vesicle size would result from deregulation of ESCRT III disassembly. We observed a decrease in the number of internal vesicles formed when either Hrs or STAM was depleted. Moreover, we observed an increase in endosomal diameter and a decrease in internal vesicle size when STAM was depleted as would be predicted by the concentric ring model. However, Hrs depletion resulted in no change in endosome diameter but significantly increased internal vesicle size. While Vps4E236Q also decreased the number of internal vesicles, it did not alter the mean endosome diameter and had no significant effect on internal vesicle size. As Vps4-mediated contraction of the ESCRT III ring has been suggested to drive vesicle formation, we hypothesized that addition of Vps4E236Q into the reaction might inhibit the invagination process (39), similar to that observed at the plasma membrane with dynamin mutants (41), such that invaginated profiles might be arrested. However, these profiles were only occasionally observed and their presence was not more frequent than in reactions lacking Vps4E236Q.

During the course of these studies, a report appeared by Gruenberg et al. that used a cell-free method to study endosomes albeit with several methodological differences (31). We chose cargo protection as a measure of internalization while Falguières et al. (31) primarily examined the fluorescence of the pH-sensitive fluorescent molecule 8-hydroxypyrene-1,3,6-trisulfonate (HPTS). The time–courses of our sorting events differed. We found that trypsin protection of the EGFR epitope was signifi-cant after a 60-min reaction and was maintained after a 3-h reaction, while a plateau for the sorting reaction was not reached in the Falguières study (31). However, we examined cargo movement of a membrane protein while Falguières et al. (31) examined fluorescence changes of a soluble protein. Both studies determined some molecular components that are required for the sorting events measured. We manipulated the reaction conditions by directly adding recombinant proteins or depleting them from the membranes and required cytosolic component, whereas Falguières et al. manipulated the cells from which the membranes were harvested, sometimes for days prior to the experiment. It is perhaps because of the differences in experimental approaches that some differences in molecular requirements were noted. For example, we observed a marked effect of Hrs depletion (inhibition of cargo sorting that was rescued by addition of recombinant protein), whereas Falguières et al. found little effect of partial depletion of Hrs by prior transfection of RNAi (31). However, both studies observed an inhibition of their measured events after manipulating the ATPase activity of Vps4.

The cytosol and ATP dependence of this reaction is intriguing and begs for additional molecular examination. Many proteins are hypothesized to play important roles in protein sorting at the MVB membrane and a cell-free assay will allow investigation of the function of various molecules. Examination of outward budding from the MVB compartment as must occur in order for cargo transport to compartments other than the lysosome to occur could also potentially be examined. Classical budding assays that make use of centrifugation methods to separate vesicular fractions (42) could likely be incorporated. These studies might enable an understanding of whether the requirements for the outward budding into the cytosolic compartment are different than that of inward budding into the endosome lumen. Nevertheless, the assay described herein will be useful for understanding the molecular aspects of the critical sorting step for membrane proteins at the MVB.

Materials and Methods

Materials

Antibodies were obtained from the following sources: EEA1 (Affinity Bio Reagents), 6XHIS (Sigma), Hrs (Axxora), myc (Sigma), FLAG (Sigma) and EGFR (Affinity Bio Reagents and Millipore Inc). Anti-STAM monoclonal antibodies were generated after immunizing mice with recombinant full-length STAM protein produced in baculovirus (Figure S1). After immunizing mice with full-length STAM protein, we isolated the STAM antibody used in the present studies (O8G4, IgG2b subtype) for immunoprecipitation and western analysis (Figure S1).

Protein production and purification

For Vps4B(E236Q), Escherichia coli (BL21) was transformed with a kanamycin-resistant plasmid harboring His₆-VPS4B(E236Q)-myc and grown overnight on kanamycin-containing agar plates. Bacteria were cultured at 37°C in terrific broth (TB) media until an optical density (OD) of 600 nm = 1; flasks were then cooled to 30°C and induced with 0.4 μm isopropyl-β-d-thiogalactopyranoside (IPTG) for 3 h. Bacteria were harvested by centrifugation; and lysed in buffer (50 mm HEPES pH 7.6, 150 mm KCl, 5% glycerol, 2 mm MgCl₂, 0.5 mm ATP and 2 mm DTT) with sonication, 0.25% Triton-X-100 and 1 mm phenylmethylsulphonyl fluoride (PMSF). Bacterial lysates were clarified by centrifugation at 12 000 × g for 20 min and bound to Nickel nitrilotriacetic acid (Ni-NTA) resin (Qiagen) in the presence of 5 mm imidazole. Resin was then washed in 750 mm NaCl to remove bacterial contaminant DNA, and proteins were eluted with imidazole step gradients. Fractions containing Vps4B(E236Q) were collected and subjected to size exclusion chromatography using a Superdex 200 (GE Healthcare) gel filtration column equilibrated in lysis buffer. Column fractions containing VPS4B(E236Q) were identified by SDS–PAGE and Coomassie staining. Fractions were pooled and quantified using a Bradford protein assay with BSA protein standards. Protein aliquots were then snap-frozen in liquid nitrogen, and stored at -80°C. Hrs and STAM were expressed in insect cells as previously described (9,43). No detectable STAM was present in purified Hrs preparations, nor was there any detectable Hrs in purified STAM preparations.

Cell culture

HeLa cells were cultured as a monolayer in 10-cm plastic plates in DMEM containing 10% FBS under 5% CO₂ at 37°C. Before each experiment, cells were split with trypsin/ethylenediaminetetraacetic acid (EDTA) and seeded into 10-cm tissue culture plates.

