Proteins involved in vesicular transport and membrane fusion.

In the past year, new information about proteins involved in vesicular transport has been plentiful. Particularly noteworthy are the complementary findings that Sec17p is required for vesicle consumption in endoplasmic reticulum-to-Golgi transport in yeast and that an analogous activity in mammalian cells, termed SNAP, is required for transport from the cis to the medial cisternae of the Golgi apparatus.

Receptor-mediated uptake of antigen/heat shock protein complexes results in major histocompatibility complex class I antigen presentation via two distinct processing pathways.

Heat shock proteins (HSPs) derived from tumors or virally infected cells can stimulate antigen-specific CD8(+) T cell responses in vitro and in vivo. Although this antigenicity is known to arise from HSP-associated peptides presented to the immune system by major histocompatibility complex (MHC) class I molecules, the cell biology underlying this presentation process remains poorly understood. Here we show that HSP 70 binds to the surface of antigen presenting cells by a mechanism with the characteristics of a saturable receptor system. After this membrane interaction, processing and MHC class I presentation of the HSP-associated antigen can occur via either a cytosolic (transporter associated with antigen processing [TAP] and proteasome-dependent) or an endosomal (TAP and proteasome-independent) route, with the preferred pathway determined by the sequence context of the optimal antigenic peptide within the HSP-associated material. These findings not only characterize two highly efficient, specific pathways leading to the conversion of HSP-associated antigens into ligands for CD8(+) T cells, they also imply the existence of a mechanism for receptor-facilitated transmembrane transport of HSP or HSP-associated ligands from the plasma membrane or lumen of endosomes into the cytosol.

Transport of newly synthesized vesicular stomatitis viral glycoprotein to purified Golgi membranes.

In a previous communication we reported that the newly synthesized membrane glycoprotein of vesicular stomatitis virus could be transported in crude extracts of CHO cells from endoplasmic reticulum-derived membranes to membranes of the Golgi complex. This conclusion was an indirect one, based on the terminal glycosylation of this glycoprotein, a reaction that was dependent upon a Golgi-specific enzyme, UDP-GlcNAc transferase I. We show here that the Golgi fraction of rat liver will substitute for members of CHO cells as a source of transferase I in this reaction. The use of highly purified fractions of liver Golgi membranes, coupled with the ability to recover these membranes from incubations, has now permitted a direct demonstration of net transport of G protein to these heterologous Golgi membranes. This transport reaction is specific, in that the smooth endoplasmic reticulum fraction will not substitute for the Golgi fraction, is quantitatively significant, involving at least 30% of the viral glycoprotein, and is sustained only in the presence of both ATP and a soluble, cytosol fraction of liver cells.

Transient activity of Golgi-like membranes as donors of vesicular stomatitis viral glycoprotein in vitro.

Previous reports demonstrated that the vesicular stomatitis viral glycoprotein (G protein), initially present in membranes of a Chinese hamster ovary mutant cell line (clone 15B) that is incapable of terminal glycosylation, can be transferred in vitro to exogenous Golgi membranes and there glycosylated (E. Fries and J. E. Rothman, 1980, Proc. Natl. Acad. Sci. U. S. A. 77:3870-3874; and J. E. Rothman and E. Fries, 1981, J. Cell Biol. 89:162-168). Here we present evidence that Golgi-like membranes serve as donors of G protein in this process. Pulse-chase experiments revealed that the donor activity of membranes is greatest at approximately 10 min of chase, a time when G protein has been shown to have arrived in Golgi stacks (J. E. Bergmann, K. T. Tokuyasu, and S. J. Singer, 1981, Proc. Natl. Acad. Sci. U. S. A. 78:1746-1750). Additional evidence that the G protein that is transferred to exogenous Golgi membranes in vitro had already entered the Golgi membranes in vivo was provided by observations that its oligosaccharides had already been trimmed, and that its distribution in a sucrose density gradient was coincident with that of enzymatic markers of Golgi membranes. The capacity of this Golgi-like membrane to serve as donor is transient, declining within 5 min after "trimming" in vivo as the G protein enters a "nontransferable" pool. The rapidity of the process suggests that both the "transferable" and "nontransferable" pools of G protein reside in Golgi-like membranes.

