The Golgi apparatus receives a varied mixture of newly synthesized proteins and lipids from the endoplasmic reticulum (ER) and sends them to their correct destination in the cell. These molecules share a common pathway through the Golgi apparatus, passing, in sequence, through an ordered array of compartments, each capable of carrying out a specific set of covalent post-translational modifications. In the last compartment, they are separated from each other in preparation for final delivery.
1. Morphology
First described in 1898 by the Italian anatomist Camillo Golgi (1843-1926) using a silver impregnation technique (Fig. 1a) (1), the Golgi apparatus has the same basic structure in all eukaryotic cells, comprising a stack of closely apposed and flattened cisternae (Figs. 1d and 1e). Cisternae are often cup-shaped, especially toward the trans side (2, 3), and there are typically three to six cisternae in the stack, although up to 40 have been reported (4). Each cisterna is about 1 |im in diameter and comprises a core region, involved in stacking and containing resident enzymes, and a fenestrated rim from which transport vesicles bud and with which they fuse (5). In plants and fungi, there are multiple copies of the Golgi dispersed throughout the cytoplasm (Fig. 1d) (3). In animal cells these stacks are linked laterally, forming a bifurcating, ribbon-like structure (Figs. 1e and 1f) (6, 7), which forms a compact reticulum most often found in the pericentriolar region of the cell, adjacent to the cell nucleus (Fig. 1a and 1c) (8).
Figure 1. Different views of the Golgi apparatus. (a) The original "appareil reticulaire" in a Purkinje cell, described by C 1898 (1) and revealed by his silver impregnation technique. (b, c) The Golgi reticulum in a living HeLa cell revealed by Golgi enzyme with the green fluorescent protein (138). (b) Phase image. (c) Fluorescence image. (d) Longitudinal sectio yeast cell (Schizosaccharomycespombe) fixed with permanganate. Note the dispersed Golgi stacks (arrowheads). (e) Epi the Golgi region of a HeLa cell showing two stacks in the Golgi ribbon. (f) Three-dimensional reconstruction of part of t animal cells. Sections of two Golgi stacks (left and right) are shown linked by tubules and networks connecting equivale adjacent stacks.
Each face of the stack is apposed to a complex and extensive tubular network, best characterized in animal cells (and shown schematically in Fig. 1f). At the cis or entry face is the cis-Golgi network (CGN) (9), and at the trans or exit face is the trans-Golgi network (TGN) (10, 11). An intermediate compartment is interposed between the ER and the CGN, which is thought by many to ferry newly synthesized cargo molecules from the exit sites on the ER (the transitional element region) (12) to the CGN (13).
The Golgi apparatus is embedded in a matrix called the "zone of exclusion" because other cytoplasmic structures down to the size of ribosomes are excluded (14). Candidate components of this matrix have been identified, many of which have sequences predicting rod-like, fibrous proteins (15).
2. Biogenesis
The growth and division of the Golgi apparatus has been little studied, with the exception of the partitioning of Golgi membranes during mitosis in animal cells (16). The Golgi ribbon is converted during the early stages of mitosis into clusters of small vesicles and tubules, as well as free vesicles, a process that is thought to aid the partitioning process (17, 18). Cell-free assays (19, 20) have provided a molecular explanation of at least part of this process (21, 22).
3. Compartmentation
The Golgi apparatus comprises an ordered array of compartments through which newly synthesized proteins pass in sequence. The CGN is the entry point and is thought to be the last quality-control step on the pathway (23, 24). Properly folded cargo proteins proceed on to the stack. Misfolded proteins are returned to the ER for further rounds of folding (25) or degradation (26).
