Residues important for GTP binding
We have already discussed the conservation of G1 to G4 GTP binding sequeces in ras, DymA and GBP (see structures). We now address in more detail some of the important residues that were highlighted in the previous discussion (see table) concentrating on DymA and ras GTPase domains. The core structure of a GTPase domain is a beta-sheet surrounded by alpha-helices (see structures below). Many of the residues that coordinate GTP/GDP + Mg2+ are on flexible loops and many of the protein interactions of the ras GTPase domain involve these loops. This allows for binding proteins to sense the GTP binding state or to promote hydrolysis or exchange. We would love to be able to hypothesise how dynamin works from comparisons of ras and the DymA structure but one would need the GTP bound form of DymA and preferably the complete structure rather than simply the GTPase domain bound to GDP to understand mechanochemical coupling. However, as we said previously, the long terminal alpha-helix in DymA and a long helix in a similar position in GBP are in key positions to play a role as a lever arm, if there were an overall conformational change in the GTPase domain on GTP hydrolysis.

Comparison of ras and dynamin GTPase domain core structures
ras bound to GTP (PDB:121p)
DymA bound to GDP (PDB:1jwy)
Highlights of the extra features in DymA compared to ras
Endocytosis, Dynamin, Dynamic dynamin, Shibire, Vesicle scission, Harvey McMahon, Dlp, OPA, Dmn, DymA, GBP, GTPase, Poppase, Pinchase, GTP, PIP2, PtdIns(4,5)P2, phosphatidylinositol phosphates, synaptic vesicle endocytosis Endocytosis, Dynamin, Dynamic dynamin, Shibire, Vesicle scission, Harvey McMahon, Dlp, OPA, Dmn, DymA, GBP, GTPase, Poppase, Pinchase, GTP, PIP2, PtdIns(4,5)P2, phosphatidylinositol phosphates, synaptic vesicle endocytosis Endocytosis, Dynamin, Dynamic dynamin, Shibire, Vesicle scission, Harvey McMahon, Dlp, OPA, Dmn, DymA, GBP, GTPase, Poppase, Pinchase, GTP, PIP2, PtdIns(4,5)P2, phosphatidylinositol phosphates, synaptic vesicle endocytosis
6 stranded beta-sheet surrounded by 5 alpha-helices
8 stranded beta-sheet surrounded by 9 alpha-helices


The residues important for GTP binding
Movies compiled using Aesop, courtesy of Martin Nobel
Double click pictures to play movies......

The above movies show where the GTP binding residues are in ras and DymA. In many cases they are found on loops but one can also note that K16 and S17 are attached to the core structure (to alpha-helix 1), and in DymA the equivalent residues, K38 and S39, are also on the corresponding alpha helix.
A 2D representation is shown below for GDP bound to DymA (using LigPlot).

Endocytosis, Dynamin, Dynamic dynamin, Shibire, Vesicle scission, Harvey McMahon, Dlp, OPA, Dmn, DymA, GBP, GTPase, Poppase, Pinchase, GTP, PIP2, PtdIns(4,5)P2, phosphatidylinositol phosphates, synaptic vesicle endocytosis see enlarged LigPlots of:

GTP-ras (using a nonhydrolysable GTP analogue GppCp),

GDP-ras,

GTP-hGBP (using GppNp)

GDP-DymA

From the crystal structures and these plots of GTP/GDP coordination we can now explain the phenotypes of many of the dynamin I mutants (see below). We can also begin to understand the Shibire phenotypes (see Shibire).


Key dynamin I mutations: K44, S45, T65, D208...The most widely used mutations to weaken or abolish nucleotide binding are in residues of the G1-region. Thus mutations of Lysine 44 or Serine 45 in dynamin I have been used in various studies. The low nucleotide affinity of wild-type dynamin means that the accurate affinity measurements have not been determined. Qualitative data shows that GTP binds to S45N and K44A mutants of dynamin I but but with much lower affinities than wild-type. It is generally assumed from comparisons with other GTP-binding proteins that dynamin K44A and S45A are predominantly nucleotide free under cellular conditions (37 deg. C., 150 mM free GTP, 15 mM free GDP). This is likely an overassumption but in cells these mutants show a dominant negative phenotype and block endocytosis.
Turnover Rate (kcat)(s-1) % of WT activity Km(microM)
WT Dynamin 3.1±0.5 100±16 7.8±2.5
S45N 0.083±0.043 2.6±1.4 ND
T65A 0.023±0.015 0.74±0.48 9.1±1.1
Dynamin Ras
K44 K16
S45 S17
T65 T35
D208 D119

