1. Biochemistry
  2. Biophysics and Structural Biology
Download icon

Molecular mechanism for Rabex-5 GEF activation by Rabaptin-5

  1. Zhe Zhang
  2. Tianlong Zhang
  3. Shanshan Wang
  4. Zhou Gong
  5. Chun Tang
  6. Jiangye Chen
  7. Jianping Ding Is a corresponding author
  1. Institute of Biochemistry and Cell Biology, Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences, China
  2. Wuhan Institute of Physics and Mathematics, Chinese Academy of Sciences, China
Research Article
Cited
9
Views
2,143
Comments
0
Cite as: eLife 2014;3:e02687 doi: 10.7554/eLife.02687

Abstract

Rabex-5 and Rabaptin-5 function together to activate Rab5 and further promote early endosomal fusion in endocytosis. The Rabex-5 GEF activity is autoinhibited by the Rabex-5 CC domain (Rabex-5CC) and activated by the Rabaptin-5 C2-1 domain (Rabaptin-5C21) with yet unknown mechanism. We report here the crystal structures of Rabex-5 in complex with the dimeric Rabaptin-5C21 (Rabaptin-5C212) and in complex with Rabaptin-5C212 and Rab5, along with biophysical and biochemical analyses. We show that Rabex-5CC assumes an amphipathic α-helix which binds weakly to the substrate-binding site of the GEF domain, leading to weak autoinhibition of the GEF activity. Binding of Rabaptin-5C21 to Rabex-5 displaces Rabex-5CC to yield a largely exposed substrate-binding site, leading to release of the GEF activity. In the ternary complex the substrate-binding site of Rabex-5 is completely exposed to bind and activate Rab5. Our results reveal the molecular mechanism for the regulation of the Rabex-5 GEF activity.

https://doi.org/10.7554/eLife.02687.001

eLife digest

Cells need to allow various molecules to pass through the plasma membrane on their surface. Some molecules have to enter the cell, whereas others have to leave. Cells rely on a process called endocytosis to move large molecules into the cell. This involves part of the membrane engulfing the molecule to form a ‘bubble’ around it. This bubble, which is called an endosome, then moves the molecule to final destination inside the cell.

A protein called Rab5 controls how a new endosome is produced. However, before this can happen, various other molecules—including two proteins called Rabex-5 and Rabaptin-5—must activate the Rab5 protein. Exactly how these three proteins interact with each other was unknown.

Zhang et al. used X-ray crystallography to examine the structures of the complexes formed when Rabex-5 and Rabaptin-5 bind to each other, both when Rab5 is present, and also when it is absent. Biochemical and biophysical experiments confirmed that the Rabex-5/Rabaptin-5 complex is able to activate Rab5.

Zhang et al. also found that Rabex-5, on its own, folds so that the site that normally binds to Rab5 instead binds to a different part of Rabex-5, thus preventing endocytosis. However, when Rabaptin-5 forms a complex with Rabex-5, the Rab5 binding site is freed up.

The Rabex-5/Rabaptin-5 complex can switch between a V shape and a linear structure. Binding to Rab5 stabilizes the linear form of the complex, which then helps activate Rab5, and subsequently the activated Rab5 can interact with other downstream molecules, triggering endocytosis.

https://doi.org/10.7554/eLife.02687.002

Introduction

Endocytosis is a major process which eukaryotic cells use to absorb extracellular materials (Doherty and McMahon, 2009; Grant and Donaldson, 2009; Huotari and Helenius, 2011). In this process, small GTPase Rab5 functions as a master regulator of the early endosomal biogenesis (Stenmark, 2009; Mizuno-Yamasaki et al., 2012; Zeigerer et al., 2012). Rab5 is localized to early endosomal membrane via its isoprenylated C-terminus and regulates early endosomal fusion through interactions with an array of effectors including Rabaptin-5 (Stenmark et al., 1995), Rabenosyn-5 (Nielsen et al., 2000), EEA1 (Mills et al., 1998; Simonsen et al., 1998), PI3Ks (Li et al., 1995; Christoforidis et al., 1999), and APPLs (Miaczynska et al., 2004). Like other small GTPases, Rab5 exists mainly in two states, the GTP-bound active state and the GDP-bound inactive state, and requires guanosine nucleotide exchange factor (GEF) for activation and GTPase-activating protein (GAP) for inactivation.

Rabex-5 is a specific GEF for Rab5, Rab17, and Rab21 (Horiuchi et al., 1997; Delprato et al., 2004; Delprato and Lambright, 2007; Mori et al., 2013). The GEF domain is located in the middle and consists of a helical bundle (HB) domain and a Vps9 domain (Figure 1—figure supplement 1). Besides, the N-terminal region comprises two distinct ubiquitin-binding domains, a zinc finger domain and a motif interacting with ubiquitin domain, which can interact with ubiquitinated cargoes or adaptors to recruit Rabex-5 to early endosomal membrane (Lee et al., 2006; Mattera et al., 2006; Penengo et al., 2006; Mattera and Bonifacino, 2008) and function as an E3 ligase for Ras ubiquitination to promote Ras endosomal localization (Xu et al., 2010; Yan et al., 2010). The following membrane binding motif domain and the HB domain together compose an early endosomal targeting domain that can direct Rabex-5 to early endosomal membrane (Zhu et al., 2007). The C-terminal region consists of a coiled-coil (CC) domain and a proline rich region; the CC domain is involved in autoinhibition of the GEF activity and binding of Rabaptin-5 (Lippe et al., 2001; Mattera et al., 2006; Delprato and Lambright, 2007). Rabaptin-5 is a key effector of Rab5 and plays an important role in both homotypic and heterotypic fusions of early endosomes (Stenmark et al., 1995; Stenmark, 2009). It is a scaffold protein consisting of primarily four coiled-coil domains, namely C1-1, C1-2, C2-1, and C2-2 domains (Figure 1—figure supplement 1). The C2-1 domain is responsible for interaction with and recruitment of Rabex-5 to early endosomal membrane to activate Rab5 (Lippe et al., 2001; Mattera et al., 2006; Delprato and Lambright, 2007). Besides, the N-terminal region can mediate interactions with Rab4 and Rab8 (Vitale et al., 1998; Omori et al., 2008); the middle region can interact with the GAE and GAT domains of GGAs that function as effectors of the Arf family small GTPases in the tethering and fusion of trans Golgi network (TGN) (Mattera et al., 2003; Miller et al., 2003; Zhu et al., 2004a); and the C-terminal region can interact with the GTP-bound Rab5 that recruits Rabaptin-5 to early endosomal membrane (Vitale et al., 1998; Zhu et al., 2004b). In addition, Rabex-5 and Rabaptin-5 have been shown to function as neoplastic tumor suppressors and are implicated in human cancers (Magnusson et al., 2001; Wang et al., 2009; Christoforides et al., 2012; Thomas and Strutt, 2014), and Rabex-5 has also been shown to determine the neurite localization of its substrate Rab proteins and thus plays an important role in the development of hippocampal neurons (Mori and Fukuda, 2013; Mori et al., 2013).

Previous structural, biochemical, and biological data have demonstrated that Rabex-5 and Rabaptin-5 function together to activate Rab5 in endocytosis; the GEF activity of Rabex-5 could be autoinhibited by its CC domain and activated by binding of the Rabaptin-5 C2-1 domain (Rabaptin-5C21) to the CC domain (Lippe et al., 2001; Delprato and Lambright, 2007; Zhu et al., 2007, 2010). However, the underlying molecular mechanism is unclear. We report here the crystal structures of a Rabex-5 variant in complex with the dimeric Rabaptin-5C21 (Rabaptin-5C212) and in complex with Rabaptin-5C212 and Rab5. The structural data together with the in vitro functional data reveal the molecular mechanism for the regulation of the Rabex-5 GEF activity.

Results

Structure of the Rabex-5Δ-Rabaptin-5C212 complex

To investigate the molecular mechanism of the regulation of the Rabex-5 GEF activity, we were intent to determine the crystal structures of Rabex-5 containing the GEF and CC domains (residues 132–455, Rabex-5) alone and in complex with the Rabaptin-5 C2-1 domain (residues 552–642, Rabaptin-5C21). We were able to obtain Rabex-5 and the Rabex-5-Rabaptin-5C21 (R2) complex with high purity, stability, and homogeneity, but unfortunately failed to grow any crystals for either Rabex-5 or the R2 complex. Partial digestion of the R2 complex with trypsin shows that Rabex-5 could be proteolyzed in the linker between the GEF and CC domains (Figure 1—figure supplement 2), indicating that the linker is surface exposed with high flexibility which may prevent proper crystal packing. Thus, we constructed a series of Rabex-5 variants containing different forms of linker deletion (Rabex-5Δ). Although none could be crystallized alone, one Rabex-5Δ variant (residues 132–455Δ393–407) led to successful crystallization of the Rabex-5Δ-Rabaptin-5C21 (R2Δ) complex.

The crystal structure of the Rabex-5Δ-Rabaptin-5C21 complex was determined at 3.10 Å resolution (Table 1), containing one Rabex-5Δ and two Rabaptin-5C21 or one Rabex-5Δ-Rabaptin-5C212 complex per asymmetric unit (Figure 1A and Figure 1—figure supplement 3A). The N- and C-terminal regions of each Rabaptin-5C21 form two α-helices linked together by a short loop to assume a ‘V’ shaped conformation with an inclination angle of about 40°; and the N- and C-terminal α-helices of the two Rabaptin-5C21 dimerize with each other to form two two-helix bundles. In addition, two symmetry-related complexes further dimerize through the N-terminal α-helices of Rabaptin-5C212 (Figure 1B).