A large number of covalent modifications are carried out by the Golgi apparatus, including glycosylation, acylation, phosphorylation, sulfation, and proteolytic cleavage (27-30). N-Glycosylation is the best characterized and comprises a sequence of processing reactions that mark the passage of proteins through the Golgi stack (27, 31). The trimming of the high-mannose oligosaccharides, which are attached co-translationally in the ER, is completed in the early part of the Golgi apparatus. Construction of complex oligosaccharides is initiated in the middle part of the Golgi stack and completed in the late part. The enzymes that carry out these steps are present in two or more adjacent cisternae, generating an overlapping distribution that might help to ensure more complete glycosylation (32). Construction of complex oligosaccharides is completed in the TGN, which also functions to sort proteins and lipids. Those destined for lysosomes and secretory granules are packaged separately from those destined for the cell surface (see text below).
O-linkedoligosaccharides (see O-Glycosylation) and glycolipids (33) are also assembled in a stepwise fashion as they pass, in sequence, through the stacked cisternae. The precise locations at which these steps occur must await cloning and localization of the enzymes involved.
4. Transport Through the Golgi Apparatus
Several models have been proposed to explain the ordered transport of proteins (and some lipids) through the Golgi apparatus. The two extremes are cisternal maturation (34) and vesicle-mediated transport (12), illustrated in Figure 2.
Figure 2. Models for transport through the Golgi stack. (a) Cisternal maturation envisages a dynamic stack of cisternae i of cargo molecules, matures into the next one through the retrograde transport of Golgi enzymes. (b) Vesicle-mediated ti cisternae through which a mixture of cargo molecules passes, in sequence, transported from cisterna to cisterna by vesicl.
1. Cisternal maturation postulates a dynamic stack of cisternae. The cis-most cisterna is assembled by the fusion of transport vesicles from the ER, carrying newly synthesized cargo. Golgi processing enzymes are then delivered back from the next cisterna in the stack, which in turn receives enzymes from the next, and so on throughout the stack. In this way, each cisterna matures into the next one.
2. Vesicle-mediated transport postulates a stable stack of cisternae containing a fixed and ordered array of processing enzymes. The cargo is delivered to each cisterna in turn by vesicles that bud from one cisterna and fuse with the next in the stack.
There is still controversy as to whether either model, or a hybrid one, is correct. There is, however, general agreement that the vesicles mediating these pathways are COPI (Coat protomer complex I)-coated vesicles (35).
4.1. COPI-Mediated Budding
Assembly of the coat is thought to be initiated by the GTPase ARF (ADP-ribosylation factor) (36, 37) (Fig. 3). ARF exists in the cytoplasm bound to GDP, which is exchanged for GTP upon binding to Golgi membranes. This is catalyzed by a GEF (guanine nucleotide exchange factor) (38) that is sensitive to brefeldin A (39, 40). Exchange exposes a myristoyl group at the N-terminus of ARF, which aids Golgi binding (41). ARF then recruits coatomer , a complex of seven polypeptides (42, 43). This is either direct, through protein-protein interactions (44), or indirect, through the activation of phospholipase D by ARF, which generates phosphatidic acid to which the coatomer can bind (45). Stepwise binding of coatomer complexes is thought to cause incremental deformation of the membrane, generating a coated bud like that thought to occur for clathrin-coated vesicles (46). Scission generates a completed COPI vesicle and requires acyl-CoA, but it is otherwise uncharacterized (47). A GAP (GTPase activating protein) catalyzes the hydrolysis of the GTP bound to ARF (48), leading to disassembly of the coat (49) and recycling of the complexes for further rounds of budding.
Figure 3. COPI-mediated budding of transport vesicles. Stepwise assembly of coat subunits (ARF and coatomer) generate a COPI vesicle. GTP hydrolysis releases the coat subunits for further rounds of budding, leaving an uncoated vesicle that can then fuse with the target membrane. Note that proteins such as v-SNAREs, which have to be incorporated into vesicles during budding, have been omitted for clarity.
COPI vesicles are also involved in retrograde transport from the Golgi to the ER (see text below) and perhaps also from the ER to the Golgi apparatus (50). COPI coats are found on the TGN but appear not to be involved in transport from this compartment.