The capacity to retrieve escaped ER proteins extends to the trans-most cisterna of the Golgi stack.

To explore how far into the Golgi stack the capacity to retrieve KDEL proteins extends, we have introduced an exogenous probe (the peptide YHPNSTCSEKDEL) into the TGN of living cells. For this purpose, a CHO cell line expressing a c-myc-tagged version of the transmembrane protein TGN38--which cycles between the TGN and the cell surface--was generated. The cells internalized peptides that were disulfide bonded to anti-myc antibodies and accumulated the peptide-antibody complexes in the TGN. Peptides released from these complexes underwent retrograde transport to the ER, as evidenced by the transfer of N-linked carbohydrate to their acceptor site. The KDEL-tagged glycopeptides (approximately 10% of the endocytosed load) behaved like endogenous ER residents: they stayed intracellular, and their oligosaccharide side chains remained sensitive to endoglycosidase H. An option thus exists to extract ER residents even at the most distant pole of the Golgi stack, suggesting that sorting of resident from exported ER proteins may occur in a multistage process akin to fractional distillation.

Transport of protein between cytoplasmic membranes of fused cells: correspondence to processes reconstituted in a cell-free system.

Mixed monolayers containing vesicular stomatitis virus-infected Chinese hamster ovary clone 15B cells (lacking UDP-N-acetylglucosamine transferase I, a Golgi enzyme) and uninfected wild-type Chinese hamster ovary cells were formed. Extensive cell fusion occurs after the monolayer is exposed to a pH of 5.0. The vesicular stomatitis virus encoded membrane glycoprotein (G protein) resident in the rough endoplasmic reticulum (labeled with [35S]methionine) or Golgi complex (labeled with [3H]palmitate) of 15B cells at the time of fusion can reach Golgi complexes from wild-type cells after fusion; G protein present in the plasma membrane cannot. Transfer to wild-type Golgi complexes is monitored by the conversion of G protein to an endoglycosidase H-resistant form upon arrival, and also demonstrated by immunofluorescence microscopy. G protein in the Golgi complex of the 15B cells at the time of fusion exhibits properties vis a vis its transfer to an exogenous Golgi population identical to those found earlier in a cell-free system (Fries, E., and J. E. Rothman. 1981. J. Cell Biol., 90: 697-704). Specifically, pulse-chase experiments using the in vivo fusion and in vitro assays reveal the same two populations of G protein in the Golgi complex. The first population, consisting of G protein molecules that have just received their fatty acid, can transfer to a second Golgi population in vivo and in vitro. The second population, entered by G protein approximately 5 min after its acylation, is unavailable for this transfer, in vivo and in vitro. Presumably, this second population consists of those G-protein molecules that had already been transferred between compartments within the 15B Golgi population, in an equivalent process before cell fusion or homogenization for in vitro assays. Evidently, the same compartment boundary in the Golgi complex is detected by these two measurements. The surprisingly facile process of glycoprotein transit between Golgi stacks that occurs in vivo may therefore be retained in vitro, providing a basis for the cell-free system.

Compartmentation of asparagine-linked oligosaccharide processing in the Golgi apparatus.

Golgi-associated processing of complex-type oligosaccharides linked to asparagine involves the sequential action of at least six enzymes. By equilibrium sucrose density gradient centrifugation of membranes from Chinese hamster ovary cells, we have partially resolved the set of four initial enzymes in the pathway (Mannosidase I, N-acetylglucosamine (GlcNAc) Transferase I, Mannosidase II, and GlcNAc Transferase II) from two later-acting activities (galactosyltransferase and sialyltransferase). In view of the recent demonstration that galactosyltransferase is restricted to the trans face of the Golgi complex in HeLa cells (Roth, J., and E.G. Berger, 1982, J. Cell Biol., 93:223-229), our results suggest that removal of mannose and attachment of peripheral N-acetylglucosamine may occur in some or all of the remaining cisternae on the cis side of the Golgi stack.

