Role of Dynamin in the Formation of Transport Vesicles from the Trans-Golgi Network

See allHide authors and affiliations

Science  23 Jan 1998:
Vol. 279, Issue 5350, pp. 573-577
DOI: 10.1126/science.279.5350.573


Dynamin guanosine triphosphatases support the scission of clathrin-coated vesicles from the plasmalemma during endocytosis. By fluorescence microscopy of cultured rat hepatocytes, a green fluorescent protein–dynamin II fusion protein localized with clathrin-coated vesicles at the Golgi complex. A cell-free assay was utilized to demonstrate the role of dynamin in vesicle formation at the trans-Golgi. Addition of peptide-specific anti-dynamin antibodies to the assay mixture inhibited both constitutive exocytic and clathrin-coated vesicle formation. Immunodepletion of dynamin proteins also inhibited vesicle formation, and budding efficiency was restored upon readdition of purified dynamin. These data suggest that dynamin participates in the formation of distinct transport vesicles from the trans-Golgi network.

The dynamins comprise a family of 100-kD guanosine triphosphatases that have been implicated in severing clathrin-coated invaginations from the plasma membrane based on the shibire ts1 mutant of Drosophila melanogaster (1) and studies of a mutant dynamin isoform overexpressed in mammalian epithelial cells (2-4). Originally dynamin was thought to be a neuronal specific protein. However, three distinct dynamin genes recently have been identified in mammals: dynamin I (Dyn1) is expressed exclusively in neurons (5, 6); dynamin II (Dyn2) is found in all tissues (6); and dynamin III (Dyn3) is restricted to the testis, the brain, and the lung (7). Each dynamin gene encodes at least four alternatively spliced isoforms (8). Whether all these dynamin gene products function solely at the plasma membrane or also mediate other vesicle scission events at distinct cellular sites is unknown (8). Recently, a dynamin has been localized to the Golgi complex of mammalian cells by biochemical, immunological, and morphological techniques (9,10). To provide additional evidence supporting the Golgi localization of a specific dynamin isoform, we linked Dyn2 (spliced form “aa”) to green fluorescent protein (GFP) and expressed it in a rat hepatocyte cell line. Subsequently, its distribution was followed in vivo by fluorescence microscopy (11-13) (Fig.1). In parallel, untransfected cells were labeled with a Dyn2-specific antibody and a Pan-dynamin antibody (MC63), which recognizes a conserved region of the dynamins (14). A prominent punctate staining at the plasma membrane and the Golgi region was observed with both experimental protocols [GFP-Dyn2 in vivo (Fig. 1, B and D) and endogenous Dyn2 after fixation and immunolocalization (Fig. 1, A, B′, C, and D′)]. Thus, the transfection process did not alter the distribution of the endogenous Dyn2 compared with untransfected cells. Importantly, the overlap between the two images (Fig. 1) suggests that a Dyn2 isoform is localized to vesicles at both the plasma membrane and the Golgi complex.

Figure 1

A GFP-Dyn2 protein expressed in cultured hepatocytes localizes to the plasma membrane and perinuclear vesicles. (A and C) Immunofluorescence microscopy of nontransfected cultured clone 9 cells stained with affinity-purified antibodies specific for Dyn2 (A) or the Pan-dynamin antibody MC63 (C), which recognizes all dynamin isoforms [for details of antibodies see (14)]. Numerous punctate vesicles are seen at the cell periphery near the plasma membrane (arrowheads) and around the nucleus (arrows). (B and D) Clone 9 cells transfected with a GFP-Dyn2 construct and viewed in vivo with a cooled CCD video camera. The GFP-Dyn2 localizes to vesicles at the plasma membrane and in the Golgi region, a pattern similar to the localization obtained with antibodies against dynamin in untransfected cells (A and C). (B′ and D′) The GFP-Dyn2–expressing cells stained with the Dyn2-specific (B′) or MC63 (D′) antibodies show colocalization of overexpressed and endogenous forms of dynamin. Bars = 10 μm.

