[
G T P a s e d o m a in□
□
a 8 6 4 a a pH d o m a in coiied c o ilXmaln'PRD^ b 8 5 1 a a Dynamin I; 4 Isoforms[
G T P a se d o m a in[
G T P a s e d o m a in□
PH d o m a in 8 7 1 a a Dynamin 2; 4 Isoforms□
PH d o m a in a 8 4 6 a a b 8 5 7 Dynamin 3; 4 IsoformsFigure 1.9 Domain structure of the dynam ins
The three dynam in genes are characterised by m ultiple dom ains and several sites of alternative splicing o f the mRNA. The PH, coiled-coil, and proline rich (PR) dom ains are labelled in dynamin. The letters a and
G TPase domain PH dom ain PRD
I---
1 Dyanmin 1 , 2 ,3 Vps1P/Spo15 D nm ip Mx1, 2 ,3 M gmip ÎVol-tÎA ^ A rabldopsisFigure 1.10 The family o f dynam in-like proteins
M em bers are shown in decreasing order of relatedness to dynamin. Each has the c h a ra c te ristic G T Pase dom ain. T he PH dom ain and PR D are unique to the dynamins.
these m olecules show little amino acid similarity. This C-term inal variability m ay in part specify and explain the functional diversity of these proteins.
Protein Length (aa) MW (kDa) % Wentity with dynam in 1 % identity with GTPase dtnnain o f dynamin 1 Possible funeticMi
Dynamin 1 864 96 100 100 vesicle fission in synaptic
v esicle re trie v a l and
r e c e p t o r m e d ia te d endocytosis Dynamin 2 870 97 79 88 Dynamin 3 864 96 89 90 S h ib ire 883 98 68 82 e le g a n s 865 96 64 75
Vpsplp 704 79 45 64 Yeast traffic G olgi to
vacuole Dnm Ip 760 85 41 50 Y east traffic PM to endosome M xlp 652 72 31 40 Mx2p 659 73 Mx3p 659 73 M gmlp 843 92 28 32 M itochondrial genom e maintenance 588 65 24 30
Table 1.1 Features of the dynam in-like proteins determ ined from am ino acid
sequence analysis
Regulators o f dynamin GTPase
D ynam in differs from m ost G T Pases in that it is large (100 kD a) and has high endogenous GTPase activity of ~ 100 nmol/min per mg (Shpetner and Vallee, 1992). This G T P a se a c tiv ity can be reg u la te d in v itr o by sev eral fu n c tio n a lly d iv e rs e m acrom olecules. Stim ulators of the GTPase activity includes m icrotubules, SH3 dom ain containing proteins, such as Grb2 and acidic phospholipids (Gout et al., 1993; Lin et al., 1997; Shpetner and Vallee, 1992; Tuma et al., 1993). The m echanism of this activation is poorly understood, however based on kinetic data showing a dependence of activation on enzym e concentration (Lin and Gilman, 1996; Tum a and Collins, 1994; W arnock et al., 1996; W arnock et al., 1995), dynamin self-association appears to be a key elem ent of the
activation process. It has been proposed that two potent activators of dynamin 1, microtubules and phosphatidylserine (PtdSer), stimulate activity by facilitating dynamin self assembly (Tuma and Collins, 1994). They may do so by providing multivalent negatively charged surfaces that bind to the highly basic PRD ( p i- 12.5) of dynamin molecules. The ionic nature of these interactions is evident by their marked sensitivity to ionic strength; both microtubule and PtdSer stimulated activities are nearly abolished at physiological salt concentrations (Lin et al., 1997; Tuma and Collins, 1994). The GTPase activity of dynamin 1 is also stimulated by the binding of SH3 domains of Grb2 to the PRD (Gout et al., 1993; Herskovits et al., 1993a). Furthermore, crosslinking dynamin molecules via interlinked monoclonal antibodies that recognise epitopes within the dynamin PRD, stimulates its GTPase activity in the absence of other activators (Wamock et al., 1995). Neither the isolated SH3 domains of Grb2 (Gout et al., 1993) or the monovalent Fab fragments of the above antibodies (Warnock et al., 1995) activated dynamin GTPase.
Analysis of PRD deletion mutants has demonstrated that this domain is not only responsible for interactions with the GTPase regulators PtdSer, Grb2 and microtubules (Herskovits et al., 1993a; Tuma et al., 1993) but it also contains a PKC phosphorylation site which results in the stimulation of intrinsic dynamin GTPase activity in vitro (Liu et al., 1994). However, dynamin 1 is not phosphorylated by PKC in stably transformed cell
in vivo (Damke, 1996), arguing against a physiological role for this interaction.
Apart from PtdSer, the GTPase of dynamin 1 has been shown to be stimulated by rat brain lipid vesicles and acidic phospholipids including phosphatidylglycerol, Ptdlns and Ptdlns(4,5)P2 (Lin and Gilman, 1996; Tuma et al., 1993). The level of stimulation was comparable to that obtained with microtubules and was also found to require the presence of the PRD (Tuma et al., 1993). However, the reported binding of phosphoinositides to several PH domains, specifically Ptdlns(4,5)P2 to the PH domain of dynamin 1 (discussed in Chapter 5) led to reinvestigation of the role of the PRD in the phosphoinositide activation of dynamin GTPase. Lin et al., (1997) have recently shown that Ptdlns(4,5)P2 mediated activation of dynamin 1 and dynamin 2 is independent of the PRDs, whilst the GTPase of a dynamin 1 PH domain deletion mutant was not activated by Ptdlns(4,5)P2 (Chapter 4). The discrepancy betw een the aforem entioned phosphoinositide-PH domain and the PRD mediated phosphoinositide activation of dynamin GTPase (Tuma et al., 1993) is not easily reconcilable. However, what is clear is that the PH domain is another allosteric site of regulation for dynamin GTPase.