There are other types of coated vesicle involved in transport from the ER to the cis-Golgi and from the trans-Golgi. Two have been characterized at the molecular level. At the cis-side of the Golgi apparatus, COPII coats mediate the budding of cargo-carrying vesicles from the ER (50, 51). On the trans side, clathrin coats package lysosomal and other proteins during transport to lysosomes or upon recycling from immature granules (see text below). Although the protein constituents of both these coats differ from COPI coats, many of the underlying principles of operation appear to be the same (52). Lipids other than phosphatidic acid, and lipid exchange proteins (53), have also been implicated in budding processes. Diacylglycerol, in particular, appears to play a central role (54), although it is unclear whether this role is structural or regulatory.
4.2. Vesicle Docking and Fusion
After budding, Golgi transport vesicles must dock and fuse with the correct compartment. The SNARE hypothesis was put forward to explain the specificity of these transport steps (55) (Fig. 4). This hypothesis postulates a v- (or vesicle) SNARE (SNAP receptor) on each type of transport vesicle that interacts specifically with the t- (or target) SNARE in the recipient membrane compartment. SNAREs were first identified at the presynaptic membrane with which synaptic vesicles fuse during neurotransmission (56-60). Using genetic and biochemical techniques, they have now been identified at many other vesicle-mediated steps in the cell (61).
Figure 4. Targeting and fusion of transport vesicles. Transport vesicles are tethered and then docked specifically using v process. In one model (illustrated), SNAPs mediate the binding of the NSF ATPase to the v-t SNARE pair, and hydrolys another model, SNAPs and NSF prime the SNAREs before they can interact with each other (139).
All but one of the SNAREs studied to date are type II membrane proteins, with most of their mass protruding into the cytoplasm (62); the exception has a Cys-Ala-Ala-X box at the C-terminus, so it is probably anchored by a long, hydrophobic prenyl group (63) (see Prenylation). Their sequences predict a coiled-coil structure suitable for forming paired complexes (64). Many, perhaps all, t-SNAREs are hetero-oligomers containing, in addition, a member of the SNAP25/sec9 family of proteins (65, 66). Transport vesicles can contain more than one type of v-SNARE, suggesting that multiple SNARE pairs might be used to improve the fidelity of vesicle targeting (63). Assembly of the SNARE pair is a highly regulated process. In some cases, the two membranes are initially brought together by the vesicle docking protein, p115, a process that might facilitate the efficiency with which cognate SNAREs find each other (67).
Some, perhaps all, t-SNAREs exist in an inactive form, complexed to a member of the sec1 family of proteins (68, 69), which is removed during the activation process (70). Activation of SNAREs is triggered by a member of the Ypt/rab family of small GTPases (70, 71), one or more of which is present at each vesicle-mediated transport step (72). They exist in two forms: (a) an inactive, GDP form, complexed with GDI (GDP dissociation inhibitor) in the cytoplasm and (b) an active, GTP form bound to the membrane by a prenyl group (73). Delivery of the GDP form to the correct membrane is mediated by a GDF (GDI displacement factor) (74), followed by exchange of GDP for GTP catalyzed by a GEF (75, 76). Hydrolysis of GTP by a GAP (77, 78) completes the cycle, permitting the ypt/rab to be extracted by GDI and recycled (79-81).
Formation of the cognate SNARE pair is thought to initiate the fusion process catalyzed by the ATPase NSF (N-ethylmaleimide-sensitive factor) (82). Binding of NSF to the SNARE pair is mediated by SNAPs (soluble NSF attachment proteins) (83) and leads to ATP hydrolysis, membrane fusion, and break-up of the SNARE pair (84, 85). The precise order of events is still disputed, especially the step at which NSF acts (86, 87). It is also unclear how these events lead to the physical merging of the two lipid bilayers.
More recently, the NSF-like ATPase p97 (88) has been implicated (together with NSF) in the rebuilding of the Golgi apparatus from fragments generated either by mitotic conditions (89) or by drug treatment (90). The fusion role played by p97 is distinct from that played by NSF, but the molecular mechanism is unknown.