To define more precisely the localization of Dyn2 at the Golgi region, cells expressing GFP-Dyn2 were labeled with antibodies to either clathrin or an antigen of the trans-Golgi, TGN38 (14). A nearly identical fluorescence pattern was observed between GFP-Dyn2 and the clathrin labeling (Fig.2, A to A"). The significant overlap between GFP-Dyn2 and clathrin at the Golgi region suggests that Dyn2 may associate with nascent clathrin-coated buds forming at the trans-Golgi network (TGN). These images also demonstrate a colocalization of GFP-Dyn2 and clathrin at the plasma membrane and provide an important control, as others have shown a colocalization of dynamin and clathrin by double labeling and conventional immunofluorescence microscopy (3, 15). The localizations of TGN38 and GFP-Dyn2 were nearly identical in the central region of the cells; however, the overlap on peripheral vesicles was significantly less than with clathrin (Fig. 2, B to B"). Thus, at least one spliced variant of dynamin (Dyn2aa) associates with clathrin-coated vesicles at the TGN in these cells and probably is more widely distributed than the clathrin-coated vesicles in the TGN. A similar distribution was observed in cells that expressed the spliced variant Dyn2ba (16). Therefore, Dyn2 may function in vesicle budding events at the TGN.

Figure 2

GFP-Dyn2 associates with clathrin-coated vesicles at both the plasma membrane and the TGN. Double-label fluorescence microscopy of clone 9 cells expressing GFP-Dyn2. (A andB) GFP-Dyn2 is localized to punctate vesicles at the plasma membrane and in the Golgi region. Immunolocalization of the same cell with a monoclonal clathrin antibody (A′) shows a nearly identical localization with the GFP-Dyn2 at the plasma membrane and the Golgi complex. Immunolocalization with an antibody to TGN38 (B′) shows a nearly identical localization of the GFP-Dyn2 with the prominently labeled Golgi. Bars = 10 μm.

To support the morphological localization of Dyn2 at the Golgi complex, we conducted complementary biochemical studies. We determined the amount of dynamin binding to a highly enriched rat liver Golgi fraction (17) under three conditions, all containing added cytosol (100,000g liver supernatant): alone in the absence of adenosine triphosphate (ATP), with added ATP, and with an ATP regeneration system either without or with guanosine 5′-O-(2-thiodiphosphate) (GTP-γ-S) (18). After the incubation, Golgi fractions were separated from cytosol and subjected to SDS–polyacrylamide gel electrophoresis (SDS-PAGE) and immunoblot analysis with the Pan-dynamin antibody MC63. When dynamin was incubated with cytosol alone, a minimal amount of it was found associated with the Golgi fraction (Fig.3A). When ATP and an ATP regeneration system were added to the incubation mixture, a 3-fold increase in dynamin binding was observed and the addition of GTP-γ-S resulted in a 10-fold increase in dynamin binding. These experiments support the morphological observations that dynamin associates with the Golgi and demonstrate that this association is energy-dependent.

Figure 3

Antibodies to dynamin inhibit vesicle formation from the Golgi complex. (A) Dynamin binding to a highly enriched Golgi fraction. The SGF1 was incubated with a cytosolic fraction in the absence of ATP (lane 1); in the presence of ATP and an ATP-regenerating system (lane 2); or in the presence of ATP, an ATP-regenerating system, and 10 μM GTP-γ-S (lane 3) at 37°C for 15 min. The reaction mixture was centrifuged through a sucrose cushion and the Golgi pellet was immunoblotted and analyzed for the presence of dynamin. The dynamin-immunoreactive band at 100 kD is shown in the upper panel, and quantitation of the amount of dynamin bound (PhosphorImager units) is shown in the lower panel. Note the increase in the association of dynamin with the Golgi membranes in the presence of either ATP or GTP-γ-S. Each assay was carried out in triplicate and the standard error is plotted. (B) Antibodies to dynamin inhibit formation of both the pIgA-R–containing exocytic and clathrin-coated vesicles. (Top) Domains of Dyn2 are diagrammed and include three GTP-binding consensus sequence elements in the NH2-terminus, a pleckstrin homology (PH) domain in the COOH-terminal region, and a proline-rich domain at the COOH-terminus. The regions used to generate the polyclonal MC60, MC63, and Dyn2-specific antibodies are noted. The MC60 and MC63 epitopes are present in all dynamin isoforms, whereas the Dyn2-specific epitope is unique to Dyn2. (Middle and bottom) Cell-free assays of vesicle budding from the TGN were carried out in the presence of increasing amounts (0 to 16 μg) of affinity-purified antibodies. The budding efficiencies of the pIgA-R–containing vesicles (middle) and clathrin-coated vesicles (bottom) are plotted against the antibody concentration. Control antibodies to the pIgA-R (not shown) and kinesin heavy chain (MC44) have no effect on vesicle budding, whereas the MC63 and Dyn2-specific antibodies were strongly inhibitory. Antibody MC60 induced a more modest inhibition. Antibodies were preincubated with the cytosolic fraction for 30 min on ice before they were added to the cell-free assay mixture. The antibody used in each reaction is listed above the corresponding graph: MC44 (against kinesin), MC60, MC63, and DYN 2.

