The structural basis of Rab:bMERB interaction
In order to understand the mode of interaction of Rab proteins and the C-terminal Rab-binding domain of Micals/EHBPs, we first aimed at determining the structure of the RBD of one member of these families. We succeeded in crystallizing a selenomethionine derivative of Mical-31841-1990 containing the whole predicted bMERB domain and solved the structure with a resolution of 2.7 Å (data and refinement statistics are shown in Table 2).
10.7554/eLife.18675.009Table 2. Data-collection and processing statistics (values in parentheses are for the outer shell).
DOI: http://dx.doi.org/10.7554/eLife.18675.009 *All data sets were collected from one single crystal on beamline X10SA of the Swiss Light Source (Paul Scherrer Institute, Villigen, Switzerland). †Data collections statistics for SAD data refer to unmerged Friedel pairs.
The asymmetric unit contains two copies of Mical-3 that form a central 4-stranded coiled-coil composed of α-helices 2 and 3 of each monomer flanked by α-helices 1 on both sides (Figure 4a). Interactions between the monomers mainly occur via two hydrophobic patches and some additional charged interactions (Figure 2b). Overall, the structure shows that each monomer consists of a central helix (α2, residues K1891-R1937) and N- and C-terminal helices folding back on this central helix.
10.7554/eLife.18675.010Figure 4. Structure of the Rab binding domain of Mical-3.
Mical-3 folds into three α-helices, the central α-helix 2 and α-helices 1 and 3 folding back on the central helix. The dimer observed in the asymmetric unit is formed mostly by hydrophobic interactions involving the same hydrophobic patches in both monomers, and α-helices 2 and 3 from each monomer form a central 4-stranded coiled-coil.
DOI: http://dx.doi.org/10.7554/eLife.18675.010
The completely α-helical fold of this protein is common to many Rab effector proteins and most of them bind the interacting Rab proteins via two α-helices (Mott and Owen, 2015). In order to test whether this is also true for the bMERB domains, we screened for crystallization conditions of bMERB:Rab complexes. Well diffracting crystals were found using the RBD of Mical-cL (residues 534–683) in complex with different Rab proteins (Rab1, Rab8 and Rab10). All structures were solved using a single chain of Mical-3 and the structure of Rab1 (pdb id 3nkv) or Rab8 (pdb id 4lhw) as search models for molecular replacement (the data collection and refinement statistics are shown in Table 2).
In all structures (Rab1:Mical-cL, Rab8:Mical-cL and Rab10:Mical-cL) we found one Rab protein bound to one molecule of Mical-cL (Figure 5a), in agreement with the previous observations that all Rab proteins tested bind only one site in Mical-cL. Most interactions were visible between the Rab proteins and α-helix 3 of Mical-cL with some additional contributing residues from α-helix 2, forming extensive contacts involving residues within switch I and II of the Rab proteins (Figure 5b). In all cases, hydrophobic interactions between the hydrophobic patch II in Mical-cL and residues from Rabs forming a hydrophobic pocket (residues Ile43, Phe70 and Ile73 in Rab8) and a triad of aromatic amino acids (Phe45, Trp62, Tyr77 in Rab8) known from all Rab:effector complexes solved to date (Itzen and Goody, 2011) were also observed in the Rab:Mical-cL structures.
10.7554/eLife.18675.011Figure 5. The specificity of Rab proteins binding to bMERB domains.
(a) A superposition of the complex structures of Rab1:Mical-cL, Rab8:Mical-cL, Rab10:Mical-cL and Rab1R8N:Mical-cL shows that Rabs bind Micals via their N-terminus (including RabSF1), RabSF2 as well as the switch regions (Rabs are shown in cartoon representation, switch I – red, switch II – blue, RabSF1 – orange, RabSF2, magenta; Micals are shown in cartoon representation and colored in dark green (Mical-cL interacting with Rab1), yellow (Mical-cL interacting with Rab8), light green (Mical-cL interacting with Rab10) or brown (Mical-cL interacting with Rab1R8N). The sequence alignments of different Rab proteins clearly shows that the interacting residues of Rab proteins with Micals (red residues) are highly conserved (orange residues) in Rab1 and Rab8 family members (Rab1a, Rab1b, Rab35, Rab8, Rab10, Rab13, Rab15), but not in other Rabs (below the black line). (b) In all structures of Rab proteins in complex with Mical-cL, the N-termini of the Rab proteins point towards a negatively charged patch of Mical-cL (Rabs are shown in cartoon representation as above; the surface of Mical-cL is colored by charge, red – negative charge, blue – positive charge). The sequence of the N-terminal residues of each Rab protein is shown below the corresponding structure: Whereas Rab1 contains a negatively charged glutamate at position 4, Rab8 and Rab10 contain one or two lysine residues at position 3 or at position 3 and 4, respectively. Consistently, the negatively charged N-terminus of Rab1 seems to repel α-helices 1 and 2 of Mical-cL and they adopt a conformation slightly further away from Rab1 compared to Rab8 and Rab10 (also see (a)). However, after mutating the 4 N-terminal residues of Rab1 to the corresponding sequence of Rab8 (the resulting chimera is called Rab1R8N), the structure of Rab1R8N:Mical-cL shows a similar conformation of α-helices 1 and 2 as in the structure of Rab8:Mical-cL. (c) Consistently, ITC measurements show that the affinity of binding increases approximately five-fold after mutating the N-terminal residues (Rab1:Mical-cL: KD = 5.2 µM; Rab1R8N:Mical-cL: KD = 1.1 µM; Rab8:Mical-cL: KD = 0.23 µM).