Table 1

Summary of diffraction data and structure refinement statistics

https://doi.org/10.7554/eLife.02687.003
Rabex-5CCRabex-5CC-Rabaptin-5C212Rabex-5Δ-Rabaptin-5C212Rab5-Rabex-5Δ-Rabaptin-5C212
Diffraction data
 Wavelength (Å)0.92001.00000.97930.9785
 Space groupP21C2P3121P41212
 Cell parameters
a (Å)46.890.087.2174.8
b (Å)40.328.987.2174.8
c (Å)51.6108.0168.9149.0
 α (°)90.090.090.090.0
 β (°)95.1102.290.090.0
 γ (°)90.090.0120.090.0
 Resolution (Å)50.0–2.0050.0–2.2050.0–3.1050.0–4.60
(2.07–2.00)*(2.28–2.20)(3.21–3.10)(4.76–4.60)
 Observed reflections38,44547,48279,255124,340
 Unique reflections (I/σ(I) > 0)12,74813,81613,73012,699
 Average redundancy3.0 (3.0)3.4 (3.0)5.8 (6.0)9.8 (9.0)
 Average I/σ(I)23.6 (14.0)21.2 (3.4)20.1 (2.4)18.2 (2.8)
 Completeness (%)96.4 (97.7)97.7 (85.8)98.1 (100.0)97.6 (95.8)
Rmerge (%)5.3 (9.3)6.0 (28.0)8.2 (64.3)11.7 (94.3)
Refinement and structure model
 Reflections (Fo ≥ 0σ(Fo))
 Working set11,43712,43310,80611,982
 Test set622691601631
Rwork/Rfree (%)19.1/23.419.3/23.526.4/31.525.1/34.3
 No. of atoms1726181430749679
 Protein1621162730749679
 Water105187
 Average B factor (Å2)
 All atoms22.858.572.0187.3
 Main-chain atoms17.951.872.6186.9
 Side-chain atoms25.964.070.7187.7
 Water34.963.5--
 RMS deviations
 Bond lengths (Å)0.0180.0140.0050.015
 Bond angles (°)1.611.371.271.87
 Ramachandran plot (%)
 Most favored99.599.592.193.8
 Allowed0.50.57.65.8
 Generously allowed0.00.00.30.5
  1. *

    Numbers in parentheses represent the highest resolution shell.

  2. Rmerge = ∑hkl∑i|Ii(hkl)i−<I(hkl)>|/∑hkl∑iIi(hkl).

  3. R = ∑hkl‖Fo|−|Fc‖/∑hkl|Fo|.

Figure 1 with 5 supplements see all
Crystal structure of the Rabex-5Δ-Rabaptin-5C212 complex.

(A) A ribbon representation of the overall structure of the Rabex-5Δ-Rabaptin-5C212 complex. The HB and Vps9 domains of the Rabex-5 GEF domain are colored in light blue and dark blue, respectively, and the Rabex-5 CC domain (Rabex-5CC) in green. The two Rabaptin-5C21 are designated with superscripts A and B, and colored in orange and dark yellow, respectively. The disordered loops of the HB domain (residues 149, 161–162, 174, 190–204, and 220–230) and the linker between the GEF and CC domains (residues 369–412Δ393-407) are indicated with dotted lines. The autoinhibitory residues of Rabex-5CC are marked in red. (B) The dimeric Rabex-5Δ-Rabaptin-5C212 complex. (C) An electrostatic surface representation of the amphipathic α-helix of Rabex-5CC. The autoinhibitory residues are located on the N-terminal of the nonpolar surface as indicated with a yellow circle. (D) Interactions between Rabex-5CC and Rabaptin-5C212. Left panel: Rabex-5CC is shown in ribbon representation and Rabaptin-5C212 in electrostatic surface representation. Right panel: Rabex-5CC is shown in electrostatic surface representation and Rabaptin-5C212 in ribbon representation. The interacting residues are shown with side chains. The detailed interactions is available in the Figure 1—source data 1. (E) An electrostatic surface representation of the interactions between the GEF and CC domains of Rabex-5Δ. (F) A close-up view of the interactions of the GEF domain with Rabex-5CC and Rabaptin-5C21. The interacting residues are shown with side chains and the hydrogen bonds are indicated with dotted lines.

https://doi.org/10.7554/eLife.02687.004

Rabex-5CC (residues 413–452) forms a long amphipathic α-helix (about 60 Å) (Figure 1A–D) that is in agreement with the prediction by Delprato and Lambright (2007). It packs in parallel with the C-terminal α-helices of Rabaptin-5C212 to form a tight three-helix bundle with its nonpolar surface buried in a hydrophobic surface groove of Rabaptin-5C212. The interactions are dominantly hydrophobic that bury a total of solvent accessible surface area of 2664 Å2. The residues that were suggested to be involved in the autoinhibition of the GEF activity, including Asn413, Leu414, Leu417, Leu420, Arg423, and Ile427 (Delprato and Lambright, 2007), are located in the N-terminal half of the nonpolar surface, and several of them (Leu420, Arg423, and Ile427) are involved in the interactions with Rabaptin-5C212 and buried in the interaction interface (Figure 1C,D).

The HB domain (residues 132–229) and Vps9 domain (residues 230–368) of the Rabex-5 GEF domain are well defined except a few surface exposed loops (Figure 1A). The C-terminal helix αC and the following linker (residues 369–412 with the deleted residues 393–407) are also disordered, consistent with our trypsin digestion results (Figure 1—figure supplement 2). The distance between the visible C-terminal end of the GEF domain and N-terminal end of the CC domain is about 10 Å, which is large enough to accommodate the disordered 29 residues with a loop conformation, suggesting that the positions and conformations of the GEF and CC domains are unlikely constrained by the shortened linker.

The overall structure of the GEF domain is very similar to that in the free form (Delprato et al., 2004) and in complex with Rab21 (Delprato and Lambright, 2007) (RMSD of ∼0.90 Å for 228 Cα atoms) (Figure 1—figure supplement 3B). At the substrate-binding site, there is a surface groove composed of largely nonpolar residues, which exhibits good chemical and geometrical complementarities with the nonpolar surface of the amphipathic helix of Rabex-5CC (Figure 1E). The GEF domain packs along the three-helix bundle formed by Rabex-5CC and Rabaptin-5C212 (Figure 1A). The interactions involve a small portion of the substrate-binding site, a small portion of the N-terminal region of Rabex-5CC adjacent to the nonpolar surface, and a small portion of the N-terminal region of one Rabaptin-5C21 C-terminal α-helix (Figure 1F). The interaction interface buries a total of solvent accessible surface area of 1040 Å2, which is much smaller than that between Rabex-5 and Rab21 (2400 Å2) (Delprato and Lambright, 2007). Rab21 uses switch I, switch II, and the interswitch region to interact with the substrate-binding site of Rabex-5 (Delprato and Lambright, 2007). Although the binding sites for switch II and a small portion of the interswitch region of Rab21 are occupied by the three-helix bundle, the binding sites for switch I and a large portion of the interswitch region of Rab21 are exposed to the solvent (Figure 1—figure supplement 3C). Hence, we consider that the substrate-binding site of Rabex-5 is largely exposed to the solvent and partially accessible by the substrate.

To explore the conservations of the conformations of Rabex-5CC and Rabaptin-5C21 and the interactions between Rabex-5CC and Rabaptin-5C21, we also determined the crystal structures of Rabex-5CC alone at 2.00 Å resolution and in complex with Rabaptin-5C212 at 2.20 Å resolution (Table 1). In the Rabex-5CC structure, Rabex-5CC also forms a long α-helix (about 65 Å) and the four Rabex-5CC in the asymmetric unit form a tight four-helix bundle via the nonpolar surface (Figure 1—figure supplement 4). In the structure of the Rabex-5CC-Rabaptin-5C212 complex, the asymmetric unit contains one complex (Figure 1—figure supplement 5). Notably, each Rabaptin-5C21 forms a long α-helix (about 125 Å) and two Rabaptin-5C21 dimerize in parallel to form a twisted linear two-helix bundle, which is different from the V-shaped conformation in the R2Δ complex. Otherwise, two symmetry-related complexes also dimerize through the N-terminal regions of Rabaptin-5C212. Rabex-5CC also forms a long α-helix (52 Å) and packs in parallel with the C-terminal regions of Rabaptin-5C212 to form a tight three-helix bundle. The interactions between Rabex-5CC and Rabaptin-5C212 are essentially the same as those in the R2Δ complex. These results confirm that Rabex-5CC assumes a stable amphipathic α-helix and tends to bury its nonpolar surface via interactions with other proteins; Rabaptin-5C21 may adopt two different conformations and forms a stable homodimer; and the interactions between Rabex-5CC and Rabaptin-5C212 are conserved in different complexes.

Structure of the Rab5-Rabex-5Δ-Rabaptin-5C212 complex

In the structure of the R2Δ complex, the substrate-binding site of Rabex-5 is partially occupied by Rabex-5CC and Rabaptin-5C212, implying that further conformational change(s) of the R2Δ complex will be required to bind Rab5. To investigate the molecular mechanism for Rab5 activation by the R2 complex, we solved the crystal structure of the Rab5-Rabex-5Δ-Rabaptin-5C21 (R3Δ) complex to 4.60 Å resolution (Table 1; Figure 2A, and Figure 2—figure supplement 1). The asymmetric unit contains two Rab5-Rabex-5Δ-Rabaptin-5C212 complexes related by a twofold non-crystallographic symmetry. The four Rabaptin-5C21 and two Rabex-5 are well defined in the electron density map; however, only one Rab5 is fairly defined while the other is poorly defined, indicating that the bound Rab5 has high flexibility which may explain the poor diffraction quality of the crystal. No nucleotide and/or metal ion are found at the active site of Rab5 and thus the bound Rab5 is in nucleotide-free form.

Figure 2 with 2 supplements see all
Crystal structure of the Rab5-Rabex-5Δ-Rabaptin-5C212 complex.

