Does clathrin pull the fission trigger?

Clathrin-mediated vesicle budding is essential for a variety of intracellular membrane transport reactions in the secretory and endocytic pathways (1, 2). This process can be dissected in several sequential steps that include the recruitment of the clathrin coat, membrane invagination and concomitant propagation of the clathrin lattice, constriction of the vesicle neck, and finally fission of the coated vesicle, followed by its rapid uncoating (1, 3, 4). Although clathrin and clathrin adaptors, assisted by other accessory factors, drive this reaction up to the stage of a deeply invaginated coated pit, a distinct factor, the GTPase dynamin, is thought to play a critical role in fission (5, 6). To provide new clues concerning all sites where clathrin functions in mammalian cell physiology and on the vesicular traffic pathways that are dependent on clathrin, Iversen et al. (7) have engineered a mammalian cell line in which expression of clathrin can be suppressed by an inducible antisense RNA. Unexpectedly, this study has suggested a role for clathrin heavy chain in the regulation of dynamin expression and function.

This study suggests a role for clathrin heavy chain in the regulation of dynamin expression and function.

Whereas the use of genetics has significantly advanced the understanding of clathrin function in unicellular organisms (8, 9), perturbation of clathrin-mediated transport in mammalian cells has so far relied on indirect approaches. A first approach consisted of the overexpression of dominant negative mutants of dynamin that block clathrin-mediated endocytosis by producing an arrest of the fission reaction (5, 6). However, several considerations complicate the interpretation of these experiments. First, it was discovered that dynamin can function in clathrin independent pathways of endocytosis, including internalization via caveolae (10–15). Second, it remains controversial whether dynamin is absolutely required in all clathrin-dependent transport reactions, for example in clathrin-mediated budding from the trans-Golgi network (6). Third, dominant dynamin mutants may have effects on cell function independent of endocytosis, such as on signaling pathways (16, 17) and actin dynamics (18, 19), and at least some of these effects may be isoform-specific. Thus, alternate approaches to interfere with clathrin and actin have been developed that capitalize on the elucidation of critical interactions required for clathrin recruitment and assembly. Overexpression of the hub domain of clathrin heavy chain blocks clathrin function by preventing, via a competitive mechanism, the assembly of triskelia, i.e., the unit elements of the clathrin lattice (20). Likewise, peptide fragments have become available that compete the interaction of the “foot” of the clathrin triskelion with the clathrin adaptors. This interaction involves binding of the NH₂-terminal region of clathrin heavy chain with short amino acid motifs, the so-called “clathrin boxes,” present in the adaptors and in a variety of clathrin accessory factors (2, 21, 22). Expression of protein fragments containing clathrin boxes, such as fragments of AP180, epsin, auxilin, or amphiphysin, have been successfully used to titrate out assembly competent triskelia (23–28). In these experiments, however, indirect effects caused by sequestration by unassembled clathrin of other factors, for example of the molecular chaperone Hsc70, cannot be ruled out.

To circumvent this problem, Iversen et al. have generated a BHK cell line that can inducibly lower the expression of clathrin heavy chain. As expected, they observed a progressive depression of clathrin mRNA and of clathrin levels in the cell. Correspondingly, a blockade of transferrin uptake and of trans-Golgi network-to-lysosome/endosome transport was observed, although, surprisingly, the onset of these defects preceded the occurrence of a significant decrease in the levels of clathrin. Strikingly, electron microscopy of antisense expressing cells revealed that the block of endocytosis correlated with a dramatic morphological phenotype. Numerous endocytic coated pits were observed that were connected to the plasma membrane by long narrow tubular necks decorated by regularly spaced dynamin-like rings. The presence of dynamin on these tubules was confirmed by immunogold labeling. Thus, the down-regulation of clathrin appears to produce a block in dynamin function.