The residue which corresponds to Dynamin's S45 in ras is S17. S17N has reduced hydrolytic activity. The nucleotide affinity of S17N is decreased leading to a stronger interaction with the GEF. Moreover, depletion of the GEF causes a dominant negative effect of this mutation because ras can now be GTP bound.
Endocytosis, Dynamin, Dynamic dynamin, Shibire, Vesicle scission, Harvey McMahon, Dlp, OPA, Dmn, DymA, GBP, GTPase, Poppase, Pinchase, GTP, PIP2, PtdIns(4,5)P2, phosphatidylinositol phosphates, synaptic vesicle endocytosis The residues that are generally used to generate permanently GTP-bound Ras-related proteins are not conserved in the dynamin family, thus a GTPase negative mutant was only recently identified by our lab. Mutation of the residue Threonine 65 (T35 in ras) to an alanine blocks GTP-hydrolysis but seems to have only little effect on GTP-binding. This permanently GTP-bound mutant also blocks endocytosis and leads to the formation of long tubular structures throughout the cell, implicating that GTP-hydrolysis as necessary for vesicle budding (see Marks et al 2001). Even if S45N in dynamin results in a lower affinity for GTP, it also gives a similar tubulation phenotype on a much smaller scale to T65A (see paper). This is likely because it can still hydrolyse GTP just like T65A but at a much slower rate.

What about D208?
Some residues that were mutated in dynamin showed little or no effect despite their conservation and phenotype in ras and their conservation in other GTP-binding proteins. For example exchange of the conserved Aspartic acid 208 that is involved in the specific recognition of the guanine base to asparagine in dynamin I did not inhibit endocytosis. It seems that this mutation does not weaken the nucleotide affinity to the same extent as the K44A in the G1 and indeed this residue is not close to the guanine base in DymA. In our studies we have used D208W, where a larger hydrophobic residue is substituted for aspartic acid and this gives a dominant negative phenotype likely by sterically blocking nucleotide binding.
In contrast, the mutation of a residue corresponding to a temperature-sensitive gain-of-function mutation in ras from C.elegans gave a similar phenotype to the permanently GTP-bound mutant of T65A described before in a dynamin related protein.

Interesting Ras mutants not present in dynamin
The intrinsic GTPase activity of Ras is low (0.028 min-1). Upon binding of GAP it is stimulated up to five orders of magnitude. Several mutations in Ras inhibit the intrinsic GTPase activity and, more importantly, can also not be stimulated by GAP. G12V (in G1) and Q61L (in Switch II) both make ras permanently GTP bound and thus they are oncogenic. These residues are conserved in G-alphas but not in dynamin family members. Apart from a lower intrinsic GTPase activity of these mutants and an inability to be GAP stimulated, they can still bind to effectors and so can still signal. It is this last feature that distinguishes G12V and Q61L for T35 mutants. Mutations of this residue also result in a lower GTPase activity but this mutant cannot interact with effectors and so is not oncogenic. We have not tested if dynamin binding partners interact less well with dynamin T65A versus WT protein, but so far none of the known binding partners interact with the Switch 1 region.

Is there a rational for the low nucleotide affinity of dynamin?

Dynamin Ras
KM 8-15mM GDP dissociation is limiting
kcat 190-260 min-1 intrinsic 0.028 min-1

GAP stimulated 20 s-1 = 1200 min-1

Kd 0.2-1mM (GTP)

19mM (GDP

5pM (GTP)

21pM (GDP)

29mM (GMP)

Ras has a very high affinity for GTP and GDP with dissociation constants in the sub-nanomolar range. In contrast, GMP interacts only weakly with Ras and nucleotide free protein is instable (John et al. 1990). The high affinity is mainly caused by the slow dissociation rate in the range of 10-5 s-1.
In the case of GTPases from the dynamin family, the dissociation rates are in the range from 10-2 s-1 to 102 s-1 (Richter 1995, Praefcke 1999, Binns 2000). These numbers are similar to Ras like GTPases in presence of their GEFs which accelerate this step by a factor of 500 - 10000 (CDC25, Lenzen et al. 1995, Rudolph 1999). The main interactions formed by the beta-phosphate in the GDP bound state of high affinity GTPases are via the G1 region and the Mg2+ ion. Therefore these are the interactions the GEFs interfere with (reviewed in Cherfils and Cardin 1999), with additional minor rearrangements of the switch regions.
In the structures of the nucleotide free forms of DymA and hGBP1, the conserved lysine in G1 interacts with residues from the G3 (DxxG) and this interaction is believed to be the main stabilizing interaction for GTPases in the nucleotide free state. This type of interaction is also present in structures of various high affinity GTPases in complex with their GEFs and intriguingly in the GDP-bound form of DymA. This clearly shows that dynamin related proteins have their GEF already inbuilt.