(A) A ribbon representation of the overall structure of the Rab5-Rabex-5Δ-Rabaptin-5C212 complex. Only one Rab5-Rabex5Δ-Rabaptin5C212 complex in the asymmetric unit is shown with Rab5 in purple and Rabex-5 and Rabaptin-5C21 in the same colors as in Figure 1A. (B) Comparison of the Rab5-Rabex-5Δ-Rabaptin-5C212 complex and the Rabex-5Δ-Rabaptin-5C212 complex based on superposition of the three-helix bundle formed by Rabex-5CC and the C-terminal regions of Rabaptin-5C212. (C) Interactions between the Rabex-5 GEF domain and Rabaptin-5C212. The GEF domain is shown in ribbon representation in blue and the interacting residues are shown with side chains. Rabaptin-5C212 is shown in surface representation with the interacting residues colored in pink. (D) Interactions between Rab5 and the Rabex-5 GEF domain. Rab5 is shown in coil representation with the P-loop, switch I, switch II, and interswitch region colored in purple, orange, blue, and dark green, respectively. Several key residues are shown with side chains. The GEF domain is shown in surface representation with Asp313 colored in yellow. For comparison, Rab21 in its complex with the Rabex-5 GEF domain (Delprato and Lambright, 2007) is shown in coil representation in light blue.

https://doi.org/10.7554/eLife.02687.012

In the R3Δ complex, each Rabaptin-5C21 forms a long α-helix; two of them form a twisted linear two-helix bundle; and the two Rabaptin-5C212 dimerize through the middle regions (residues 590–600) (Figure 2A and Figure 2—figure supplement 1A). The conformation of Rabaptin-5C212 is similar to that in the Rabex-5CC-Rabaptin-5C212 complex and the previously reported GAT-Rabaptin-5C212 complex (Zhu et al., 2004a) but significantly different from that in the R2Δ complex (Figure 2—figure supplement 2). Similar to that in the R2Δ complex, Rabex5CC also forms a long α-helix and interacts with the C-terminal regions of Rabaptin-5C212 to form a three-helix bundle (Figure 2A). However, the orientations and positions of the Rabex-5 GEF domain and the N-terminal regions of Rabaptin-5C212 in relation to the three-helix bundle are dramatically different (Figure 2B). When the two complexes are superimposed based on the three-helix bundle, the N-terminal regions of Rabaptin-5C212 rotates downwards by about 120° to transform from the V-shaped conformation to the linear conformation. Meanwhile, the GEF domain rotates by about 270° along the vertical axis and about 100° along the horizontal axis, and is dislodged from the three-helix bundle without any interaction. As a result, the substrate-binding site is completely exposed to the solvent for Rab5 binding.

The GEF domain of Rabex-5 interacts via a small portion of the opposite side of the substrate-binding site with a small portion of the N-terminal regions of Rabaptin-5C212 (Figure 2A). The Rabex-5-Rabaptin-5C21 interaction involves only Glu232, Leu236, Gln239, Arg243, and Arg246 of αV1 and Val269 and Ser280 of αV2 of the GEF domain, and Asn568, Glu572, Gln579, and Glu582 of one Rabaptin-5C21 and Ile608 and Asp623 of the other, and the interaction interface buries a total of solvent accessible surface area of 1700 Å2 (Figure 2C). The overall structure of the nucleotide-free Rab5 differs from the nucleotide-bound Rab5 in the conformations of the P-loop and the switch regions (Merithew et al., 2001; Zhu et al., 2004b). Nevertheless, the interactions between Rabex-5 and Rab5 are similar to those between Rabex-5 and Rab21, suggesting that Rabex-5 may activate Rab5 via a similar mechanism as for Rab21 (Figure 2D) (Delprato and Lambright, 2007; Langemeyer et al., 2014).

Solution structure of the Rabex-5-Rabaptin-5C212 complex

In the R2Δ complex, Rabaptin-5C212 assumes a V-shaped conformation and the substrate-binding site of the Rabex-5 GEF domain is partially occupied by Rabex-5CC and Rabaptin-5C212 (Figure 1, Figure 1—figure supplement 3C). However, in the R3Δ complex, Rabaptin-5C212 adopts a linear conformation and the Rabex-5 GEF domain is displaced with a completely exposed substrate-binding site (Figure 2A,B). A modeling study shows that if Rabaptin-5C212 assumes the linear conformation in the R2Δ complex, the middle regions of Rabaptin-5C212 would have steric conflict with part of the Rabex-5 GEF domain (αV1, αV6, and αC helices), suggesting that the conformational change of Rabaptin-5C212 is essential for the complete exposure of the substrate-binding site of the GEF domain to bind Rab5. To investigate which conformation the R2 complex may assume in solution, we performed small angle X-ray scattering (SAXS) analyses of the R2, R2Δ, R3, and R3Δ complexes.

Our SAXS data show that the experimental P(r) distributions for R2Δ and R3Δ are similar to these of R2 and R3, respectively (Figure 3—figure supplement 1A,B), suggesting that deletion of the linker does not significantly affect the structures of R2Δ and R3Δ. The maximum paired-distance of the particle (Dmax), the radius of gyration (Rg), and the Porod volume derived from the SAXS data together show that all of these complexes exist as dimers in solution, consistent with our biochemical and structural data (Figure 3—source data 1). The theoretical scattering curve calculated from the R2Δ model with the V-shaped conformation fits better with the experimental data for both R2 and R2Δ complexes (goodness of fit χ = 0.71 and 0.60, respectively) than that calculated from the R2Δ model with the linear conformation (goodness of fit χ = 0.74 and 0.65, respectively) (Figure 3A,B). In addition, although the theoretical and experimental P(r) distributions exhibit some differences, the theoretical P(r) distribution of the R2Δ model with the V-shaped conformation agrees better with the experimental P(r) distributions of both R2 and R2Δ complexes than that with the linear conformation (Figure 3—figure supplement 1A,C). As such, the solution structures of the R2 and R2Δ complexes can be best described as assuming mainly the V-shaped conformation. Nevertheless, it is plausible that the two segments of the V-shaped conformation and/or the component proteins of the complexes may bear some flexibility and adopt alternative conformation(s) in solution. Based on these results, we conclude that the R2 and R2Δ complexes assume mainly the V-shaped conformation as observed in the R2Δ structure with some flexibility in solution.

Figure 3 with 1 supplement see all
SAXS analyses of the R2 and R3 complexes.

(A and B) Comparison of the experimental data with the theoretical scattering curves calculated from the structure models of R2Δ with the V-shaped conformation observed in the R2Δ structure and the linear conformation observed in the R3Δ structure for the R2 complex (A) and the R2Δ complex (B). (C and D) Comparison of the experimental data with the theoretical scattering curve calculated from the structure model of R3Δ observed in the R3Δ structure for the R3 complex (C) and the R3Δ complex (D). The observed and calculated values of Rg, Dmax, and Porod volume are summarized in Figure 3—source data 1.

https://doi.org/10.7554/eLife.02687.015

Similarly, the theoretical scattering curve calculated from the R3Δ model fits well with the experimental SAXS data for both R3 and R3Δ complexes (goodness of fit χ = 0.68 and 0.96, respectively) (Figure 3C,D), and the theoretical P(r) distribution of the R3Δ model is also in good agreement with the experimental P(r) distributions of both R3 and R3Δ complexes (Figure 3—figure supplement 1B,D). These results indicate that the R3 and R3Δ complexes assume mainly the linear conformation as observed in the R3Δ structure in solution.

Functional analyses of the Rabex-5-Rabaptin-5 complex in Rab5 activation

To investigate the biological relevance of the R2Δ and R3Δ structures, we performed in vitro functional assays. We first measured the in vitro GEF activity of different Rabex-5 variants and mutants. Our kinetic data show that Rabex-5 containing the GEF and CC domains possesses a basal GEF activity (0.93 ± 0.03 × 104 M−1·s−1); the Rabex-5 GEF domain alone exhibits a 3.2-fold higher activity (2.93 ± 0.06 × 104 M−1·s−1); and the R2 complex exhibits a 3.3-fold higher activity (3.07 ± 0.08 × 104 M−1·s−1) (Figure 4A,B). Moreover, the Rabex-5 mutants containing mutations on the nonpolar surface of Rabex-5CC exhibit relatively higher GEF activity (1.4–1.9 folds) compared with the wild-type Rabex-5 (Figure 4—figure supplement 1A). These results indicate that the GEF domain itself is constitutively active; the CC domain slightly autoinhibits the GEF activity; and the binding of Rabaptin-5C21 to Rabex-5CC relieves the autoinhibition, which are largely in agreement with the previous biochemical data (Lippe et al., 2001; Delprato et al., 2004; Zhu et al., 2007; Langemeyer et al., 2014). However, the magnitude of the autoinhibition by Rabex-5CC is smaller than that reported by Delprato and Lambright (2007), which may be caused by differences of the assay systems, for examples, different Rab5 and Rabex-5 constructs, different concentration of the proteins, and different sensitivity of the instruments.

Figure 4 with 1 supplement see all
In vitro functional analyses of the Rabex-5-Rabaptin-5 complex.

(A) GEF activity of Rabex-5 in different forms. Catalytic efficiency (kcat/Km) was obtained from the slope of a linear least-squares-fit of the kobs values against the concentrations of Rabex-5 from two independent measurements. (B) Histogram of the catalytic efficiencies of Rabex-5 variants alone and in complexes with different Rabaptin-5C21 mutants or variant. Values are means ± SEM of two independent measurements. R2M1: the R2 complex in which Rabaptin-5C21 contains a quadruple mutation N568A/E572A/Q579A/E582A; R2M2: the R2 complex in which Rabaptin-5C21 contains a double mutation I608A/D623A; R2ΔN: the R2 complex in which the N-terminal half of Rabaptin-5C21 (residues 552–592) is deleted. The complexes in (A) and (B) were co-expressed and co-purified. (C) GST pull-down assays for the interactions between the wild-type and mutant GST-Rabex-5 and the wild-type His6-Rabaptin-5C21. The gel was stained by Coomassie blue. (D) GST pull-down assays for the interactions between the wild-type GST-Rabex-5 and the wild-type and mutant His6-Rabaptin-5C21. (E) Histogram of the catalytic efficiencies of the wild-type and mutant Rabex-5 alone and in complexes with the wild-type Rabaptin-5C21. (F) Histogram of the catalytic efficiencies of the wild-type Rabex-5 alone and in complexes with the wild-type and mutant Rabaptin-5C21. For the assays in (E) and (F), Rabex-5 and Rabaptin-5C21 were expressed and purified separately and then mixed together in a 1:2 molar ratio overnight prior to the assay. Tables of the GEF activities are available in the Figure 4—source data 1–3.

https://doi.org/10.7554/eLife.02687.018

In the R2Δ complex, the linker between the Rabex-5 GEF and CC domains was removed to facilitate the crystallization. We then investigated whether the linker deletion has any effects on the functions of Rabex-5 and the R2 complex. Our kinetic data show that Rabex-5Δ possesses a slightly higher activity (1.9-fold) than Rabex-5 and the R2Δ complex exhibits a similar activity (0.9-fold) as the R2 complex (Figure 4A,B). Meanwhile, our in vitro GST pull-down assay results show that Rabex-5Δ can bind tightly to Rabaptin-5C21; however, deletion of either the N- or C-terminal half of Rabex-5CC disrupts the interaction (Figure 4—figure supplement 1B). These results indicate that the linker plays a minor role in the autoinhibition of the GEF activity but is not involved in the Rabex-5-Rabaptin-5C21 interaction, and the linker deletion does not affect the function of the R2 complex in the Rab5 activation.