To test whether Dyn2 acts in vesicle formation from the TGN as it does at the plasma membrane, we used a cell-free assay of vesicle formation from the TGN (19, 20). This assay measures the formation of both polymeric immunoglobulin A receptor (pIgA-R)–containing (constitutive) exocytic vesicles and clathrin-coated vesicles from the TGN by using a purified rat liver stacked Golgi fraction. Consistent with previous work (19), in the complete system the budding efficiency of the mature form of the pIgA-R was ∼70%, and in the absence of cytosol had a background of less than 5% (Fig. 4C). In the absence of cytosol, the background of clathrin-coated vesicle budding also was less than 5% (Fig. 4D). First, the effect of dynamin antibodies on vesicle formation was tested. Three different affinity-purified peptide antibodies against conserved or isoform-specific domains of the dynamins (Fig. 3B) [see (14) for discussion of antibody specificity] were added to the assay mixture in increasing concentrations (0 to 16 μg). Antibodies against a conserved region of the kinesin heavy chain (MC44) or vesicular stomatitis virus G protein (P5D4) (17) were used at the same concentrations and served as controls. Although control antibodies did not inhibit the budding of pIgA-R–containing or clathrin-coated vesicles at any of the concentrations tested, the formation of both vesicle populations was inhibited significantly by the MC63 and Dyn2-specific antibodies (Fig.3B). Inhibition with MC63 was apparent with 1 to 2 μg of antibody and nearly complete inhibition was achieved with ∼8 μg of antibody. The Dyn2-specific antibody was slightly less efficient at inhibiting the formation of both vesicle populations and the MC60 antibody only partially inhibited vesicle budding. The MC63 and Dyn2-specific antibodies showed the highest immunoreactivities by immunoblot analysis and immunofluorescence microscopy (21), which may explain the differences observed between antibodies in these functional inhibition studies. MC63 specifically labels the Golgi by immunofluorescence microscopy and efficiently immunoisolates Golgi components (9). Because antibodies against two distinct dynamin domains effectively inhibited the formation of both vesicle populations, whereas control antibodies had no effect, these reagents demonstrate a specific block of dynamin function at the TGN.

Figure 4

Vesicle formation from the TGN is dependent on dynamin and cytosolic factors. (A) Immunodepletion of dynamin from rat liver cytosol using immunoaffinity columns. A rat liver cytosolic fraction was passed over two successive antibody columns and the nonbound fractions were collected and concentrated. Fractions of the starting (SM) and immunodepleted (DPL) cytosolic fraction were resolved by SDS-PAGE and stained with Coomassie blue (top) or transferred to nitrocellulose filters and blotted with the Pan-dynamin antibody MC63 (bottom). (B) Generation of an enriched dynamin fraction from rat brain for use in restoration experiments. A rat brain dynamin-enriched fraction (elute) was obtained by ion-exchange column chromatography as described (9). Proteins were resolved by SDS-PAGE and either stained with Coomassie blue (top) or transferred to nitrocellulose filters for immunoblot analysis (bottom). Other fractions shown are a high-speed supernatant (HSS) and the void volumes collected from DEAE (DE) and phosphocellulose (PC) anion-exchange columns. (C andD) Dynamin-dependent vesicle budding from the TGN by a reconstituted cell-free assay. Cell-free assays to measure budding of pIgA-R–containing (C) and clathrin-coated (D) vesicles were carried out under the following conditions: in the absence of ATP and cytosol (lane 1); in the presence of ATP and cytosol (lane 2); with a dynamin-depleted cytosolic fraction (lane 3); with a dynamin-depleted cytosolic fraction plus increasing concentrations of a dynamin-enriched fraction (lanes 4 to 7); and with the dynamin-enriched fraction alone (lane 8). Lanes 4 to 7 contain 5, 10, 25, and 50 μg of dynamin-enriched fraction, respectively. Whereas no vesicle budding occurred with dynamin-depleted cytosol, formation of both pIgA-R–containing (C) and clathrin-coated (D) vesicles was restored to near control levels when the dynamin fraction was added back to the reaction mixture. Importantly, the dynamin preparation alone did not support vesicle budding. Each assay was carried out in triplicate and the standard error is plotted. Immunoblot analysis of a representative experiment showing the 116-kD form of the pIgA-R (C) and the 180-kD clathrin heavy chain (D) is shown above each bar graph.