Using Fluorescently labelled Nucleotides: Fluorescently labelled nucleotides have been widely used to study the interactions of nucleotide-binding proteins with their substrates. In the case of small Ras-related GTPases the modification of the hydroxyl group of the ribose moiety of GTP with methyl-anthraniloyl (mant-) derivates was used as a tool to determine the nucleotide affinities in various studies. The addition of the mant-group did not severely impair the binding properties of the proteins in Ras-related GTPases since the hydroxyl moieties of the ribose are mainly solvent exposed. Nevertheless, the affinity of the labelled nucleotides was found to be different from the real substrate with up to tenfold weaker or tighter binding, depending on the protein.

Endocytosis, Dynamin, Dynamic dynamin, Shibire, Vesicle scission, Harvey McMahon, Dlp, OPA, Dmn, DymA, GBP, GTPase, Poppase, Pinchase, GTP, PIP2, PtdIns(4,5)P2, phosphatidylinositol phosphates, synaptic vesicle endocytosis
Mant GTP

In case of dynamin and related proteins the situation is more complex since the proteins undergo nucleotide dependent oligomerisations that can influence the affinity for the nucleotide. The first study with a dynamin-related protein was done with human MxA. The authors showed that the protein has weak affinities for nucleotides in the micromolar range, a preference for GTP over GDP and that the mant-labelled nucleotides bind 5 to 20 fold tighter than non-labelled ones (Richter et al .1995).
A transient kinetic study by Binns et al. 1999 showed that dynamin2 also interacts weakly with mant-nucleotides. However they estimate that the mant-labelled nucleotides interacts about tenfold weaker. Several steps of the association and dissociation showed multi-phasic behaviour that was attributed to a mixture of assembled and non-assembled forms of dynamin. From the structure of DymA GTPase domain it is very likely that Mant-GTP is sterically restricted in the binding site.

Dynamin II mdGTP: 91 mM, mdGDP 19 mM,
Ten-fold slower dissociation of GTP than mGTP
MxA mGTP 0.75 mM, mGDP 20 mM
GTP: 20 mM, GDP: 100 mM

What is an "Arginine Finger"?
The structure of Ras in complex with its GAP and an analogue of the transitions state of the GTPase reaction (GDP and aluminium fluoride) gives insights into the mechanism of GTP hydrolysis. The GTPase activation protein inserts the side chain of a conserved arginine into the active site of Ras. The positive charge of this “arginine finger” compensates the negative charges of the oxygen atoms of the gamma phosphate in the transition state. In G-alpha proteins, a similar scenario has been seen. However, the arginine is supplied by the helical domain that is inserted into their GTPase domain. This is a reason for the higher intrinsic GTPase activity of G-alpha proteins compared to Ras proteins. The RasGAP has further functions. A second positively charged residue is stabilizing the conformation of the catalytic arginine and the GAP also positions the conserved Gln61 which in turn is important for the correct positioning of the attacking water molecule. Similar stabilizing effects are also the reason for the stimulation of the GTPase activity of G-alpha proteins by RGS proteins. The “arginine finger” model was thought to be the case for all GTP binding proteins since similar scenarios have also been found for Rho, Rab and Arf GTPases. However, the universal impact of this theory has recently been challenged. Structural and mutational analysis of the GAP stimulated GTPase of Ran and Rap showed no catalytic arginine. Instead, the GTP binding region is likely stabilized giving rise to higher GTPase activity (Seewald, M. 2002, Brinkmann T. 2002).
In dynamin, the helical structure of the GED and the conservation of certain positively charged residues lead to the idea that this region might contribute a catalytic residue into the active site similar to Ras and G-alpha proteins. Indeed, mutations of two of those residues were found to have impaired GTPase activity and a block of the biological function has been shown for dynamin, Dlp and Mgm1. We have not been able to reproduce the effects on GTPase activity and we believe that the mutation have an indirect effect by impairing the oligomerisation properties of the proteins. Also, in Ras related proteins the active site is open and the nucleotide is bound in a groove. In contrast, the structure on hGBP1 bound to GTP shows that the nucleotide is completely covered by two loops from the solvent and the corresponding loops have similar length in dynamin. This does not exclude the likely possibility that a catalytic residue is contributed by the GTPase domain and is moved into position on oligomerisation.

Back to top

Back to GTPase domain
Back to Dynamin Home Page