In the R2Δ and Rabex-5CC-Rabaptin-5C212 complexes, the interactions between Rabex-5CC and Rabaptin-5C21 are well conserved. To validate the biological relevance of these interactions, we mutated several key residues of both Rabex-5CC and Rabaptin-5C21 at the interaction interface and analyzed their effects on the Rabex-5-Rabaptin-5 interaction. Our in vitro GST pull-down assay results show that mutations L434D, L438D, and W441A of Rabex-5CC, and mutations L599D, L610D, L613D, and L617D of Rabaptin-5C21 abolish the interaction, and mutations L420D and I427D of Rabex-5CC and mutation V624D of Rabaptin-5C21 substantially impair the interaction. In contrast, mutations R423E of Rabex-5CC and E607K of Rabaptin-5C21 have no significant effect on the interaction as these two residues form a salt bridge on the solvent-exposed surface and thus their mutations do not affect the hydrophobic core of the interaction. As negative controls, mutations I439D of Rabex-5CC and I608D of Rabaptin-5C21 have no effect on the interaction as these two residues are located on the polar surface of the three-helix bundle and are not involved in the interaction (Figure 4C,D). In agreement with the GST pull-down results, our kinetic data show that the GEF activity of the L434D, L438D, and W441A Rabex-5 mutants cannot be activated by Rabaptin-5C21, whereas that of the I439D Rabex-5 mutant can be potentiated by Rabaptin-5C21 (Figure 4E). Similarly, the L610D, L613D, and L617D Rabaptin-5C21 mutants cannot relieve the autoinhibition of Rabex-5; whereas the I608D Rabaptin-5C21 mutant can activate the GEF activity of Rabex-5 (Figure 4F). It is noteworthy that due to partial aggregation of Rabaptin-5C21, the GEF activity of the mixed R2 complex is only 1.6-fold higher compared with Rabex-5 which is weaker than the co-expressed and co-purified R2 complex (3.3-fold).

In the R3Δ complex, the Rabex-5 GEF domain interacts via a small surface opposite the substrate-binding site with a small portion of the N-terminal regions of Rabaptin-5C212 (Figure 2C). To explore the functional role of this interaction, we constructed two Rabaptin-5C21 mutants containing a quadruple mutation (N568A/E572A/Q579A/E582A) and a double mutation (I608A/D623A) of the residues on the interaction interface and detected their effects on the GEF activity of the R2 complex. Our kinetic data show that the GEF activity of these mutant complexes are unaffected (Figure 4B), indicating that this interaction is not essential for the activation of the Rabex-5 GEF activity. Intriguingly, when the N-terminal half of Rabaptin-5C21 (residues 552–592) was removed, the GEF activity of Rabex-5 could not be activated (Figure 4B), indicating that the full-length Rabaptin-5C21 is required for the function of the R2 complex in the Rab5 activation. Taken together, these data indicate that the structures of the R2Δ and R3Δ complexes are functionally relevant, and the interaction between Rabex-5 and Rabaptin-5 is important for the activation of the Rabex-5 GEF activity in Rab5 activation, which is in accord with the previous biochemical data (Lippe et al., 2001).

Discussion

The previous biochemical and biological data showed that Rabex-5 functions together with Rabaptin-5 to activate Rab5 and then to promote the fusion of early endosomes in endocytosis (Lippe et al., 2001; Delprato and Lambright, 2007; Zhu et al., 2007). The GEF activity of Rabex-5 is autoinhibited by its CC domain and is activated by the binding of Rabaptin-5 via its C2-1 domain. However, the molecular mechanism is unknown. In this work, we determined the crystal structures of Rabex-5Δ in complex with Rabaptin-5C21 and in complex with Rabaptin-5C21 and Rab5, which are validated by biophysical and biochemical analyses.

Our structural data show that at the substrate-binding site of the Rabex-5 GEF domain, there is a surface groove composed of largely nonpolar residues. Rabex-5CC forms a stable amphipathic α-helix that tends to bury its nonpolar surface via oligomerization or interaction with the C-terminal regions of Rabaptin-5C212. The nonpolar surface of Rabex-5CC has good chemical and geometrical complementarities with the nonpolar surface groove of the GEF domain and thus might be able to bind there to block the substrate binding and hence autoinhibit the GEF activity as proposed by Delprato and Lambright (2007). Nonetheless, our structural and biochemical data show that although Rabex-5CC alone exists as a stable helix in both solution and crystal structure, it cannot form a stable complex with the GEF domain as shown by both GST pull-down assay and ITC analysis (data not shown). In addition, Rabex-5 and Rabex-5Δ cannot be crystallized alone and Rabex-5 can be easily proteolyzed in the linker region (Figure 1—figure supplement 2). These results suggest that the CC domain and the linker have high flexibility. Moreover, Rabex-5 itself has a basal GEF activity which is slightly weaker (about 1/3) than that of the constitutively active GEF domain (Figure 4A,B) but is not so weak compared with some other GEFs including DSS4 (a GEF for Ypt1p) (Esters et al., 2001) and MSS4 (a GEF for Rab8) (Zhu et al., 2001; Itzen et al., 2006). Mutations of the residues on the nonpolar surface of Rabex-5CC can enhance the GEF activity by 1.4–1.9 folds (Figure 4—figure supplement 1A). These results together indicate that the binding of Rabex5CC to the GEF domain is not tight, and Rabex-5 alone is not completely autoinhibited.

On the other hand, in the structure of the R2Δ complex, Rabaptin-5C212 forms 2 two-helix bundles with a V-shaped conformation, and Rabex-5CC interacts via the nonpolar surface with the C-terminal regions of Rabaptin-5C212 to form a tight three-helix bundle. Meanwhile, the GEF domain folds along the three-helix bundle with a partially occupied substrate-binding site, suggesting that further conformational change is required to completely expose the substrate-binding site for Rab5 binding and activation. Indeed, in the structure of the R3Δ complex, although Rabex-5CC still forms a tight three-helix bundle with the C-terminal regions of Rabaptin-5C212, Rabaptin-5C212 forms a linear two-helix bundle which is different from the V-shaped conformation but similar to that in the Rabex-5CC-Rabaptin-5C212 complex and the GAT-Rabaptin-5C212 complex (Zhu et al., 2004a). The GEF domain is dislodged from the three-helix bundle and interacts with the N-terminal regions of Rabaptin-5C212, and the substrate-binding site is completely exposed to bind Rab5. Meanwhile, our SAXS analysis results indicate that the R2 and R2Δ complexes mainly assume the V-shaped conformation as observed in the R2Δ structure but have some flexibility in solution, and our biochemical data show that the R2 and R2Δ complexes have full GEF activities as the constitutively active GEF domain and the N-terminal regions of Rabaptin-5C212 are required for the function of the R2 complex in the activation of Rab5.

Based on the structural and biological data in this work and those reported previously, we can propose the molecular mechanism for the regulation of the Rabex-5 GEF activity despite the lack of an intact Rabex-5 structure (Figure 5). In the free-form Rabex-5, Rabex-5CC may bind weakly via the nonpolar surface to the substrate-binding site of the GEF domain, leading to the blockage of the substrate-binding site and thus a weak autoinhibition of the GEF activity (Figure 5, State I). As the binding of Rabex-5CC to the GEF domain is not tight, it might assume alternative conformations. The linker between the GEF and CC domains might help to modulate the conformational flexibility and/or the relative conformations of the two domains and thus plays a minor role in the autoinhibition. One plausible alternative conformation of Rabex-5CC might be similar to that observed in the R2Δ structure with a largely exposed substrate-binding site of the GEF domain. As the interaction of Rabex-5CC with the GEF domain via the nonpolar surface is much tighter than that via the adjacent surface, Rabex-5CC might exist mainly in the autoinhibitory conformational state and partially in the alternative conformational state. This may explain why the free-form Rabex5 exhibits some basal GEF activity. Rabaptin-5C21 might bind to the C-terminal region of the nonpolar surface of Rabex-5CC in the autoinhibitory conformational state and induce conformational change of the α-helix to transform into the conformational state as observed in the R2Δ structure, or directly to the exposed nonpolar surface of Rabex-5CC in the alternative conformational state, leading to the relief of the autoinhibition and the release of the GEF activity (Figure 5, State II).

Molecular mechanism of the regulation of the Rabex-5 GEF activity.