Intercompartmental transport in the Golgi complex is a dissociative process: facile transfer of membrane protein between two Golgi populations.

The transfer of the vesicular stomatitis virus-encoded glycoprotein (G protein) between Golgi populations in fused cells (Rothman, J. E., L. J. Urbani, and R. Brands. 1984. J. Cell Biol. 99:248-259) is exploited here to study and to help define the compartmental organization of the Golgi stack and to characterize the mechanism of intercompartmental transport. We find that G protein that has just received its peripheral N-acetylglucosamine in the Golgi complex of one cell is efficiently transferred to the Golgi complex of another cell to receive galactose (Gal). Remarkably, this transport occurs at the same rate between these two compartments whether they are present in the same or different Golgi populations. Therefore, a dissociative (presumably vesicular) transport step moves G protein from one part of the Golgi in which N-acetylglucosamine is added to another in which Gal is added. Minutes later, upon receiving Gal, the same G protein molecules are very poorly transferred to an exogenous Golgi population after cell fusion. Therefore, once this intercompartmental transfer has already taken place (before fusion), it cannot take place again (after fusion); i.e., transport across the compartment boundary in the Golgi complex that separates the sites of N-acetylglucosamine and Gal incorporation is a vectorial process. We conclude that transfers between Golgi cisternae occur by a stochastic process in which transport vesicles budding from cisternae dissociate, can diffuse away, and then attach to and fuse with the appropriate target cisterna residing in the same or in a different stack, based on a biochemical pairing after a random encounter. Under these circumstances, a transported protein would almost always randomize among stacks with each intercisternal transfer; it would not progress systematically through a single stack. Altogether, our studies define three sequential compartments in the Golgi stack.

A novel 115-kD peripheral membrane protein is required for intercisternal transport in the Golgi stack.

We have used an in vitro Golgi protein transport assay dependent on high molecular weight (greater than 100 kD) cytosolic and/or peripheral membrane proteins to study the requirements for transport from the cis- to the medial-compartment. Fractionation of this system indicates that, besides the NEM-sensitive fusion protein (NSF) and the soluble NSF attachment protein (SNAP), at least three high molecular weight protein fractions from bovine liver cytosol are required. The activity from one of these fractions was purified using an assay that included the second and third fractions in a crude state. The result is a protein of 115-kD subunit molecular mass, which we term p115. Immunodepletion of the 115-kD protein from a purified preparation with mAbs removes activity. Peptide sequence analysis of tryptic peptides indicates that p115 is a "novel" protein that has not been described previously. Gel filtration and sedimentation analysis indicate that, in its native state, p115 is a nonglobular homo-oligomer. p115 is present on purified Golgi membranes and can be extracted with high salt concentration or alkaline pH, indicating that it is peripherally associated with the membrane. Indirect immunofluorescence indicates that p115 is associated with the Golgi apparatus in situ.

Release of clathrin from coated vesicles dependent upon a nucleoside triphosphate and a cytosol fraction.

Calf-brain coated vesicles were incubated with ATP and a cytosol fraction. As much as 90% of the clathrin was selectively released within 10 min at 37 degrees C without detectable proteolysis. This uncoating process required the presence of both ATP and cytosol. Empty cages of clathrin could also be dissociated in a similar manner. A nonhydrolyzable analogue, 5'-adenylylimidodiphosphate (AMP-PNP), would not substitute for ATP. Clathrin was dissociated from coats in a form unable to reassemble into cages under standard conditions. These reactions may reflect a segment of a clathrin-coated vesicle cycle in which coats are removed from vesicles after budding.

Two distinct populations of ARF bound to Golgi membranes.

