GEF-mechanism of the Rab8-GEFs Rabin8/GRAB 1 Intermediates in the Guanine Nucleotide Exchange Reaction of Rab8 Catalyzed by Rabin8/GRAB

Small G-proteins (G) of the Ras-superfamily control the temporal and spatial coordination of intracellular signaling networks by acting as molecular on/off switches. Guanine nucleotide exchange factors (GEFs) regulate the activation of these G-proteins through catalytic replacement of guanosine diphosphate (GDP) by guanosine triphosphate (GTP). During nucleotide exchange, three distinct substrate-enzyme complexes occur: A ternary complex with GDP at the start of reaction (G:GDP:GEF), an intermediary nucleotide-free binary complex (G:GEF), and a ternary GTP-complex after productive G-protein activation (G:GTP:GEF). Here we show structural snapshots of the full nucleotide-exchange reaction sequence together with the G-protein substrates and products using Rabin8/GRAB (GEF) and Rab8 (Gprotein) as a model system. Together with a thorough enzymatic characterization, our data provide a detailed view into the mechanism of Rabin8/GRAB mediated nucleotide exchange. INTRODUCTION One of the hallmarks of eukaryotic cells is the intracellular movement of vesicles that transport material and allows communication between cellular compartments. The spatial and temporal regulation of vesicular trafficking is achieved by proteins of the Rab subfamily of small GTPases (1,2). Rabs are molecular switches and cycle between inactive GDP-bound and active GTP-bound states. When inactive, Rab proteins exist in the cytosol in complex with the GDP dissociation inhibitor http://www.jbc.org/cgi/doi/10.1074/jbc.M113.498329 The latest version is at JBC Papers in Press. Published on September 26, 2013 as Manuscript M113.498329 Copyright 2013 by The American Society for Biochemistry and Molecular Biology, Inc. by gest on N ovem er 9, 2017 hp://w w w .jb.org/ D ow nladed from GEF-mechanism of the Rab8-GEFs Rabin8/GRAB 2 (GDI) but are localized to a distinct membrane when in the active state. In order to exert their function, Rabs need to be activated by a process requiring GDP/GTP exchange factors (GEFs). These enzymes accelerate GDP release from and allow the binding of GTP to a Rab protein. Rabs can interact with effector proteins that preferentially bind the active GTP but not the GDP state. GTPase activating proteins (GAPs) stimulate the very low intrinsic GTPase activity of Rab proteins and thus convert them back into the inactive form. The Rab subfamily consists of approximately 60 members in humans, and each family member has a specific intracellular localization (2). The correct activation of a certain Rab requires the action of a cognate RabGEF at the proper location and at the appropriate time (3). Consequently, GEFs have evolved to have mechanisms that guarantee their correct membrane targeting as well as the specific recognition of one distinct Rab protein over structurally and sequentially similar family members. The Rab protein Rab8 is involved in events such as the delivery of secretory vesicles to the plasma membrane and in polarized membrane transport in epithelial cells (4,5). Rab8 also regulates cell shape, and the interaction with its GEF Rabin8 appears to be crucial to this function (6). Rabin8 and Rab8 appear to be important in cilium formation by acting in concert with the BardetBiedl-Syndrome complex (7-9). Rabin8 is a 460 amino acid protein that consists of several domains, of which only 2 are functionally characterized (5,10,11). Rabin8 contains a central Sec2 domain having GEF activity towards Rab8 (5). Amino acid sequence and structure comparison to the yeast homologue Sec2 predicts that the central domain of Rabin8 consists of a homodimeric parallel coiled-coil. Rabin8 is thought to be recruited to its target location by active Rab11 which interacts with a C-terminal Rab11-effector domain (10). Another factor possessing a Sec2-like GEFdomain is the 382 amino acid protein GRAB. Despite a high sequence homology of the Sec2domain of GRAB to Rabin8, it has previously been reported that GRAB is a GEF for Rab3A rather than Rab8 (12). However, a recent study that analyzed the activity profiles of various RabGEFs (with a focus on the DENN-domain family which is structurally unrelated to GRAB) has indicated that GRAB has GEF activity towards Rab8 but not Rab3 (13). The mode of action of GTPases involved in signal transduction or regulation includes activation