In the free-form Rabex-5, the CC domain binds weakly via the nonpolar surface to the substrate-binding site of the GEF domain, leading to occlusion of the substrate-binding site and thus a weak autoinhibition of the GEF activity (State I). The free-form Rabex-5 can directly target to the early endosomes to activate Rab5 at the basal level. In the cells, most Rabex-5 forms a binary complex with Rabaptin-5 which can be recruited to early endosomes via the binding of the C-terminal region of Rabaptin-5 to GTP-bound Rab5. The binding of Rabaptin-5C21 to Rabex-5 pulls Rabex-5CC away from the GEF domain to form a binary complex with the V-shaped conformation and a largely exposed substrate-binding site, leading to the relief of the autoinhibition (State II). The binding of Rab5 induces further conformational changes of the Rabex-5-Rabaptin-5 complex such that Rabaptin-5C212 transforms from the V-shaped to the linear conformation, and the substrate-binding site of the GEF domain is completely exposed to the solvent to bind and activate Rab5 (State III). The positive feedback loop among Rab5, its effector Rabaptin-5, and its GEF Rabex-5 can lead to a robust activation of Rab5, which then promotes the fusion of early endosomes efficiently.

https://doi.org/10.7554/eLife.02687.023

When the Rabex-5-Rabaptin-5 complex is recruited to the early endosomal membrane via the interaction of the C-terminal region of Rabaptin-5 with the GTP-bound Rab5, Rabex-5 can activate Rab5 locally in a very efficient way. The binding of the GDP-bound Rab5 to Rabex-5 induces further conformational changes of the Rabex-5-Rabaptin-5 complex such that Rabaptin-5C212 transforms from the V-shaped to the linear conformation, the GEF domain is dislodged from Rabex-5CC and the C-terminal regions of Rabaptin-5C212, and the substrate-binding site is completely exposed to the solvent to bind Rab5 as observed in the R3Δ structure (Figure 5, State III). In this conformational state, the Rabex-5-Rabaptin-5 complex can facilitate the exchange of GDP- to GTP-bound Rab5. The not-so-tight interaction of the Rabex-5 GEF domain with Rabex-5CC and Rabaptin-5C212 as observed in the R2Δ structure allows the conformational changes easily during the substrate binding. The conformational changes of the Rabex-5-Rabaptin-5 complex induced by the binding of Rab5 leading to the full activation of the Rabex-5 GEF activity may provide another leverage to ensure the high substrate specificity.

In the context of the early endosomal membrane, Rabex-5 can directly target to the early endosomes either through the interaction of the N-terminal ubiquitin binding domain (UBD) with the ubiquitinated cargoes or through the early endosomal targeting domain (EET) (Zhu et al., 2007; Mattera and Bonifacino, 2008). In this case, as the GEF activity of Rabex-5 is autoinhibited by Rabex-5CC, Rabex-5 can only activate Rab5 at the basal level. In the cells, most Rabex-5 forms a stable complex with Rabaptin-5 which can be recruited to early endosomes via the binding of Rabaptin-5 through the C-terminal region to GTP-bound Rab5 (Lippe et al., 2001; Zhu et al., 2010). In this case, Rabaptin-5 not only assists the recruitment of Rabex-5 to the early endosomal membrane, but also activates the GEF activity of Rabex-5. The positive feedback loop among Rab5, its effector Rabaptin-5, and its GEF Rabex-5 can lead to a robust activation of Rab5, which then promotes the fusion of early endosomes efficiently.

As a scaffold protein, Rabaptin-5 comprises several coiled-coil regions which can mediate interactions with different proteins to exert different functions. In addition to acting as the effector of Rab5 to function in the fusion of early endosomes, Rabaptin-5 can interact with Rab4 via the N-terminal region and thus may serve as the effector of Rab4 to function in the endocytic recyling process (Vitale et al., 1998). As the C2-1 domain of Rabaptin-5 can interact with the CC domain of Rabex5 and the GAT domain of GGA1, and Rabaptin-5 is the effector of Rab5 and GGA1 is the effector of Arf1, this dual interaction might mediate the crosstalk between Rab and Arf GTPases to promote the tethering and fusion of early endosomes and trans-Golgi network-derived vesicles (Mattera et al., 2003; Zhu et al., 2004a; Kawasaki et al., 2005).

In the GAT-Rabaptin-5C212 structure, Rabaptin-5C212 assumes a linear conformation, the N-terminal regions of Rabaptin-5C212 bind one GAT, and two symmetry-related Rabaptin-5C212 dimerize through the C-terminal regions (Zhu et al., 2004a), which are the binding site for Rabex-5CC. Interestingly, Rabaptin-5C212 assumes a linear conformation in the Rabex-5CC-Rabaptin-5C212 structure but a V-shaped conformation in the R2Δ structure, and in both structures, the C-terminal regions of Rabaptin-5C212 bind one Rabex-5CC and two symmetry-related Rabaptin-5C212 dimerize through the N-terminal regions which are the binding site for GAT. Moreover, in the R3Δ structure, Rabaptin-5C212 also assumes a linear conformation but two Rabaptin-5C212 dimerize through the middle regions (residues 590–600) between the GAT-binding site and the Rabex-5CC-binding site. As the GAT-binding site is unoccupied, Rabaptin-5 may bind Rabex-5 and GGA1 simultaneously. These results demonstrate that Rabaptin-5C21 always forms a dimer which can assume either a V-shaped or a linear conformation and can bind other proteins via different regions. In addition, the dimeric Rabaptin-5C212 prefers to further dimerize to form a dimer of dimers via the regions that are not involved in the interactions with other proteins. In the context of full-length Rabaptin-5, it is possible that Rabaptin-5C212 might exist in either the V-shaped or the linear conformation depending on the functional state. The dimerization of Rabaptin-5C212 might not only avoid exposure of the hydrophobic surface and minimize the overall energy of the protein in aqueous environment, but also play some functional roles in the tethering and fusion of early endosomes and/or with other vesicles.

Materials and methods

Cloning, expression, and purification of proteins

The cDNAs corresponding to the Rabex-5 CC domain (residues 409–455, Rabex-5CC), the Rabex-5 GEF domain (residues 132–392), the Rabex-5 GEF and CC domains (residues 132–455, Rabex-5), the Rabaptin-5 C2-1 domain (residues 552–642, Rabaptin-5C21), and Rab5 (residues 15–184) were all amplified by PCR from the cDNA library of human brain cells. The Rabex-5CC and Rabex-5 GEF ORFs were cloned into the pET-28a plasmid (Novagen, Germany) with a His6 tag inserted at the N-terminus. The Rab5 ORF was cloned into a modified pET-28a plasmid (Novagen) with a His6-sumo tag inserted at the N-terminus. The Rabex-5CC and Rabaptin-5C21 ORFs, and the Rabex-5 and Rabaptin-5C21 ORFs were cloned into the pET-Duet1 plasmid (Novagen) with a His6 tag inserted at the N-terminus of Rabex-5CC and Rabex-5, respectively. The Rabex-5Δ variants containing different deletion forms of the linker between the GEF and CC domains were generated using the Takara MutanBEST Mutagenesis kit (TakaRa Biotechnology, Japan). The Rabex-5 and Rabaptin-5C21 mutants containing point mutations were generated using the QuikChange Site-Directed Mutagenesis kit (Agilent Technologies, Santa Clara, CA).

All recombinant proteins were expressed in E. coli BL21 (DE3) Codon-Plus strain (Novagen). The transformed cells were grown at 37°C in LB medium containing 0.05 mg/ml ampicillin or kanamycin until OD600 reached 0.8, and then induced with 0.25 mM IPTG at 16°C for 24 hr. All the proteins were purified by affinity chromatography using a Ni-NTA column (Qiagen, Germany) and gel filtration chromatography using a Superdex 200 16/60 column (GE Healthcare, Sweden) in a buffer containing 20 mM Tris–HCl, pH 8.0, 150 mM NaCl, and 1 mM PMSF. The resultant samples were of >95% purity as evaluated by SDS-PAGE.

Trypsin digestion analysis

A trypsin stock solution (2.5 mg/ml) was diluted to 10−1 to 10−6 times. The Rabex-5-Rabaptin-5C21 complex (1 mg/ml) was mixed with the trypsin solution of different concentrations. The digestion reaction proceeded for 30 min at 4°C and 16°C, respectively, and was then stopped by addition of 10 μg/ml aprotinin to inhibit the activity of trypsin. The reaction mixture was loaded onto Ni-NTA beads, and both the beads and the flow-through were analyzed by SDS-PAGE with Coomassie blue staining and Western blot with anti-His antibody (1:3000, TIANGEN, China).

In vitro GST pull-down assay

For in vitro GST pull-down assay, the Rabaptin-5C21 ORF was cloned into the pET-3E-His plasmid (Novagen) with an N-terminal His6 tag, and the Rabex-5 ORF into the pGEX 6P-1 plasmid (GE Healthcare) with an N-terminal GST tag. His6-Rabaptin-5C21 was purified by Ni-NTA affinity chromatography and GST-Rabex-5 by glutathione sepharose beads (GE Healthcare). 20 μg GST-Rabex-5 immobilized onto the glutathione sepharose beads were incubated with 100 μg His6-Rabaptin-5C21 at 4°C for 2 hr. The beads were analyzed by SDS-PAGE with Coomassie blue staining.

Nucleotide exchange assay

The Rabex-5 GEF activity for Rab5 was determined using the method described previously (Delprato et al., 2004). Briefly, Rab5 was mixed with 20-fold excess fluorescent 2’(3’)-bis-O-(N-methylanthraniloyl)-GDP (mantGDP, Invitrogen, Carlsbad, CA). The mixture was incubated for 2 hr and the free mantGDP was removed by gel filtration using a HiTrap De-salting column (GE Healthcare). The mantGDP-bound Rab5 was diluted to 500 nM in a buffer containing 20 mM Tris–HCl (pH 8.0), 150 mM NaCl, and 2 mM MgCl2. Nucleotide exchange reaction was initiated by addition of GTP to a final concentration of 1 mM and varied concentrations (50–500 nM) of Rabex-5 or Rabex-5-Rabaptin-5C21. Dissociation of mantGDP was monitored by measuring the decrease of fluorescence. Samples were excited at 360 nm and the emission was monitored at 440 nm. Fluorescence data were recorded using a Varian Cary Eclipse spectrofluorimeter (Agilent Technologies). Observed pseudo first-order exchange rate constant (kobs) was obtained by a nonlinear least-squares-fit of the data at each concentration of Rabex-5 to the exponential equation

I(t)=(I0I)exp(kobst)+I

where I(t) is the emission intensity at time t, I0 the initial emission intensity, and I the final emission intensity. Catalytic efficiency (kcat/Km) was obtained from the slope of a linear least-squares-fit of the kobs values to the linear equation

kobs=(kcat/Km)[Rabex5]+kintr

where kintr is the intrinsic nucleotide exchange rate in the absence of Rabex-5. The intrinsic exchange rate (kintr) of Rab5 is measured to be 0.00064 ± 0.00002 s−1.