Cell-free reconstitution of multivesicular body formation

HeLa cells were grown in DMEM with 10% bovine serum in 10-cm plates (usually eight plates) to 75–80% confluence. Before harvesting, the cells were serum starved for 2 h. Cells were stimulated with EGF (100 ng/mL, 10 min at 37°C), and subsequently placed on ice where the media were removed and cells were washed with ice-cold PBS (3× with 5 mL). The cells were scraped from the plate and washed with 2 mL ice-cold PBS, followed by centrifugation at 1500 × g for 10 min at 4°C. The cell pellet was resuspended in 170 μL of homogenization buffer (20 mm HEPES 7.4, 0.25 m sucrose, 2 mm EGTA, 2 mm EDTA and 0.1 mm DTT) and a protease inhibitor cocktail (112 μm PMSF, 3 μm aprotinin, 112 μm leupeptin, 17 μm pepstatin). The cells were then drawn through a 30-gauge needle into a 1-mL syringe 30× within approximately 6 min. The lysate was centrifuged at 800 × g for 5 min to remove debris and the supernatant was then centrifuged for 15 min at 1500 × g. Membranes were recovered from the resulting supernatant and used for reconstitution reactions (usually eight reactions from eight plates of HeLa cells).

A standard reaction (50 μL) contained 15 μL of membranes, 15 μLof rat brain cytosol [6.8 mg/mL, prepared as described in Ref (32)], 6 μL ATP regeneration system [2 mm MgATP, 50 μg/mL creatine kinase, 8 mm phosphocreatine and 1 mm DTT of final concentrations; (32)] and 14 μL homogenization buffer. The reactions are incubated for various times at 37°C. During the incubation period, the reactions were gently mixed by tapping the tube every 30 min. After incubation, the reactions were placed on ice and trypsin was added (6 μL, 0.27 μg/μL) and incubated for an additional 30 min. The reactions were centrifuged (15 000 × g for 30 min at 4°C) and the pellet was resuspended in sample buffer. The samples were boiled and proteins were separated by SDS–PAGE, followed by transfer to nitrocellulose using standard conditions. After blocking with 5% non-fat dry milk in PBS, the blot was probed with an antibody that recognizes the intracellular domain of EGFR (ABR Inc, 1:1000 dilution, overnight at 4°C), followed by goat anti-rabbit polyclonal antibody conjugated to horse-radish peroxidase (HRP) (Sigma-Aldrich Inc, for 1 h at room temperature). For some experiments, an antibody directed against an ectodomain epitope (Millipore Inc) was employed. Proteins were detected with enhanced chemiluminescence (ECL; Pierce) and exposed to X-ray film.

For experiments in which detergent was added to the reaction, Triton-X-100 (1%) was added to the reaction for 30 min at 4°C prior to trypsin treatment. To determine the requirement for rat brain cytosol, the cytosol was replaced in the reaction by homogenization buffer. To examine whether the reaction was dependent on ATP, we did not include ATP or its regenerating system in the reactions and inhibited endogenous ATP with apyrase (6 units/mL of final concentration). Comparing the result of incubating the reactions at 37°C versus on ice assessed the effect of temperature. Time dependence was examined by allowing the reactions to proceed for varying lengths of time prior to stopping the reactions with sample buffer.

Hrs and STAM depletion and reconstitution

To deplete Hrs from membranes, the membranes were incubated with a monoclonal antibody against Hrs (9) (2 μg per reaction) and a fragment of Hrs containing the coiled-coil region (29.2 μm) for 30 min on ice followed by centrifugation to remove the released protein (32). Pilot experiments that determined this combination resulted in maximal depletion of Hrs from starting membranes. As Hrs is also found in the cytosol, we immunodepleted Hrs from the rat brain cytosol used in these experiments by incubation with a monoclonal anti-Hrs antibody for 1 h at 4°C. The Hrs-antibody complexes were recovered with protein A–Sepharose and the resulting supernatant was depleted of Hrs and used in the reaction. To reconstitute Hrs activity after depletion, we added recombinant Hrs protein into the reaction. We have previously shown that this recombinant Hrs protein binds to endosomal membranes in a SNAP-25-dependent manner (32). To deplete STAM from reactions, we incubated the STAM monoclonal antibody with starting membranes and rat brain cytosol as described earlier for Hrs depletion. Pilot experiments were used to determine the optimal amount of antibody (15 μg per reaction) for this purpose.

Electron microscopy

Membranes were harvested (15 000 × g for 30 min) after incubation under various reaction conditions. Pellets were fixed with 3% glutaraldehyde in cacodylate buffer for 2–16 h at 4°C, embedded in epon, and sections were cut and viewed on a microscope (Jeol model 1010). Images were captured directly with a camera (Hamamatsu Orca). For purposes of quantitation in this study, endosomes were defined as double membrane-delineated structures > 300 nm in diameter that did not contain membrane whorls or electron dense material. We quantified internal vesicle number, vesicle diameter and endosome diameter by manual counting or measurement (double-blind) on images taken from multiple fields per condition. For endosome or vesicle diameter, we used the measuring tool in Adobe Photoshop and corrected for the scale using the scale bar on the micrograph and the magnification. The number of internal vesicles was counted. Potential differences in vesicle number, size or endosome diameter were assessed using a one-way anova with post hoc Tukey test for group comparisons (http://faculty.vassar.edu/lowry/VassarStats.html).