ADP-ribosylation factor (ARF) is a small molecular weight GTP-binding protein (20 kD) and has been implicated in vesicular protein transport. The guanine nucleotide, bound to ARF protein is believed to modulate the activity of ARF but the mechanism of action remains elusive. We have previously reported that ARF binds to Golgi membranes after Brefeldin A-sensitive nucleotide exchange of ARF-bound GDP for GTP gamma S. Here we report that treatment with phosphatidylcholine liposomes effectively removed 40-60% of ARF bound to Golgi membranes with nonhydrolyzable GTP, presumably by competing for binding of activated ARF to lipid bilayers. This revealed the presence of two different pools of ARF on Golgi membranes. Whereas total ARF binding did not appear to be saturable, the liposome-resistant pool is saturable suggesting that this pool of ARF is stabilized by interaction with a Golgi membrane-component. We propose that activation of ARF by a guanine nucleotide-exchange protein results in association of myristoylated ARF GTP with the lipid bilayer of the Golgi apparatus. Once associated with the membrane, activated ARF can diffuse freely to associate stably with a target protein or possibly can be inactivated by a GTPase activating protein (GAP) activity.

Transport of the membrane glycoprotein of vesicular stomatitis virus to the cell surface in two stages by clathrin-coated vesicles.

The G protein of vesicular stomatitis virus is a transmembrane glycoprotein that is transported from its site of synthesis in the rough endoplasmic reticulum to the plasma membrane via the Golgi apparatus. Pulse-chase experiments suggest that G is transported to the cell surface in two successive waves of clathrin-coated vesicles. The oligosaccharides of G protein carried in the early wave are of the "high-mannose" (G1) form, whereas the oligosaccharides in the second, later wave are of the mature "complex" (G2) form. the early wave is therefore proposed to correspond to transport of G in coated vesicles from the endoplasmic reticulum to the Golgi apparatus, and the succeeding wave to transport from the Golgi apparatus to the plasma membrane. The G1- and G2-containing coated vesicles appear to be structurally distinct, as judged by their differential precipitation by anticoated vesicle serum.

Binding of an N-ethylmaleimide-sensitive fusion protein to Golgi membranes requires both a soluble protein(s) and an integral membrane receptor.

An N-ethylmaleimide (NEM)-sensitive fusion protein (NSF) has recently been purified on the basis of its ability to restore transport to NEM-inactivated Golgi membranes in a cell-free transport system. NSF is a peripheral membrane protein required for the fusion of transport vesicles. We now report the existence of two novel components that together bind NSF to Golgi membranes in a saturable manner. These components were detected by examining the requirements for reassociation of purified NSF with Golgi membranes in vitro. One component is an integral membrane receptor that is heat sensitive, but resistant to Na2CO3 extraction and to all proteases tested. The second component is a cytosolic factor that is sensitive to both proteases and heat. This soluble NSF attachment protein (SNAP) is largely resistant to NEM and is further distinguished from NSF by chromatography. SNAP appears to act stoichiometrically in promoting a high-affinity interaction between NSF and the membrane receptor. Because NSF promotes vesicle fusion, it seems likely that these two new factors that allow NSF to bind to the membrane are also part of the fusion machinery.

Purification of coated vesicles by agarose gel electrophoresis.

We have applied agarose gel electrophoresis as a novel step in the purification of clathrin-coated vesicles. Preparations of coated vesicles obtained by sedimentation velocity and isopycnic centrifugation are resolved into two distinct fractions upon electrophoresis. The slower migrating fraction contains smooth vesicles, whereas the faster contains only coated vesicles and empty clathrin coats. The faster mobility of the coated vesicles is primarily caused by the acidic nature of clathrin. Coated vesicles from three different cell types have different mobilities. In each case, however, all of the major polypeptides previously attributed to coated vesicles comigrate with the now homogeneous particles, even though a powerful ATPase activity is completely removed.

Fatty acylation promotes fusion of transport vesicles with Golgi cisternae.

Two different methods, stimulation of transport by fatty acyl-coenzyme A (CoA) and inhibition of transport by a nonhydrolyzable analogue of palmitoyl-CoA, reveal that fatty acylation is required to promote fusion of transport vesicles with Golgi cisternae. Specifically, fatty acyl-CoA is needed after the attachment of coated vesicles and subsequent uncoating of the vesicles, and after the binding of the NEM-sensitive fusion protein (NSF) to the membranes, but before the actual fusion event. We therefore suggest that an acylated transport component participates, directly or indirectly, in membrane fusion.

Dissociation of clathrin coats coupled to the hydrolysis of ATP: role of an uncoating ATPase.