of GDP release catalyzed by a specific guanine nucleotide exchange factor (GEF) in almost all cases known. The basic mechanistic feature is the weakening of the otherwise very tight binding of GDP (Kd values in the nM to pM range) by interaction with GEFs, which also bind with similarly high affinity to their cognate GTPases (14). The effect is both thermodynamic and kinetic, implying the formation of a ternary complex between GEF, GTPase and GDP, with a dramatic reduction in the affinities of both GEF and GDP in the ternary complex in comparison with the respective binary complexes. Several GTPase-GEF interactions have been examined thoroughly at the kinetic level, and a large number of GTPase:GEF complexes has been characterized structurally. In the present work, we have examined the interaction of the Ras superfamily protein Rab8 with 2 structurally highly similar GEF molecules (Rabin8 and GRAB) by kinetic and structural methods. GRAB is a GEF for Rab8 with almost identical structural properties to Rabin8. The work presented leads to the identification of several intermediates in the overall GTP/GDP exchange of Rab8 in the presence of Rabin8. EXPERIMENTAL PROCEDURES Protein expression and purificationHuman Rab8a1-184 and Rab8a6-176 were expressed in E. coli and purified as described previously (15). Expression and purification of Rab3 was performed as described (16). The protein encoding sequences of full length GRAB and coiled-coil domains of Rabin8/GRAB were cloned into a modified pET19 vector containing an N-terminal His6-tag followed by a Tobacco Etch Virus (TEV) protease cleavage sequence(17). In the case of GRAB, a synthetic codon optimized gene was used. GRAB and Rabin8 variants were expressed in E. coli BL21(DE3)RIL by induction with 0.5 mM IPTG (isopropyl-β-dithiogalactopyranoside) at 20°C for 18 h and purified by Ni-NTA affinity chromatography. After removal of His6-tag with TEV protease, the GRAB and Rabin8 variants were further purified by Ni-NTA affinity chromatography followed by size exclusion chromatography. Preparative loading of Rab8a6-176 with GppNHp was performed as described previously (18). To obtain a nucleotide-free Rab8a1-184:Rabin8157-232 complex, Rab8:GDP was mixed with Rabin8 at 1:4 molar by gest on N ovem er 9, 2017 hp://w w w .jb.org/ D ow nladed from GEF-mechanism of the Rab8-GEFs Rabin8/GRAB 3 ratio in buffer containing 20 mM Hepes pH 7.5, 50 mM NaCl, 150 mM (NH4)2SO4, 50 μM ZnCl2, and 3 mM DTT. Alkaline phosphatase was added to hydrolyze GDP and the mixture was incubated for 10 h at 4°C. GDP hydrolysis was monitored by reversed phase HPLC (19) and Rab8:Rabin8 was purified by size exclusion chromatography with buffer containing 25 mM HEPES pH 7.5, 40 mM NaCl and 5 mM DTT after GDP was completely hydrolyzed. Crystallization and structure determinationAll crystals in these studies were obtained by mixing 1 μl protein and 1 μl reservoir solution at 20°C using the hanging drop approach. Crystals of Rab8a6-176:GDP were obtained by mixing the protein (20 mg/ml, buffer: 25 mM HEPES pH 7.5, 40 mM NaCl, 1 mM MgCl2, 10 μM GDP and 5 mM β-mercaptoethanol) with a reservoir consisting of 16% (w/v) PEG4000, 0.1 M CaAc2, 0.1 M HEPES pH 7.0. The crystal was protected with cryo solution containing 30% (w/v) PEG4000, 0.1 M CaAc2, 0.1 M HEPES pH 7.0 before data collection. Rab8a6-176:GppNHp (15 mg/ml; buffer: 25 mM HEPES pH 7.5, 40 mM NaCl, 1 mM MgCl2, 10 μM GppNHp and 5 mM βmercaptoethanol) crystals were produced in 15% (w/v) PEG8000, 7.5 %(v/v) MPD, 0.1 M HEPES pH 6.8. Nucleotide-free Rab8a1-184:Rabin8157-232 (10 mg/ml)) was crystallized in 18 % (w/v) PEG3350, 0.1 M Li2SO4, 0.1 M MES pH 6.6. Before data collection, the complex crystal was protected with cryo solution (30% (w/v) PEG3350, 0.1 M Li2SO4, 0.1 M MES pH 6.6). In order to produce nucleotide-bound forms of Rab8:Rabin8 complexes, the nucleotidefree Rab8:Rabin8 crystals were soaked with cryo solution containing 30 % (w/v) PEG3350, 0.1 M Li2SO4, 0.1 M MES pH 6.6 and 1 mM respective nucleotide GDP/GTP for one hours at 4°C. Rab8:GRAB (10 mg/ml) was crystallized in solution containing 1.6 M ammonium sulphate and sodium acetate at pH 5.3. The crystals were protected with cryo solution containing 20 % glycerol in the reservoir solution before data collection. All diffraction data were collected at 100 K at station X10SA of the Swiss Light Source (SLS, Villigen, Switzerland). Data were processed with XDS (20). The