Crystallization, data collection, and structure determination

Crystallization was performed using the hanging drop vapor diffusion method at 16°C by mixing equal volumes (1.0 μl) of protein solution (20 mg/ml) and reservoir solution. Crystals of the R2Δ complex were grown from drops consisting of the reservoir solution of 2.0 M NaH2PO4/K2HPO4 (pH 7.0) and 0.05% n-octyl-β-D-galactopyranoside. Crystals of Rabex-5CC were grown from drops consisting of the reservoir solution of 0.10 M NaAc (pH 5.4), 17.5% MPD, and 2% PEG4000. Crystals of the Rabex-5CC-Rabaptin-5C21 complex were grown from drops consisting of the reservoir solution of 0.15 M MgAc2 and 20% PEG3350. Crystals of the R3Δ complex were grown from drops consisting of the reservoir solution of 1.0 M NaH2PO4/K2HPO4 (pH 5.0). All of the diffraction data were collected at −175°C at beamline 17U of Shanghai Synchrotron Radiation Facility, and processed with HKL2000 (Otwinowski and Minor, 1997).

The structure of the R2Δ complex was solved using the molecular replacement (MR) method as implemented in Phenix (Adams et al., 2010) with the structure of the Rabex-5 GEF domain (PDB code 1TXU) (Delprato et al., 2004) as the search model. The structure of the Rabex-5CC-Rabaptin-5C21 complex was solved by MR with the structure of the Rabaptin-5 C2-1 domain (PDB code 1X79) (Zhu et al., 2004a) as the search model. The structure of Rabex-5CC was solved by MR with the structure of Rabex-5CC in its complex with Rabaptin-5C21 as the search model. The structure of the R3Δ complex was solved by MR with the structure of the Rabex-5 GEF-Rab21 complex (PDB code 2OT3) (Delprato and Lambright, 2007) as the search model.

Structure refinement was carried out using Phenix (Adams et al., 2010), Refmac5 (Murshudov et al., 1997), and CNS (Brunger, 2007), and model building using Coot (Emsley and Cowtan, 2004). Due to the low resolution of the diffraction data, the structure models of Rab5, Rabex-5, and Rabaptin-5C21 in the R3Δ complex were refined as rigid bodies with deformable elastic network and group B-factor restraints (Schroder et al., 2010) and thus the side-chain orientations in this complex are somewhat uncertain. Stereochemistry of the structure models was analyzed using Procheck (Laskowski et al., 1993). Structural analyses were carried out using programs in CCP4 (Winn et al., 2011) and the PISA server (Krissinel and Henrick, 2007). All structure figures were generated using PyMOL (http://www.pymol.org). The statistics of the structure refinement and final structure models are summarized in Table 1.

Small angle X-ray scattering (SAXS) analysis

Protein samples were concentrated to 5 mg/ml in 20 mM Tris–HCl (pH 8.0) and 150 mM NaCl. Solution scattering experiments were performed at 293 K on a SAXSess mc2 platform (Anton Paar, Austria) equipped with a sealed tube source and a CMOS diode array detector. The SAXS data were collected with 2 hr exposure time in 1-hr frame to ensure absence of radiation damage during the course of the experiment. The SAXS data for the buffer were recorded for background subtraction. Inverse Fourier transformation was performed with the GIFT program in the PCG software package (Anton Paar). The maximum paired-distance (Dmax) value was extrapolated from the P(r) distribution. The radius of gyration (Rg) and the Porod volume were calculated using PRIMUS (Konarev et al., 2003) at the low angle region (q × Rg ≤1). The theoretical P(r) distribution and Rg value for each structure model were calculated using PTRAJ from the AMBER12 package (Case et al., 2012). The structure model was assessed against the corresponding solution scattering data using CRYSOL from the ATSAS software (Svergun et al., 1995) with constant subtraction. The his6-tag and disordered residues were built back into the crystal structures using PyMOL (http://www.pymol.org), and the yielded structure models were optimized using Xplor-NIH (Schwieters et al., 2006) for optimal packing. For each crystal structure, a total of 800 structure models were generated with Monte Carlo simulated annealing while fixing the coordinates of the atoms observed in the crystal structure.

Accession codes

The crystal structures of Rabex-5CC, the Rabex-5CC-Rabaptin-5C212 complex, the Rabex-5Δ-Rabaptin-5C212 complex, and the Rab5-Rabex-5Δ-Rabaptin-5C212 complex have been deposited with the Protein Data Bank under accession codes 4N3X, 4N3Y, 4N3Z, and 4Q9U, respectively.

References

  1. 1
  2. 2
  3. 3
  4. 4
  5. 5
  6. 6
  7. 7
  8. 8
  9. 9
    Coot: model-building tools for molecular graphics
    1. P Emsley
    2. K Cowtan
    (2004)
    Acta Crystallographica Section D, Biological Crystallography 60:2126–2132.
    https://doi.org/10.1107/S0907444904019158
  10. 10
  11. 11
  12. 12
  13. 13
  14. 14
  15. 15
  16. 16
  17. 17
  18. 18
  19. 19
  20. 20
  21. 21
  22. 22
  23. 23
  24. 24
  25. 25
  26. 26
  27. 27
  28. 28
  29. 29
  30. 30
  31. 31
  32. 32
  33. 33
  34. 34
  35. 35
  36. 36
  37. 37
  38. 38
  39. 39
  40. 40
  41. 41
  42. 42
  43. 43
  44. 44
  45. 45
  46. 46
  47. 47
  48. 48
  49. 49
  50. 50
  51. 51
  52. 52
  53. 53
  54. 54
  55. 55
  56. 56

Decision letter

  1. Suzanne R Pfeffer
    Reviewing Editor; Stanford University, United States

eLife posts the editorial decision letter and author response on a selection of the published articles (subject to the approval of the authors). An edited version of the letter sent to the authors after peer review is shown, indicating the substantive concerns or comments; minor concerns are not usually shown. Reviewers have the opportunity to discuss the decision before the letter is sent (see review process). Similarly, the author response typically shows only responses to the major concerns raised by the reviewers.

Thank you for sending your work entitled “Molecular mechanism of the activation of Rab5 by Rabex-5 and Rabaptin-5 in endocytosis” for consideration at eLife. Your article has been favorably evaluated by a Senior editor and three reviewers, one of whom, Suzanne Pfeffer, is a member of our Board of Reviewing Editors. The Reviewing editor and the other reviewers discussed their comments before we reached this decision, and the Reviewing editor has assembled the following comments to help you prepare a revised submission.

Zhang et al. report crystal structures of: the Rabex-5 GEF and coiled coil domains with an internal flexible loop deleted (Rabex-5delta) in complex with the coiled coil Rabex-5 binding regions of Rabaptin-5 (Rabaptin-5C212); the coiled coil region of Rabex-5 alone; the latter in complex with Rabex-5 binding region of Rabaptin-5; and a low resolution ternary complex of Rabex-5delta:Rabaptin-5C212 bound to nucleotide free Rab5. Collectively, these structures represent an important contribution to the field by defining the mode of interaction between the coiled coil regions of Rabex-5 and Rabaptin-5, identifying alternative conformational states and interactions with the substrate site that may have a functional role, and providing an overview of the quaternary organization of the active complex with a substrate Rab. The structural information is complemented by kinetic, binding, and cell morphologic analyses of the constructs used for the crystallographic studies as well as the effects of various site specific mutants. Finally, the authors also investigated whether the alternative conformations are compatible with SAXS experiments in solution.

1) In the kinetic assays, how was the intrinsic exchange rate for Rab5 taken into account? It should be roughly 0.0005 s-1; however, the fitted lines in Figure 3A all appear to extrapolate to zero. Also, the concentration range analyzed is somewhat narrow (50-150 nM or 3 fold). A broader range may be necessary to measure both high and low catalytic efficiencies. For example, low catalytic efficiencies may be poorly estimated in the range appropriate for high catalytic efficiencies and a higher/broader range may be required to distinguish catalyzed from intrinsic exchange. Since the longest Rabex-5 construct is used as a reference for the mutants but has the weakest activity, it is important to measure its activity in an appropriate concentration range.

2A) The statement on that “the P(r) functions of the R2 and R2delta complexes exhibit a saddle-shaped peak especially for R2, suggesting that both R2 and R2delta complexes assume more than one conformation in solution” is confusing. How can the presence of multiple conformations be inferred from features in the P(r) distribution without having the P(r) distribution of the relevant individual conformations for comparison? The presence of multiple peaks in the P(r) is not unexpected for proteins with multiple domains or proteins that form oligomeric complexes (see, for example, Putnam et al., 2007, Quart. Rev. Biophys 40, 191-285). Before ascribing significance to specific features in the P(r) distributions for the experimental data, it would seem important to compare them with the theoretical P(r) distributions for the R2 and R2delta structures.

2B) If a constant was not subtracted when fitting the data in Crysol, does constant subtraction improve the fits such that one of the conformations can adequately describe the data? A better analysis (e.g. with Bunch) would include the his-tag and any other disordered regions. A more complicated model with two conformations may fit the data better but that doesn't prove that the model is correct. It is important to account for missing elements before concluding that a more complicated model is required.

2C) Dmax is not the only measure of molecular size and its estimation can be confounded by tags or disordered regions. Is the Porod volume (should be approximately half the molecular mass) also consistent with a dimer? Is the Guinier constant Rg larger than expected for monomers? Does it correspond to the calculated Rg for either the linear or V-shaped structures?