The occurrence of endocytic pits with a collared neck was first observed in nerve terminals of shibire mutants of Drosophila after a shift to the restrictive temperature (29, 30). The striking temperature-sensitive paralytic defect of these flies was found to result from a block of synaptic vesicle recycling at the stage of deeply invaginated collared endocytic pits. When these observations were first made, the gene responsible for the shibire mutation was not known and the molecular identity of these collars remained elusive. Subsequently, the shibire gene was shown to encode the GTPase dynamin (31, 32), a protein with the intrinsic properties of assembling into higher-ordered structures, such as rings and spirals reminiscent of the shibire collars (Fig. 1) (6, 33). Moreover, electron microscopy analysis of endocytic intermediates generated from synaptic membranes in the presence of guanosine 5′-[γ-thio]triphosphate (GTPγS) revealed the presence of similar collars around the neck of clathrin coated pits and the enrichment of dynamin in these structures was confirmed by immunogold electron microscopy (34).

Figure 1.

Dynamin collars (arrows) observed in a variety of experimental systems. (A) Collared endocytic pits in nerve terminals of shibire mutants at the restrictive temperature (from ref. 30). (B–D) Dynamin collars around the stalks of endocytic buds on synaptic membranes incubated with brain cytosol, ATP, and GTPγS (from refs. 34 and 50). (E and F) Collared tubular endocytic structures in the cortical cytoplasm BHK cells expressing clathrin antisense RNA, as shown by Iversen et al. (7) in this issue of PNAS. These structures, as well as the bud into which they terminate, were shown to be positive for dynamin immunoreactivity by immunogold. Note their similarity of these tubules to the typical dynamin coated tubule shown in D. (G) Cryo-electron microscopy images of liposomes incubated with purified nucleotide-free dynamin. Arrows indicate cross-sectioned collars, and arrowheads indicate individual dynamin oligomers. The dynamin coat is visible also on nontubular liposomes, raising the possibility that some of the density on the vesicular portions of the endocytic structure shown in E and F may be accounted for by dynamin (from ref. 36).

Because GTPγS locks dynamin in its “GTP” state, the hypothesis was proposed that assembly of a dynamin ring, and its subsequent GTP hydrolysis-dependent conformational change, are critical steps required for the fission reaction (34, 35). This hypothesis received support form the demonstration that dynamin alone can bind and tubulate liposomes and then fragment them in the presence of GTP (36, 37). Further supporting this model, expression of GTPase-defective mutant of dynamin in nonneuronal cells induced the formation of long tubules decorated by dynamin containing periodic rings (39).

A function of dynamin as a mechanochemical enzyme contrasts with the general mechanism of action of classical regulatory GTPases. Typically, the “GTP” bound form of a GTPase represents the “on” state of the protein, whereas GTP hydrolysis functions as the “off” switch. Thus, an alternative model to explain dynamin's role in endocytosis has been considered and there is experimental evidence to support this possibility as well. This model proposes that a yet elusive interactor of dynamin-GTP is the effector of the fission reaction (40). So far, however, the only known interactor of the GTPase module of dynamin is the GED (GTPase effector domain) of dynamin itself, consistent with a critical role of oligomerization in dynamin function (40). An interesting possibility is that the actin cytoskeleton may be an effector, or one of the effectors, of dynamin. The binding of dynamin's proline-rich COOH-terminal domain (PRD) to a variety of proteins that directly or indirectly regulate the actin cytoskeleton points to a potential role of actin downstream to dynamin (41–43), though these proteins may also act upstream of dynamin. This possibility was suggested by cryo-EM studies demonstrating that a mutant dynamin lacking the PRD can constrict lipid tubules even in the presence of nonhydrolyzable analogs of GTP, indicating that interactors of the PRD may regulate constriction (38). Perturbation of dynamin was shown to affect parameters of actin dynamics in a variety of systems (43). There is strong evidence for a role of actin in endocytosis, and recent studies have suggested a very close spatial and temporal relationship between recruitment of dynamin and of actin at endocytic sites (44).

Irrespective of its mechanism of action, the role of dynamin in the scission of endocytic necks or of other tubular membranes is likely to be highly regulated. For example, endogenous or overexpressed wild-type dynamin is present on the narrow tubules induced in nonneuronal cells by overexpression of an amphiphysin isoform (45). Yet these tubules do not undergo obvious fragmentation. Long and convoluted tubules with a striped surface reminiscent of a dynamin coat, and often connected to a clathrin coated bud, are present in specialized neuronal cells, such as nerve terminals of retinal photoreceptors (46). The molecular signals controlling dynamin-mediated fission are unknown.