structure was determined by molecular replacement with PHASER (21) of the CCP4 suite using Sec4p in the case of Rab8:GDP/GppNHp and Sec2p:Sec4p in the case of Rab8:Rabin8/GRAB complexes as searching model. The model was then corrected by alternating rounds of refinement in REFMAC5 (22) and manual adjustment in COOT (23). The nucleotide was added in the final rounds of refinement. Full data collection and refinement statistics are summarized in Table S1. GEF activity measurementsFluorescence measurements were carried out at 25C in the buffer containing 50 mM HEPES (pH 7.5), 50 mM NaCl, 5 mM MgCl2, 5 mM DTE. Rab8a was loaded with the fluorescent GDP/GTP derivatives (mantGDP/mantGppNHp) as described for Sec4 (24). The fluorescence of mant was excited at 365 nm and detected using a 420 nm cut-off filter in a stopped-flow apparatus (Applied photophysics). RESULTS Exchange activities of the GEF-domains of Rabin8 and GRABUsing human Rab86-176 we have investigated the GEF properties of Rabin8 and GRAB in detail. The displacement of a fluorescent GDP analog (mantGDP) from its complex with Rab8 catalyzed by Rabin8153-237 in the presence of excess GDP or GTP (see discussion below concerning the choice of this fragment) showed a marked acceleration with respect to the intrinsic dissociation rate (Figure 1a). A similar effect was seen when Rabin8 was replaced by GRAB73-154 or full length GRAB. The difference between the observed rate constants of Rabin8153-237 (kobs=0.171 s) and GRAB73-154 (kobs=0.207 s) is about 20% at 10 μM GEF. The small differences in the kobs values are not enough to exclude the possibility that they arise from errors in the determination of concentrations the GEF-domains of GRAB and Rabin8. Thus, we conclude that the catalytic activities of the two GEFs are similar and more detailed kinetic investigations were carried out with the GEF-domain of GRAB. Interestingly, no acceleration of mantGDP-release from Rab3a could be observed even at elevated GRAB concentration (10 μM) (Figure 1a). We conclude from this that GRAB is actually a GEF for Rab8 rather than Rab3a, at least under in vitro conditions. A similar observation has been made recently by Yoshimura et al. (13). The observed rate constant of mantGDP dissociation of from Rab8:mantGDP showed a hyperbolic dependence on the GRAB concentration (Figure 1b), allowing the determination of KD3 and k ́ ́-4 (see Scheme 1). KD3 represents the dissociation constant defining the interaction of GRAB by gest on N ovem er 9, 2017 hp://w w w .jb.org/ D ow nladed from GEF-mechanism of the Rab8-GEFs Rabin8/GRAB 4 with Rab8:mantGDP (KD3=43 μM), while k ́ ́-4 is the maximal rate constant for mantGDP dissociation (k ́ ́-4=1.1 s). Thus, the overall catalytic efficiency GRAB towards Rab8 can be determined to kcat/Km= k ́ ́-4/ KD3= 2.6 10 Ms. Additionally, we analyzed the reverse reaction, i.e. the association of nucleotide-free Rab8:GRAB complex with mant-nucleotides. The observed pseudo-first order rate constants of the association of mantdeoxyGDP with the Rab8:GRAB complex were fitted to a hyperbolic function, yielding K ́D4=8.2 μM and k ́ ́+4=34.3 s (Figure 1c). Together with the constants already derived for the interaction of mantGDP with Rab8 (25), these data lead to a complete description of the interactions in the Rab8, GRAB and mantGDP system. The general trend of GEFs is also seen here: A marked reduction in GDP affinity in the ternary complex is caused predominantly by an increased rate constant for GDP release (k ́ ́-4>>k-1). The kcat/Km value (k ́ ́-4/K ́D3 in Scheme 1), giving a measure of the catalytic efficiency of the GEF for a specific substrate, is ca. 2.6·10 Ms. This is similar to the value of 4.7·10 Ms for the interaction of Ras with Cdc25 (26) but lower than the highly efficient Rab GEFs Sec2 (2·10 Ms (24)) or DrrA from Legionella pneumophila (2.2·10 Ms (27)). One interesting aspect of the constants derived is the relatively high affinity of GDP in the ternary complex (KD4 = 0.26 μM), suggesting the possibility of generating the ternary complex for structural studies (see below). Structure of Rab8-nucleotide complexesIn order to obtain insight into the exchange mechanism by GRAB and Rabin8, we first determined the crystal structures of active and inactive Rab8 truncated at the Nand C-termini (amino acids 6-176, Rab86-176) (Figure 2a,b). In both cases there were 5 Rab8 molecules in the asymmetric unit (Figures S1A and