ATP hydrolysis was used to power the enzymatic release of clathrin from coated vesicles. The 70,000-mol-wt protein, purified on the basis of its ATP-dependent ability to disassemble clathrin cages, was found to possess a clathrin-dependent ATPase activity. Hydrolysis was specific for ATP; neither dATP nor other ribonucleotide triphosphates would either substitute for ATP or inhibit the hydrolysis of ATP in the presence of clathrin cages. The ATPase activity is elicited by clathrin in the form of assembled cages, but not by clathrin trimers, the product of cage disassembly. The 70,000-mol-wt polypeptide, but not clathrin, was labeled by ATP in photochemical cross-linking, indicating that the hydrolytic site for ATP resides on the uncoating protein. Conditions of low pH or high magnesium concentration uncouple ATP hydrolysis from clathrin release, as ATP is hydrolyzed although essentially no clathrin is released. This suggests that the recognition event triggering clathrin-dependent ATP hydrolysis occurs in the absence of clathrin release, and presumably precedes such release.

An enzyme that removes clathrin coats: purification of an uncoating ATPase.

Uncoating ATPase, an abundant 70,000-mol-wt polypeptide mediating the ATP-dependent dissociation of clathrin from coated vesicles and empty clathrin cages, has been purified to virtual homogeneity from calf brain cytosol. Uncoating protein is present in cells in amounts roughly stoichiometric with clathrin. This enzyme is isolated as a mixture of monomers and dimers, both forms being active. ATP can support protein-facilitated dissociation of clathrin at micromolar levels; all other ribotriphosphates as well as deoxy-ATP are inactive. The clathrin that is released from cages consists of trimers (triskelions) in a stoichiometric complex with uncoating ATPase. These complexes with clathrin have little tendency to self-associate at neutral pH, and at acidic pH they interfere with the assembly of free clathrin. The possible existence and function of these complexes as clathrin carriers in cells would explain why uncoating protein is made in quantities equivalent to clathrin.

ADP-ribosylation factor and coatomer couple fusion to vesicle budding.

The coat proteins required for budding COP-coated vesicles from Golgi membranes, coatomer and ADP-ribosylation factor (ARF) protein, are shown to be required to reconstitute the orderly process of transport between Golgi cisternae in which fusion of transport vesicles begins only after budding ends. When either coat protein is omitted, fusion is uncoupled from budding-donor and acceptor compartments pair directly without an intervening vesicle. Coupling may therefore results from the sequestration of fusogenic membrane proteins into assembling coated vesicles that are only exposed when the coat is removed after budding is complete. This mechanism of coupling explains the phenomenon of "retrograde transport" triggered by uncouplers such as the drug brefeldin A.

Exclusion of golgi residents from transport vesicles budding from Golgi cisternae in intact cells.

A central feature of cisternal progression/maturation models for anterograde transport across the Golgi stack is the requirement that the entire population of steady-state residents of this organelle be continuously transported backward to earlier cisternae to avoid loss of these residents as the membrane of the oldest (trans-most) cisterna departs the stack. For this to occur, resident proteins must be packaged into retrograde-directed transport vesicles, and to occur at the rate of anterograde transport, resident proteins must be present in vesicles at a higher concentration than in cisternal membranes. We have tested this prediction by localizing two steady-state residents of medial Golgi cisternae (mannosidase II and N-acetylglucosaminyl transferase I) at the electron microscopic level in intact cells. In both cases, these abundant cisternal constituents were strongly excluded from buds and vesicles. This result suggests that cisternal progression takes place substantially more slowly than most protein transport and therefore is unlikely to be the predominant mechanism of anterograde movement.

Hydrolysis of bound GTP by ARF protein triggers uncoating of Golgi-derived COP-coated vesicles.

The cycle of nucleotide exchange and hydrolysis by a small GTP-binding protein, ADP-ribosylation factor (ARF), helps to provide vectoriality to vesicle transport. Coat assembly is triggered when ARF binds GTP, initiating transport vesicle budding, and coat disassembly is triggered when ARF hydrolyzes GTP, allowing the uncoated vesicle to fuse.