DOI: http://dx.doi.org/10.7554/eLife.18675.01110.7554/eLife.18675.012Figure 5—figure supplement 1. The N-termini of Rabs determine the specificity towards bMERB domains.
(a) Schematic presentation of the interactions between Rabs and Mical-cL (Switch I and II are shown in red and blue respectively; RabSF1, RabSF2 and RabF1 are shown in orange, magenta and green; Hydrophobic interactions are indicated by black dashed lines, ionic interactions and h-bonds are indicated by orange dashed lines). (b) Mutating the four N-terminal residues of Rab1 (1-MNPE…) to the sequence corresponding to Rab8 (1-MAKT…) leads to increased binding of EHBPs: Whereas Rab1 does not form a complex with EHBP1L1 (left) and EHBP1 (second from the right), the chimeric protein Rab1R8N forms a complex with both (EHBP1L1, second from the left; EHBP1, right) as assessed by aSEC.
DOI: http://dx.doi.org/10.7554/eLife.18675.012
10.7554/eLife.18675.013Figure 5—figure supplement 2. Comparison with other Rab:effector structures.
The main interacting helix (α3) in the structure of Rab10:Mical-cL adopts a very similar relative position as the main interacting helices of the effector proteins in the structures of Rab3:Rabphilin-3a (pdb 1ZBD) (Ostermeier and Brunger, 1999), Rab27:Slp2a (pdb 3BC1) (Chavas et al., 2008) and Rab27:Slac2-a (pdb 2ZET) (Kukimoto-Niino et al., 2008). Furthermore, a basic Arg and an acidic Asp are conserved in all structures (in Rab3:Rabphilin-3a, only the Arg is conserved) and contact the residues corresponding to Asp45 and Gln61 in Rab10.
DOI: http://dx.doi.org/10.7554/eLife.18675.013
Interestingly, the Rab-binding interface in Mical-cL has a substantial overlap with the dimer interface observed in the structure of Mical-3 above. Additionally, even though all Mical constructs used have a similar molecular weight of ~18 kDa, whereas Mical-1 runs as an apparent monomer in aSEC and binding of a Rab protein induces a clear shift to higher molecular weight, both Mical-3 and Mical-cL run as apparent dimers in aSEC and binding of a Rab protein disrupts the dimer, thus not leading to a shift in retention time upon complex formation (Figure 2—figure supplement 1). It is however not clear at this point whether the dimer formation of Mical-3 and Mical-cL and the disruption of the dimer upon Rab-binding is of functional significance.
The specificity of effector proteins towards certain Rab families is generally achieved via interaction with regions termed Rab subfamily motifs (RabSFs) 1–4 (Khan and Ménétrey, 2013; Moore et al., 1995; Pereira-Leal and Seabra, 2000). In the Rab:Mical-cL structures, we observed extensive interaction of Mical-cL with RabSF1 (Tyr6, Asp7, Leu9, Lys11 in Rab10) and (less interactions) with RabSF2 (Asp31, Ser40 in Rab10). Accordingly, the sequence alignment of different Rab proteins (Figure 5) shows strong conservation of the interacting amino acids within these motifs amongst Rab1 (Rab1a/b, Rab35) and Rab8 (Rab8a/b, Rab10, Rab13, Rab15) family members that interact with bMERB domains, but not for other Rab proteins (a comparative scheme of the residues involved in interactions in the different complexes is shown in Figure 5—figure supplement 1a). Since the interacting residues in the RabSF1 and RabSF2 regions are strongly conserved between both Rab1 and Rab8 families, this did not explain the observed preference of the RBD towards the Rab8 family. However, we observed that the main chain atoms of the N-terminal residues preceding the RabSF1 motif can be traced in the electron density and seem to interact with amino acids within α-helix 1 and 2 of Mical-cL, even though the electron density in this region did not allow a precise localization of the side chains. In contrast to the main interacting helix α3, which adopts a similar position in all three structures, we observed slightly different orientations of the α-helices 1 and 2, adopting a position slightly further away from Rab1 compared to Rab8 and Rab10 (Figure 5a). Interestingly, whereas Rab1 contains a glutamate near the N-terminus (Glu4), all Rab8 family members contain one or (in the case of Rab10) two lysine residues in this region that point towards a negatively charged patch in Mical-cL (Figure 5b). Additionally, Rab35 contains an Arg residue within this N-terminal region and also displays a slightly higher affinity towards Mical-cL compared to Rab1 (Figure 2—figure supplement 2). We therefore tested whether these N-terminal residues of the Rab proteins might determine the specificity of bMERB domains towards Rab8 and its homologues rather than Rab1. We constructed a chimera of Rab1 containing the 4 N-terminal aa of Rab8, thus exchanging the negatively charged glutamate for a positively charged lysine. The x-ray crystallographic structure of this Rab1 chimera (termed Rab1R8N) in complex with Mical-cL clearly showed that the helices 1 and 2 move closer and adopt a similar conformation as observed in the structures of Rab8:Mical-cL and Rab10:Mical-cL (Figure 5a). Additionally, ITC measurements showed that the chimera had an approximately five-fold increased binding affinity compared to Rab1 (Figure 5c). In contrast to Rab1, the chimera Rab1R8N bound both EHBP1 and EHBP1L1 in aSEC experiments (Figure 5—figure supplement 1b), thus clearly indicating that the N-terminus is important for the interaction and contributes to the observed specificity of bMERB domains towards Rab proteins.