S1B). The structures of the flexible regions, including switch I and particularly switch II in the GDP state, varied significantly between the 5 molecules (Figures S1C, S1D). As seen in many other instances of GTPases from the Ras superfamily, only switch I and switch II showed significant structural differences between active and inactive Rab8 (Figure 2c). The switch I and switch II regions were largely disordered in the GDP state, but ordered in the active GppNHp state. Details of the nucleotide-protein interactions are shown in Figure S1E and S1F. Structure of Rab8-Rabin8-complexNext, we aimed to determine the structures of the Rab8:GRAB complex and its homologue Rab8:Rabin8. Both Rabin8 and GRAB contain predicted coiled-coil domains (residues 149-247 in Rabin8, residues 70-160 in GRAB). These align well with highly similar regions in Sec2, which is a GEF for the yeast Rab8 homologue Sec4 (Figure 3). The recently determined structure of the Sec2:Sec4 complex reveals that the interaction occurs in a central region of the Sec2 coiled-coil (28). Several constructs of Rabin8 and GRAB were tested and this led to the identification of Rabin8157-232 and GRAB79-149 as the minimal constructs with full GEF activity. Complexes with Rab86-176 were produced and their crystal structures determined. There were 2 complexes (corresponding to 4 molecules of Rabin8157-232/GRAB79149 and 2 molecules of Rab8) in the asymmetric unit (Figure 4a), resulting in 2 identical complexes. As in the Sec2:Sec4 structure, the GEF region of Rabin8 adopts a parallel homodimeric structure with a length of ca. 180 Å, and the 2 helices can be structurally distinguished from each other as sharply bent and moderately bent (28). There is an intermolecular disulfide bond between the Cys179 residues in the individual Rabin8 chains. There was a presumed sulfate ion in Rab8 at the position normally occupied by the -phosphate of GDP or GTP (Figure 4b), as seen in many other structures of GTPases and ATPases in the absence of nucleotides (29,30). Also, Rabin8 interacts mainly with the switch regions of Rab8 (Figure 4b). An overview of the interactions between individual amino acids demonstrates that half of the Rab8-Rabin8interactions is with one chain of the Rabin8 coiledcoil while the other half is with the other chain in the homodimeric structure (Figure 4c). Rabin8:Rab8 shows a mode of binding reminiscent of the Sec2:Sec4 structure (28) (Figure 4d). In line with this observation, Sec2 can also function as a GEF for Rab8 in vitro (own observations; data not shown). Structure of Rab8-Rabin8-nucleotide-complexDue to the relatively high affinity of the Rab8:Rabin8/GRAB complex for nucleotides (KD4 = 0.26 μM) we reasoned that a ternary GTPaseGEF-GDP/GTP complex could be amenable for protein crystallization and structure determination. We therefore soaked Rab8:Rabin8/GRAB crystals with GDP or GTP and determined their structures. The structures of Rab86-176:Rabin8157-232:GDP and by gest on N ovem er 9, 2017 hp://w w w .jb.org/ D ow nladed from GEF-mechanism of the Rab8-GEFs Rabin8/GRAB 5 Rab86-176:Rabin8157-232:GTP were determined at 3.1 Å and 3.2 Å resolution, respectively. The overall structure of the Rab86-176:Rabin8157232 complexes does not change significantly when GDP or GTP bind (Figure 5a and 5b). The most important interactions with the nucleotide are between the base and Asp124 and between the side chain of Lys21 of the P-loop and the -phosphate of GDP or the and -phosphates of GTP. Several backbone interactions of the P-loop with the phosphates are also observed. The dramatic loss of nucleotide affinity in the Rabin8:Rab8 complex in comparison to Rab8 results from the loss of GDP/GTP interactions with the switch regions and the interaction of the guanine base with Phe33. The latter interaction is conserved in Ras superfamily proteins and was originally shown to contribute significantly to tight nucleotide binding in H-Ras (Phe28 in Ras (31)). In addition, the Mg ion essential for high affinity nucleotide binding is not present in either the GDP or GTP complex, contributing to the decrease in nucleotide affinity. These arguments also apply to the structure of the GRAB:Rab8 complex since it is essentially identical to that of Rabin8:Rab8 (Figure S2a and S2b). Structural changes occurring on interaction of Rab8 with Rabin8 are illustrated in Figure 6, Figure S1e and S1f. Starting from a disordered state of switch II