5. Protein Sorting
Proteins with very different final locations within the cell are assembled in the ER and sorted in the Golgi apparatus (Fig. 5). Cargo moves to the TGN, where sorting mechanisms separate lysosomal and secretory granule proteins from each other and from proteins destined for the cell surface. ER and Golgi proteins have retention mechanisms that inhibit movement beyond the point in the exocytosis pathway at which they usually act. Those molecules that do escape are salvaged by retrieval mechanisms. Retrieval mechanisms also recycle membrane components of the transport machinery. All these mechanisms interpret signals on the proteins undergoing transport. These signals are most often short stretches of sequence that are either read directly by the sorting mechanism or are used to construct the sorting signal.
Figure 5. Protein sorting during transport through the Golgi apparatus. A varied mixture of proteins is assembled in the ] proteins are salvaged, mostly, but not exclusively, from the early part of the Golgi and are returned to the ER. Golgi enzy Cargo proteins move to the TGN, where they are separately packaged and sent to the plasma membrane, late endosomes.
![Protein sorting during transport through the Golgi apparatus. A varied mixture of proteins is assembled in the ] proteins are salvaged, mostly, but not exclusively, from the early part of the Golgi and are returned to the ER. Golgi enzy Cargo proteins move to the TGN, where they are separately packaged and sent to the plasma membrane, late endosomes.](http://what-when-how.com/wp-content/uploads/2011/05/tmp3630_thumb1.jpg "Protein sorting during transport through the Golgi apparatus. A varied mixture of proteins is assembled in the ] proteins are salvaged, mostly, but not exclusively, from the early part of the Golgi and are returned to the ER. Golgi enzy Cargo proteins move to the TGN, where they are separately packaged and sent to the plasma membrane, late endosomes.")
5.1. The Default Pathway
The pathway from the ER to the cell surface was originally designated the default pathway because signals did not appear to be required to transport cargo through the correct sequence of Golgi compartments to the cell surface (91). This is still true for many proteins and many compartmental steps, but there is increasing evidence that some surface cargo can be selectively concentrated, especially during export from the ER (92, 93).
Two types of protein that could be involved in concentrating cargo have been described. The first are lectin-like proteins—for example, ERGIC53 (94) and Vip36 (95), which are thought to bind the oligosaccharides attached to many cargo proteins. In the case of ERGIC53, the lectin-cargo complex would be concentrated in transport vesicles because the cytoplasmic tail binds to coat proteins (96). The second are the p24 family of spanning proteins (97, 98), the large lumenal domains of which are thought to bind to cargo molecules by unknown means (97), aided by ancillary proteins (99). This family of proteins is selectively incorporated into transport vesicles, because their cytoplasmic tails contain a dibasic motif that binds coat subunits (100, 101).
5.2. Sorting in the TGN
All cargo molecules share the same pathway up to and including the TGN. It is here that lysosomal (102) and secretory granule proteins (103) are separated from surface cargo. Many soluble lysosomal enzymes are tagged in the early part of the Golgi apparatus with a mannose-6-phosphate signal that is constructed from some of the terminal mannose residues on the high-mannose oligosaccharide by two enzymes that recognize protein features unique to lysosomal enzymes (31, 104). The tagged enzyme moves through the Golgi stack to the TGN, where it binds the large mannose-6-phosphate receptor (105). The complex is specifically incorporated into clathrin-coated vesicles, because multiple signals in the cytoplasmic tail of the receptor (106) are recognized by an adaptor (107, 108). The complex is then ferried to a prelysosomal compartment (109) where, eventually, the low pH separates the enzyme from the receptor. The enzyme is delivered to lysosomes, whereas the receptor is recycled for further rounds of transport. Some lysosomal membrane proteins are also ferried in clathrin-coated vesicles. The same adaptor recognizes a signal in the cytoplasmic tail (110).