A plausible interpretation of the results of Iversen et al. (7) is that the function of clathrin and dynamin may be closely coupled, so that clathrin itself, or other accessory factors whose localization/function depends on clathrin, may be involved in triggering fission. Clathrin and these factors may act by stimulating the GTPase activity of a dynamin collar assembled at the vesicle neck or by regulating/facilitating interactions of dynamin-GTP with other effectors. Alternatively, constriction of the free edge of the clathrin coat to induce fission may be mediated by coat components, but only on activation via the interaction with dynamin-GTP. Limiting concentrations of clathrin may decrease the efficiency of the coupling between clathrin-mediated budding and dynamin-mediated fission, thus leading to the elongation of dynamin coated necks, similar to what is observed after GTPγS treatment. In the electron micrographs that accompany the study of Iversen et al., the buds connected to dynamin tubules have the appearance of clathrin coats, but because of the decreased concentration of cellular clathrin, such a coat may be abnormal and therefore unable to achieve the proper coupling to dynamin. The presence of other “coat proteins” such as adaptors and their accessory factors, as well as of dynamin itself (see for example the “coat-like” appearance of dynamin at the surface of a vesicular structure in Fig. 1G), may contribute to the electron density of the coat, even if the clathrin lattice itself is discontinuous.

Iversen et al. (7) also reports that clathrin suppression induced a massive up-regulation of dynamin which could be accounted for, at least in part, by a major increase of dynamin 2 mRNA levels. This change may reflect the existence of a clathrin-dependent, or endocytosis-dependent, signaling pathway that regulates the expression of dynamin via indirect actions in the nucleus. Alternatively, perturbation of the normal ratio of soluble versus membrane-bound dynamin due to sequestration of the protein on membrane tubules may trigger de novo synthesis of dynamin to replenish the decreased cytosolic pool. Dynamin overexpression is likely to be the consequence, and not the cause, of collared tubule formation, because the phenotype is not reproduced by overexpression of exogenous dynamin 2. Among the known major endocytic proteins, only dynamin 2 was found to be up-regulated. However, because electron-dense dynamin rings can comprise dynamin binding proteins, additional work is required to determine whether other proteins are up-regulated and present in the rings. It is of interest to note that a major up-regulation of dynamin levels was also observed by blocking clathrin-mediated endocytosis via overexpression of an Eps15 deletion mutant (47). Even in that case, dynamin was found to be largely sequestered at the cell surface, though in a pattern which seems to be different from the one reported here.

It remains very puzzling why a major block in receptor-mediated endocytosis, and even an apparent and sustained increase in the number of coated pits, were observed at a stage when the cellular level of clathrin had not yet significantly decreased. The authors suggest the possibility that newly synthesized clathrin may, for some reason, be more efficient in the endocytic reaction. This intriguing observation requires further experimentation. For example, clathrin may undergo an “aging process” that compromises its ability to be regulated by Hsc70 (48, 49). It is also possible that once an initial threshold in the delay of fission has been reached, some clathrin may be trapped on nonfunctional endocytic pits and the apparent increase in coated pit number may simply reflect impaired clathrin coat turnover. Furthermore, as discussed above, the coats of the vesicular buds that caps dynamin-coated tubules may not be uniformly represented by clathrin.

Beyond the new clues that this study provides concerning a potential role of clathrin in the regulation of dynamin, the study by Iversen et al. demonstrates that dynamin collars can occur, abundantly, in nonneuronal cells. Such collars have been difficult to detect outside nerve terminals, raising the possibility that their massive assembly may requires nerve terminal-specific factors and that dynamin collars may not reflect a physiological state of dynamin. Clearly, this is not the case.

A model in which clathrin (or clathrin-associated proteins) represents the key trigger for dynamin-mediated vesicle scission is consistent with the requirement for a tightly regulated coordination of sequential steps in the budding process. When the assembly of a clathrin-coated bud has been completed, a signal must exist to engage dynamin and the fission machinery. Similarly, following fission, coat shedding must occur and thus, the fission machinery may in turn be the trigger for uncoating. In a process that requires multiple sequential steps, a coordination of one step with the following one may ensure directionality in this process, thereby providing a significant strategic advantage.