in Rab8:GDP, a new -helix is generated by the interaction with Rabin8 in the GTPase:GEF complex. There is a large conformational change in switch I that has at least two specific effects on nucleotide binding: First, the side chain of Ile38 of Rab8 moves into the site commonly occupied by Mg, thus leading to displacement of the metal ion. Second, the conserved interaction between Phe33 of Rab8 and the guanine base is disrupted by Rabin8-mediated displacement of the amino acid side chain. The structure of the Rabin8:Rab:GTP complex (Figure 6b) is almost identical to that of the GDP complex, but a more extensive change in the conformation of switch I is required to generate the final conformation starting from the Rab8:GppNHp structure. In the latter, the switch I region closes over the nucleotide due to the interaction of the Thr40 side chain hydroxyl group and the backbone NH-group with the Mg ion and the -phosphate, respectively. The change in switch II conformation is less profound than in the transition from Rab8:GDP to the Rab8:Rabin8 complex, but a movement of the N-terminal region of switch II away from the nucleotide is still apparent. No complex structure containing both nucleotide and Mg could be obtained, since addition of nucleotides in the presence of Mg ions led to rapid loss of diffraction power and destruction of the crystals. DISCUSSION In the present work, we have investigated the GEF reactions of Rabin8 and GRAB with their substrate Rab8 in biochemical and structural detail. Kinetic analysis of the GEF Sec2-domain of GRAB and Rabin8 reveals that it possesses a moderately efficient catalytic activity. Unexpectedly, the affinity of the Rab-GEF-complex for nucleotides is relatively high (KD4= 0.26 μM for GDP), allowing the generation of ternary Rab8:Rabin8/GRAB:nucleotide complexes with GDP and GTP. The structures of these complexes allow us to describe the nucleotide exchange in detail at the molecular level. The mechanism of GDP displacement in the Rabin8/Rab8 system can be described as a disturbance of the structure of the nucleotide binding regions of Rab8, in particular switch I and switch II. In previously determined structures of GTPase:GEF complexes, the direct interaction of the GEF with switch II appears to be a universal feature. Contacts to switch I occur in many, but not all complexes, while contacts to the P-loop are less common but are seen in some instances. In Rabin8:Rab8, interaction of the C-terminal part of switch II leads to the formation of an -helix and displacement of Asp63 from its position near to the nucleotide binding site, where it is involved in an indirect (via a water molecule) interaction with Mg in the Rab8:GDP and Rab8:GppNHp structures. This interaction is highly conserved in the Ras superfamily. In the absence of GEFs, the highly conserved Gly66 interacts via its backbone NH with the -phosphate of GTP, but not with GDP. In the GDP displacement mechanism, the disturbance of the Mg coordination presumably contributes to destabilization of the metal ion at this site. Further destabilization arises from structural changes induced in switch I, leading to displacement of Mg by Ile38. Similar effects are seen in other GTPase:GEF complexes, with the residue displacing Mg coming either from the GTPase itself (32,33) or from the GEF (34,35). From earlier work on Rab7, removal of Mg leads to an accelby gest on N ovem er 9, 2017 hp://w w w .jb.org/ D ow nladed from GEF-mechanism of the Rab8-GEFs Rabin8/GRAB 6 eration of a factor of ca. 250 for GDP dissociation (36). Another universally conserved effect seen on interaction of GEFs with members of the Ras superfamily is the disturbance of the interaction of a highly conserved phenylalanine in switch I (Phe33 in Rab8) with the guanine base. In the case of Ras it was shown that mutation of this residue to isoleucine results in a ca. 100 fold increase in the rate constant for GDP dissociation (31). Assuming an additive contribution of both these mechanisms (i.e. removal of Mg and disruption of the Phe33 interaction) to the dissociation rate of GDP, this would imply an acceleration of ca. 2.5·10 for Rab8:GDP:Rabin8 relative to Rab8:GDP. The factor determined (k ́ ́-1/k ́ ́-4 in Scheme 1) is actually smaller (ca. 1.5·10), which could indicate that there is rate limitation for Mg dissociation or Phe33 movement. Rabin8 has a similar catalytic efficiency as other GEFs, with GDP-GTP exchange accelerations