As well as a direct route to lysosomes from the TGN, there is also an indirect route via the cell surface. For soluble enzymes, this is mediated by the small mannose-6-phosphate receptor (106).
Secretory granule proteins begin to condense in the TGN, probably because of the lower pH of this part of the Golgi apparatus and the presence of calcium ions (111). Condensation is a sorting mechanism, because during this process all other proteins are largely excluded, including, in some cells, the granule proteins destined for other types of secretory granule. The outside of the condensing granule core is thought to bind to sorting receptors, thereby enveloping the core with a membrane that buds to form an immature vacuole. Maturation proceeds by a process involving vacuole fusion and retrieval of excess membrane, until a mature secretory granule is formed (112).
In polarized cells, there is more than one surface domain, and signals exist to direct surface cargo to one or other of these domains. The signals used include the bound N-linked oligosaccharides (113) or sequences within the cytoplasmic tail (114, 115). GPI-anchored proteins are thought to form "rafts" that are selectively incorporated into vesicles bound for the apical membrane domain (116).
5.3. Retention in the Golgi Apparatus
Proteins destined to remain in the Golgi apparatus are also synthesized and assembled in the ER. They are transported together with other cargo to the Golgi apparatus, but must then stop in the correct cisterna(e). The retention signal that specifies location within the Golgi apparatus is contained within the membrane-spanning domain and flanking sequences (117) and is thought to inhibit further forward transport in one of two ways. The kin recognition model (118) postulates interaction with other Golgi proteins sharing the same cisterna, generating an oligomer too large to enter the forward-moving transport vesicles. The bilayer thickness model (119) exploits the increasing thickness of Golgi membrane across the stack caused by a gradient of cholesterol. The Golgi protein moves forward until the length of the spanning domain matches that of the bilayer. Different lengths would specify different locations within the stack. There is evidence both for and against each model, but they are not mutually exclusive, and a combination of the two could serve to locate the proteins more precisely in the stack (120).
5.4. Retrieval by Retrograde Transport
Two types of retrieval operate from the Golgi apparatus to the ER. The first is a salvage process, recovering proteins that stray beyond the point in the exocytic pathway at which they operate (121). The need for salvage gave rise to the distillation hypothesis, which could help explain the need for a Golgi stack (122). The second type of retrieval is recycling, mostly of components of the vesicle fusion machinery.
Salvage of ER proteins is mediated by short signals at the N- or C-terminus. Soluble proteins in the ER lumen use a C-terminal tetrapeptide (123), [KDEL in mammals, HDEL in budding yeast], whereas membrane proteins use either a double-lysine (type I proteins) (124) or a double-arginine (type II proteins) (125) motif (but see Ref. 126 ). Escaped lumenal proteins bind to the KDEL receptor (127), mostly located in the CGN (128), and this triggers return of the complex (129) in COPI vesicles (130) to the ER, where a change in pH is thought to effect release (131). Escaped membrane proteins with a double-lysine motif are also returned by COPI vesicles, because this motif binds directly to the coatomer component of the COPI coats (132, 133). The mechanism for returning ER proteins with a double-arginine motif is not known.
Salvage of Golgi proteins is less well understood, although there is now good evidence that it occurs (134, 135). Salvage is also an important part of secretory granule biogenesis. During maturation of immature granules, clathrin-coated vesicles recover and return proteins destined for other pathways (136).
Recycling is exemplified by the v-SNAREs. These are present in the target compartment after vesicle fusion and must be recycled to the preceding, donor compartment for further rounds of vesicle budding. The mechanism is not known, but some evidence suggests that they are not passively recycled. Instead, they direct the vesicle to the preceding compartment, in the same way that they directed the earlier vesicles to the target compartment. A t-SNARE would be needed in each of the two compartments linked by the vesicles, but only one v-SNARE. This parsimonious solution to recycling requires a mechanism to switch the specificity of the v-SNARE, depending on which of the two compartments is being targeted (137).