2D) The description of the methods used for SAXS analysis is incomplete. E.g., what program was used to calculate P(r)? What procedure was used to determine Dmax? Was constant subtraction used when fitting the data in Crysol?

3) The contribution of rabaptin-5 to membrane targeting of Rabex-5 is not fully considered. Previous work from the Zerial lab has demonstrated synergy between Rabex-5 and rabaptin-5. A simple explanation for this would be that rabaptin-5 binding both promotes or stabilizes membrane association of Rabex-5 and stimulates its activity by relieving auto-inhibition. Such a model would fit with the modest activation of core GEF activity seen on rabaptin binding. To test this idea the localization of Rabex-5 point mutants with compromised rabaptin interaction would be informative. This should be done at low expression levels, but is not essential for publication.

4) Figure 3 the text claims in vivo functional assays were performed. This is the weakest area of the manuscript. The images in Figure 3G and Figure 3–figure supplement 3 show Rab5 localization in BHK cells transfected with Rabex-5 constructs in the presence of the endogenous Rabex-5. 2. The experiments in 3G depend upon equal expression of Rab5 as well as the various Rabex constructs introduced and a stable Rab5 cell line would be most appropriate here. In any case, co-staining for the constructs and quantitation of expression levels would add confidence to the findings reported. Quantitations on endosome size are mentioned in the Figure 3–figure supplement 3 legend text but it is not clear how such numbers could be obtained from the data presented here. The staining could be Golgi from the images presented. First, the images don't show any obvious endosome staining and also require another marker. Second, these experiments need to be done in the absence of endogenous Rabex-5. Finally, some form of functional assay for Rab5 dependent endocytic transport should be performed, for example growth factor or transferrin uptake. The best way to fix this would be to rescue cells depleted of Rabex-5 and monitor endocytosis after transfection with various rescue constructs.

Other comments:

1) Figure 1 is very busy and confusing at first glance. It is not immediately obvious that the green helix in Figure 1A is part of Rabex-5 because the labeling overlaps the structure. The presentation and labeling should be improved.

2) The text should make it clearer that nucleotide-free Rab5 was present in the crystal structure. This is only mentioned later nd described in a way that makes it appear surprising Rab5 is in nucleotide-free form. Taking the work of Delprato and Lambright (2007) and Langemeyer (2014) together one would have expected Rab5 and Rab21 to be activated by the same mechanism and to show the same interactions to Rabex-5 when in nucleotide-free form.

3) Figure 2 should show only one complex and not both copies in the overall structure for clarity. The text indicates that one Rab5 is not well defined, so this could be shown in the supplement rather than the main figure.

4) The Discussion and model should better explain how the complex might interact with and sit at a membrane surface given the known interactions of Rabex-5 via its UIM and of rabaptin with GGAs and Rabs. 3. Also, the coiled coil of Rabaptin seems to exist as an either bent form or elongated form in solution and can form a 3-helix bundle with RABEX5 to open the enzyme. Not clear is whether this coiled coil is freely available to behave that way within the context of the full length Rabaptin-5 protein and this needs to be discussed and included in Figure 5. Also not clearly specified is why Lambright saw more dramatic inhibition and relief thereof, by the Rabaptin domain (please add this to the Discussion).

5) The Title is not a correct description of the work. The authors uncover details of the way in which rabaptin binding relieves auto inhibition of Rabex-5. It does not provide the mechanism of Rab5 activation in detail or look at endocytosis.

6) The statement that the “the substrate-binding site is largely exposed to the solvent and accessible by the substrate” appears to be contradicted by the previous sentence, which states that “the binding sites for switch II and a small portion of the interswitch region of Rab21 are occupied by the three-helix bundle” and by Figure 1A, 1B and Figure 1–figure supplement 4 in which the substrate binding site appears to be blocked by the three helix bundle.

7) Abstract: “Rabaptin-5C212” should be defined on first usage. In particular, clarify what the “2” after “Rabaptin-5C21” signifies.

8) Several abbreviations (e.g., R2, R2delta, R3, etc) are missing from the list. The authors should decrease jargon wherever possible as there are many acronyms here. Thus, please define R2 complex? R3 complex? Constructs used in 3G? It is hard for the reader.

9) Finally, because the assignment and orientation of side chains shown in Figure 2C and 2D are determined primarily by the molecular replacement models and only a little bit by the data, please add a qualifying note about the uncertainty of the side chain orientations due to the low resolution of the structure.

https://doi.org/10.7554/eLife.02687.024

Author response

1) In the kinetic assays, how was the intrinsic exchange rate for Rab5 taken into account? It should be roughly 0.0005 s-1; however, the fitted lines in Figure 3A all appear to extrapolate to zero. Also, the concentration range analyzed is somewhat narrow (50-150 nM or 3 fold). A broader range may be necessary to measure both high and low catalytic efficiencies. For example, low catalytic efficiencies may be poorly estimated in the range appropriate for high catalytic efficiencies and a higher/broader range may be required to distinguish catalyzed from intrinsic exchange. Since the longest Rabex-5 construct is used as a reference for the mutants but has the weakest activity, it is important to measure its activity in an appropriate concentration range.

We thank the reviewers for the constructive suggestion. In the revision, to broaden the concentration range of Rabex-5, we have measured the exchange rates of different Rabex-5 constructs at three additional concentrations of 200 nM, 350 nM, and 500 nM, and fitted the entire kinetic data to the linear equation “kobs = (kcat/Km)[Rabex-5]+kintr” to derive the catalytic efficiencies (kcat/Km) from the slope of the line. The new results are provided in Figure 4, Figure 4–figure supplement 1A, and Figure 4–Source data 1-3. The newly obtained catalytic efficiencies are consistent with our previous results. In addition, we have measured the intrinsic exchange rate (kintr) of Rab5 and as expected by the reviewers, the intrinsic exchange rate (kintr) of Rab5 is 0.00064 ± 0.00002 s-1. In the revision, we have elaborated the nucleotide exchange assay method in the Materials and methods section (Nucleotide exchange assay) accordingly.

2A) The statement on that “the P(r) functions of the R2 and R2delta complexes exhibit a saddle-shaped peak especially for R2, suggesting that both R2 and R2delta complexes assume more than one conformation in solution“ is confusing. How can the presence of multiple conformations be inferred from features in the P(r) distribution without having the P(r) distribution of the relevant individual conformations for comparison? The presence of multiple peaks in the P(r) is not unexpected for proteins with multiple domains or proteins that form oligomeric complexes (see, for example, Putnam et al., 2007, Quart. Rev. Biophys 40, 191-285). Before ascribing significance to specific features in the P(r) distributions for the experimental data, it would seem important to compare them with the theoretical P(r) distributions for the R2 and R2delta structures.

2B) If a constant was not subtracted when fitting the data in Crysol, does constant subtraction improve the fits such that one of the conformations can adequately describe the data? A better analysis (e.g. with Bunch) would include the his-tag and any other disordered regions. A more complicated model with two conformations may fit the data better but that doesn't prove that the model is correct. It is important to account for missing elements before concluding that a more complicated model is required.

We thank the reviewers for pointing out the misstatement about the saddle-shaped peak. As these two comments are related, we respond to them together. In the revision, we have re-analyzed the SAXS data using the methods suggested by the reviewers and revised the text accordingly. As suggested by the reviewers, we have added the his6-tag and the disordered residues back into the crystal structures, and the yielded structure models were optimized and assessed against the experimental SAXS data. We have also performed constant subtraction during the fitting.

The experimental P(r) distribution for the R2Δ mutant is similar to that of R2 (Figure 3–figure supplement 1A), suggesting that deletion of the linker does not affect the structure of R2Δ in solution significantly.

As shown in Figure 3A, B, the theoretical scattering curve calculated from the R2Δ model with the V-shaped conformation fits better with the experimental SAXS data for both R2 and R2Δ complexes (goodness of fit χ= 0.71 and 0.60, respectively) than that calculated from the R2Δ model with the linear conformation (goodness of fit χ= 0.74 and 0.65, respectively). In addition, we compared the experimental P(r) distributions for the R2 and R2Δ complexes with the theoretical P(r) distributions of the R2Δ models with the V-shaped conformation and the linear conformation (Figure 3–figure supplement 1A, C). Although the theoretical and experimental P(r) distributions exhibit some differences, the theoretical P(r) distribution of the R2Δ model with the V-shaped conformation agrees better with the experimental P(r) distributions of both R2 and R2Δ complexes than that with the linear conformation. As such, the solution structures of R2Δ and R2 can be best described as assuming mainly the V-shaped conformation. Nevertheless, it is plausible that the two segments of the V-shaped conformation and/or the component proteins of the complex may bear some flexibility and adopt alternative conformation(s) in solution. As pointed out by the reviewers, a more complicated model with two or more conformations may fit the experimental data better but that does not prove the correctness of the model. Based on these data as well as the values of the radius of gyration (Rg), the maximum paired-distance (Dmax), and the Porod volume, we conclude that the R2 and R2Δ complexes assume mainly the V-shaped conformation with some flexibility in solution.

2C) Dmax is not the only measure of molecular size and its estimation can be confounded by tags or disordered regions. Is the Porod volume (should be approximately half the molecular mass) also consistent with a dimer? Is the Guinier constant Rg larger than expected for monomers? Does it correspond to the calculated Rg for either the linear or V-shaped structures?

Based on the structure model of R2Δ, we also computed the theoretical radius of gyration (Rg) and maximum paired-distance (Dmax), and compared them with the experimental values. Based on the assessments, the R2Δ dimer with the V-shaped conformation would be best to describe the solution structures of R2 and R2Δ. In addition, the experimental Porod volumes for both R2 and R2Δ(210 nm3 and 200 nm3, respectively) are in agreement with the molecular masses of R2 and R2Δ in dimeric form (118 kDa and 114 kDa, respectively). The observed and calculated values of Rg, Dmax, and Porod volume are now summarized in Figure 3–source data 1.