in the range of ca. 10 to 10 fold, but some GEFs are considerably faster. Among these are Cdc25 (for Ras; acceleration factor ca. 2·10), DrrA from Legionella pneumophila (for Rab1b; ca. 8·10 fold) and RCC1 (for Ran; ca. 1.3·10 fold) (26,27,37). It has recently been pointed out that in many GTPase-GEF-complexes of the Ras superfamily, the conserved lysine of the P-loop (Lys21 in Rab8, Lys16 in H-Ras), which interacts with the phosphate of GDP or GTP, is intermediately stabilized by a conserved glutamate in the G3-motif at the beginning of switch II (DxxGQE motif; Glu62 in Ras) (38). In a number of cases, substitution of this glutamate by alanine results in significant impairment of GEF activity. The corresponding residue is conserved in Rab molecules, but does not appear to be involved in such an interaction and is not essential for the exchange reaction (38). In the present work, the corresponding Rab8 Ploop lysine (Lys21) interacts with GTP, GDP or sulfate in the crystal structures of Rabin8:Rab8. Since sulfate is bound in the nucleotide-free state, we do not in fact have a structure corresponding to the genuine nucleotide free situation, so that we cannot make a definitive statement concerning the possible formation of an interaction Lys21 with Glu68 in the nucleotide-free state. Although we cannot present a bona fide nucleotide-free Rab8:Rabin8 structure (the sulfate presumably mimicks the -phosphate position), the -amino group of Lys21 is close to the carboxylate of Asp63 of the DxxGQE sequence and could thus interact with Lys21 in a hypothetical Rab8:Rabin8 complex devoid of nucleotides. This interaction (i.e. lysine of the P-loop with aspartate of the DxxGQE motif) is seen in at least 2 other Rab:GEF complexes (Rabex5:Rab21 and DrrA:Rab1b) as well as in the Ran:RCC1 complex. It therefore appears that nucleotide-free states of GTPase:GEF complexes are stabilized in part by interactions of the P-loop lysine with the glutamate of the DxxGQE motif, or with the aspartate or in some cases with both acidic residues. Rabin8 and GRAB are specific GEFs for Rab8, in contrast to the rather promiscuous protein MSS4 which shows GEF activity towards Rab1, Rab3, Rab10, and Rab13 in addition to Rab8 (25,39,40). MSS4 is a catalytically inefficient GEF with a kcat/Km value about four times lower than for Rabin8/GRAB. More importantly, the Rab8:MSS4 complex displays a very small second order rate constant of GTP-reassociation (25); therefore, the Rab8:MSS4 complex remains in the nucleotide-free form for a considerably longer time than other GEFs such as Rabin8/GRAB. It is believed that the weak GEF-properties of MSS4 are a result of its enzymatic mechanism in which nucleotide release is induced by a local protein unfolding reaction of Rab8 (25): In contrast to Rabin8/GRAB, MSS4 unfolds the entire switch I region together with -helix 1 of Rab8. This mechanism requires refolding of the nucleotide binding pocket prior to GTP-rebinding and thus nucleotide association is impaired. However, Sec2-domain GEFs (such as Rabin8 and GRAB) act differently from MSS4 since they stabilize the nucleotide-binding pocket by keeping switch I, switch II, and the P-loop in structurally defined confirmations that allow unimpaired rebinding of nucleotides (28,41). The work presented describes the structures of several intermediates in the exchange mechanism of Rabin8/GRAB with respect to Rab8. What is still missing is the structure of an intermediate in which Mg is bound. Such a structure has been described recently for the GEF:GTPase pair DOCK9:Cdc42 in the presence of GTP (42). Here, the amino acid homologous to Ile38 in Rab8 (a valine from the DOCK9 molecule) is displaced from its Mg occluding position by the presence of Mg and GTP, but not in the presence of Mg and GDP. The authors interpret this as a nucleotide sensor, leading to an alleged preference of the by gest on N ovem er 9, 2017 hp://w w w .jb.org/ D ow nladed from GEF-mechanism of the Rab8-GEFs Rabin8/GRAB 7 GEF for the GDP-state of the GTPase over the GTP-form and therefore allowing an unidirectional nucleotide exchange reaction. However, a similar effect is not observed in our structure and the mode of nucleotide-binding to Rab8:Rabin8/GRAB is identical for GDP and GTP. This is in keeping with the observation that there is no general preference for GEF interaction with the GDP or the GTP forms of GTPases, as required on theoretical grounds (14).