To provide an additional test for the participation of dynamin in vesicle formation from the TGN, the cell-free assay was carried out either with cytosol immunodepleted of dynamin proteins or after readdition of a dynamin-enriched preparation to the depleted cytosol (Fig. 4). Dynamin was depleted from rat liver cytosol by using two immunoaffinity columns made from the MC63 and Dyn2-specific antibodies (22). By SDS-PAGE and Coomassie blue staining, we found that the starting rat liver cytosol and depleted cytosol did not differ significantly (Fig. 4A). In contrast, immunoblot analyses of these fractions clearly showed a complete depletion of dynamin from the cytosol (Fig. 4A). For reconstitution of the depleted cytosol, a dynamin-enriched fraction was prepared from rat brain by column chromatography (23) (Fig. 4B). This purification procedure was utilized because the conditions used to elute dynamin proteins from the immunoaffinity column inactivated the protein. Standard chromatographic methods provided a highly enriched dynamin preparation that contained all three dynamin isoforms (Dyn1, -2, and -3). Significantly, depletion of dynamin proteins from the cytosol totally inhibited both exocytic and clathrin-coated vesicle formation (Fig. 4, C and D). Readdition of the dynamin-enriched fraction to the depleted cytosol restored the budding activity of both vesicle populations in a concentration-dependent manner, with ∼25 μg sufficient to restore budding to near control levels. Addition of 50 μg of the enriched dynamin fraction alone (without the depleted cytosol) could not restore any budding activity, which supports the observations indicating that other cytosolic components are required for budding (24).

The immunolocalization studies (9, 10) combined with the GFP-Dyn2 localization and the functional experiments presented here demonstrate a requirement for dynamin proteins in the formation of two different vesicle populations from the TGN. Although dynamin rings around the necks of forming TGN vesicles have not been reported, these structures have been observed only at the plasma membrane in neurons of the shibire ts1 flies at the restrictive temperature (8). Rings at the plasma membrane have not been resolved in epithelial tissues of the same flies or in any mammalian cells under physiological conditions (without GTP-γ-S). Although the precise role of dynamin isoforms in the formation of vesicles from the TGN is undefined, it is attractive to speculate that dynamin participates in the scission of nascent vesicles (9, 25). A previous study in which a mutant Dyn1 protein was overexpressed in epithelial cells did not find an inhibition of transport of newly synthesized hydrolytic enzymes from the Golgi to the lysosome (3), suggesting that Dyn1 does not act at the TGN. On the basis of our observations of a preferential association of the Dyn2 protein with the Golgi, it is not surprising that an overexpressed mutant Dyn1 isoform would have little effect on the formation of vesicles from the TGN. We assume that Dyn2 is the dynamin acting at the TGN for several reasons. First, Dyn2 is currently the only dynamin known to be expressed in most epithelial cells (6). Second, an antibody that is specific for Dyn2 localized this isoform to the Golgi complex by immunofluorescence microscopy (Fig. 1) and inhibited vesicle formation from the TGN in the cell-free assay (Fig. 3, C and D). Finally, a GFP-Dyn2 protein expressed in cultured cells localized with both clathrin and TGN38 at the Golgi as determined by immunofluorescence microscopy (Fig. 2). The other known dynamin proteins, such as the neuronally expressed Dyn1 and Dyn3, when coupled to GFP, show a markedly different distribution from each other and from Dyn2 in the rat hepatocyte cell line (26). Although both of the Dyn2 spliced variants that we have expressed localized to the Golgi, only one of the expressed Dyn1 or Dyn3 proteins showed a modest Golgi association. Whether the modest affinity of a Dyn1 spliced form for the Golgi reveals its true localization in neuronal cells or represents a nonspecific interaction based on partial sequence homology to the Dyn2 forms is unclear. Determining the roles of the many dynamin isoforms and their spliced variants and defining the molecular interactions of Dyn2 at the TGN remain as challenges for the future.

  • * To whom correspondence should be addressed. E-mail: mcniven.mark{at}


View Abstract

Navigate This Article