A further insight into the regulation of dynamin GTPase has been provided by the demonstration that the coiled coil domain (interposed between the PH domain and the PRD) interacts with the N-terminal GTPase domain to stimulate GTP hydrolysis (Muhlberg et al., 1997). These results suggest that dynamins are subject to multiple modes of regulation, some involving interactions with PRD, PH domain, or the coiled coil domain.
Cellular Role o f dynamin
Mutational analysis of the drosophila dynamin homologue, shibire, provided the initial insights into the in vivo function of dynamin. When temperature-sensitive mutant shibire
flies were raised from 22°C to 29°C (non-permissive temperature), they became reversibly paralysed (Kosaka and Ikeda, 1983; Poodry and Edgar, 1979). Morphological analysis of the neuromuscular junctions from affected flies revealed that this temperature dependent paralysis was due to a defect in synaptic vesicle recycling: after brief incubations at the non-permissive temperature, synapses were depleted of their vesicle content and invaginated clathrin-coated and non-coated pits accumulated on pre-synaptic membranes (Koenig and Ikeda, 1989; Kosaka and Ikeda, 1983; Poodry and Edgar, 1979). Given that neurotransmitter release and the postsynaptic response were normal in these flies, it was clear that the shibire gene product played a direct and essential role in the early stages of endocytosis.
Evidence that mammalian dynamin is performing a comparable role in endocytosis has come from transfection studies incorporating wild type and GTPase defective (dominant negative) dynamin forms into cultured mammalian cells (Damke et al., 1995; Damke et al., 1994; Herskovits et al., 1993; van der Bliek et al., 1993). Analysis of cells expressing dominant-negative mutants revealed an accumulation of clathrin-coated pit structures at the plasma membrane and long membrane tubules, consistent with the defects seen with shibire mutant cells. In contrast to these observations, immunoblot analysis comparing the distribution of clathrin and dynamin immunoreactivity in subcellular fractions from rat brain does not demonstrate a convincing co-localisation (Sandvig and Van Duers,
1994). Instead, there is a prevalence of dynamin in a heterogeneous membrane pellet and a reduction in the plasma membrane-coated vesicle fraction. Furthermore, of the numerous antibodies made to the conserved and isoform-specific domains of the dynamins, only one (Damke et al., 1994) has been demonstrated to localise with clathrin in mammalian cells. Thus while the majority of data supports the involvement of dynamin in clathrin-mediated endocytosis, it is possible that dynamin is also involved with other non-clathrin mediated endocytotic pathways.
The intrinsic ability of dynamin to self-assemble into rings appears to be central to its function in endocytosis. Dynamin itself appears to exist primarily as a tetramer (Muhlberg et al., 1997), which has the ability to assemble into rings or collar around certain templates. These appear to comprise of four to six of the tetrameric assembly units (Hinshaw and Schmid, 1995). Rings were first observed as collared pits in electron micrographs of the neuromuscular junction of shibire flies (Koenig and Ikeda, 1989; Kosaka and Ikeda, 1983). Rings were next observed when purified dynamin 1 coiled around microtubules in vitro (Maeda et al., 1992). It is now recognised that the diameters of microtubules and the necks of recycling vesicles are similar, 25 nm for microtubules and 25-30 nm for vesicle necks, suggesting they may form ideally sized templates for ring formation (Takei et al., 1995). It was found that concentrated solutions of recombinant dynamin 1 could be induced to spontaneously form rings with a pronounced helical twist in vitro in the absence of any template, by dialysing the protein into buffers containing no salt (Hinshaw and Schmid, 1995; Wamock et al., 1996). The diameter of these rings remarkably matched those around the recycling vesicle in shibire nerve terminals. The clearest evidence was provided by electron microscopy images showing dynamin on membranous invaginations in isolated rat brain synaptosomes treated with GTP[yS)] (a non-hydrolysable analogue of GTP) forming densely stained, regular spaced striations around the necks of budding vesicles (Takei et al., 1995). On the basis of these results a model was proposed (Wamock and Schmid, 1996) in which dynamin polymerises and winds round the neck of an invaginated plasma membrane pit and, through a concerted conformational change associated with GTP hydrolysis, scission of a coated vesicle from the plasma membrane occurs (Figure 1.11).
Although a clear picture of dynamin’s role in synaptic vesicle retrieval is emerging, many aspects remain unresolved. Most importantly, how is dynamin initially targeted to the site of retrieval and what is its specific binding site? As previously discussed, dynamin has been shown to interact with microtubules and acidic phospholipids. However, the physiological relevance of these interactions remains controversial. Dynamin has not been found to co-localise with microtubules in vivo (Noda et al., 1993; Scaife and Margolis, 1990), and cultured cells transfected to overexpress mutant forms of dynamin do not show any alteration in microtubule organisation (Herskovits et al., 1993). Furthermore, it is unlikely that interaction with acidic membrane phospholipids is sufficient to account for the specificity and saturability of dynamin's membrane binding site. More recently, work with wild-type and mutant dynamins transfected into COS cells clearly illustrates that it is a specific site within the PRD that targets dynamin to coated pits (Shpetner et al., 1996), thus indicating that the key targeting signal for dynamin may