2D) The description of the methods used for SAXS analysis is incomplete. E.g., what program was used to calculate P(r)? What procedure was used to determine Dmax? Was constant subtraction used when fitting the data in Crysol?

We have now elaborated the SAXS analysis procedures in the Materials and methods section (Small angle X-ray scattering analysis).

3) The contribution of rabaptin-5 to membrane targeting of Rabex-5 is not fully considered. Previous work from the Zerial lab has demonstrated synergy between Rabex-5 and rabaptin-5. A simple explanation for this would be that rabaptin-5 binding both promotes or stabilizes membrane association of Rabex-5 and stimulates its activity by relieving auto-inhibition. Such a model would fit with the modest activation of core GEF activity seen on rabaptin binding. To test this idea the localization of Rabex-5 point mutants with compromised rabaptin interaction would be informative. This should be done at low expression levels, but is not essential for publication.

We concur with the reviewers that the Rabaptin-5 binding to Rabex-5 both promotes or stabilizes the membrane association of Rabex-5 and stimulates its activity by relieving the autoinhibition. To test this idea, as suggested by the reviewers, we have observed the colocalization of wild-type or mutant Rabaptin-5 and Rabex-5 in HEK 293 cells co-transfected with GFP-Rab5, Myc-Rabaptin-5, and Flag-Rabex-5. As shown in Figure 5A, we showed that both Rabex-5 (residues 132–455) and Rabex-5Δ (residues 132–455Δ393–407) colocalized well with Rabaptin-5 to the Rab5-positive early endosomes. However, in the cells co-transfected the Rabex-5 GEF domain with Rabaptin-5 or the Rabex-5 mutant containing a triple mutation L434D/L438D/W441A (Rabex-5 3M) with the Rabaptin-5 mutant containing a triple mutation L610D/L613D/L617D (Rabaptin-5 3M), although Rabaptin-5 can still bind to Rab5-positive early endosomes, the Rabex-5 GEF domain or the Rabex-5 mutant cannot be recruited to the early endosomes due to the disruption of the Rabex-5-Rabaptin-5 interaction. These results indicate that the interaction between Rabex-5 and Rabaptin-5 is important not only for the activation of the Rabex-5 GEF activity but also for the recruitment of Rabex-5 to the early endosomal membrane, which is in accord with the previous work from the Zerial lab and the expectation of the reviewers.

4) Figure 3 the text claims in vivo functional assays were performed. This is the weakest area of the manuscript. The images in Figure 3G and Figure 3–figure supplement 3 show Rab5 localization in BHK cells transfected with Rabex-5 constructs in the presence of the endogenous Rabex-5. 2. The experiments in 3G depend upon equal expression of Rab5 as well as the various Rabex constructs introduced and a stable Rab5 cell line would be most appropriate here. In any case, co-staining for the constructs and quantitation of expression levels would add confidence to the findings reported. Quantitations on endosome size are mentioned in the Figure 3–figure supplement 3 legend text but it is not clear how such numbers could be obtained from the data presented here. The staining could be Golgi from the images presented. First, the images don't show any obvious endosome staining and also require another marker. Second, these experiments need to be done in the absence of endogenous Rabex-5. Finally, some form of functional assay for Rab5 dependent endocytic transport should be performed, for example growth factor or transferrin uptake. The best way to fix this would be to rescue cells depleted of Rabex-5 and monitor endocytosis after transfection with various rescue constructs.

We thank the reviewers for this criticism and the constructive suggestions. As the Rab5 localization and function in BHK 21 cells were not obvious in our previous experiments, we obtained another GFP-Rab5 plasmid from Dr. Lan Bao’s laboratory in our institute which has been used in their researches and has shown to have good early endosomal localization, and re-performed the in vivo functional assays in HEK 293 cells.

We concur with the reviewers that our cell biological analyses might depend upon equal expression of Rab5 as well as the various Rabaptin-5 and Rabex-5 constructs introduced, and thus a stable Rab5 cell line would be most appropriate. However, we do not possess a stable Rab5 cell line from available resources. Thus, in the cell staining analyses, we co-transfected the cells with equal amounts of GFP-Rab5, Myc-Rabaptin-5, and Flag-Rabex-5 plasmids and used immunoblotting assays to examine the expression levels of the three proteins. Three independent experiments showed similar expression levels of these proteins (Figure 5C), indicating that the analysis method works properly.

According to the reviewers’ suggestion, we have co-stained all of the three proteins to detect their expression and colocalization when analyzing the size of early endosomes (Figure 5A). To confirm that the GFP-Rab5-marked vesicles are early endosomes, we co-stained the early endosomal marker EEA1. The representative images showed that the GFP-Rab5-marked vesicles were well labeled by EEA1 (Figure 5–figure supplement 1), and thus we still used the GFP-Rab5 signal to detect the size of early endosomes. Quantification of the endosome size followed the method used in literature (Zhu, et al, 2007). Specifically, three independent experiments were performed and the diameter of 100 largest GFP-Rab5-positive vesicles in 50∼60 cells was measured for each group. Statistical analyses were performed using PRISM (GraphPad Software Inc) with the two-tailed and unpaired Student’s t-test. The graph in Figure 5B shows the mean ± SEM.

We also concur with the reviewers that it would be better to deplete the endogenous Rabex-5 in our experiments. In the revision, we tried to use siRNA to deplete the endogenous Rabex-5. Unfortunately, the best siRNA among the three pairs we designed could only knock down the expression of the endogenous Rabex-5 by 55% (data not shown). Considering that the cotransfection efficiency with three (Rab5, Rabaptin-5, and Rabex-5) or four (+EGFR) plasmids is very low, this method appears to be very hard to work. In addition, for the cell morphological analyses, we have to mark the siRNA with a fluorescence label to monitor its transfection, which inevitably increases the difficulty to distinguish the siRNA signal from the fluorescence signals of the co-transfected proteins. Most of all, our results have shown that the cells co-transfected with Rabex-5 and Rabaptin-5 can enlarge the diameter of the early endosomes to 2.94 ± 0.09 μm which is significantly larger than that in the cells co-transfected with the vectors (0.97 ± 0.04 μm), suggesting that the influence of the endogenous Rabex-5 is limited and does not affect our analyses and conclusions. Thus, we remain to perform the assay in the presence of the endogenous Rabex-5.

For the functional analysis of Rab5 dependent endocytic transport, as reported in the literature, the activation of Rab5 does not affect the rate of EGF-stimulated EGFR internalization but may regulate the colocalization of internalized EGF/EGFR and Rab5 (Dinneen and Ceresa, 2004b). However, the activation of Rab5 facilitates the ligand-independent EGFR internalization (Dinneen and Ceresa, 2004a). In addition, although Rab5 can accelerate the transferrin internalization, there is no obvious difference between the wild-type and constitutively active mutant (Q79L) Rab5 (Stenmark et al, 1994). Therefore, to analyze whether the activation of Rab5 by the Rabex-5-Rabaptin-5 complex has any effect on the endocytic transport, we have performed both cell biological and biochemical experiments to detect the internalization of the unliganded EGFRs in HEK 293 cells (which do not express native EGFR) co-transfected with EGFR, Rab5, Rabex-5, and Rabaptin-5 (data not shown). Consistent with our cell biological results shown in Figure 5, the new results also showed that Rabex-5 and Rabaptin-5 can enlarge the size of early endosomes and disruption of the Rabex-5-Rabaptin-5 interaction decreases the size of early endosomes accordingly. We also observed that EGFR can colocalize well with Rab5 in each case, suggesting that the endocytic trafficking of EGFR is correlated to the function of Rab5. However, we did not observe notable difference of the EGFR internalization in the cells co-expressing the vectors and different forms of Rabex-5 and Rabaptin-5 in both the cell morphological analysis of the distribution of EGFRs and the biochemical analysis of the surface EGFRs by surface biotinylation assay (Chen et al, 2012). These results suggest that under our assay conditions, the activation of Rab5 by the Rabex-5-Rabaptin-5 complex has little effect on the internalization of EGFR but is essential for the proper function of Rab5 in the downstream processes, such as the fusion of early endosomes. Due to the indirect relevance of these results to the main topic of this paper, we decided not to include these results in the revised text.

Chen XQ, Wang B, Wu C, Pan J, Yuan B, Su YY et al (2012) Endosome-mediated retrograde axonal transport of P2X3 receptor signals in primary sensory neurons. Cell Research 22: 677-696.

Dinneen JL, Ceresa BP (2004a) Continual expression of Rab5(Q79L) causes a ligand-independent EGFR internalization and diminishes EGFR activity. Traffic 5: 606-615.

Dinneen JL, Ceresa BP (2004b) Expression of dominant negative rab5 in HeLa cells regulates endocytic trafficking distal from the plasma membrane. Experimental Cell Research 294: 509-522.

Stenmark H, Parton RG, Steele-Mortimer O, Lutcke A, Gruenberg J, Zerial M (1994) Inhibition of rab5 GTPase activity stimulates membrane fusion in endocytosis. The EMBO Journal 13: 1287-1296.

Zhu H, Zhu G, Liu J, Liang Z, Zhang XC, Li G (2007) Rabaptin-5-independent membrane targeting and Rab5 activation by Rabex-5 in the cell. Molecular Biology of the Cell 18: 4119-4128.

Other comments:

1) Figure 1 is very busy and confusing at first glance. It is not immediately obvious that the green helix in Figure 1A is part of Rabex-5 because the labeling overlaps the structure. The presentation and labeling should be improved.

As suggested by the reviewers, we have removed the labels of the secondary structures in Figure 1A and simply labeled the HB, Vps9, and CC domains of Rabex-5Δ with the same colors of the structure models. For clarity, we have also labeled the N- and C-terminals of Rabex-5Δ in black.

2) The text should make it clearer that nucleotide-free Rab5 was present in the crystal structure. This is only mentioned later nd described in a way that makes it appear surprising Rab5 is in nucleotide-free form. Taking the work of Delprato and Lambright (2007) and Langemeyer (2014) together one would have expected Rab5 and Rab21 to be activated by the same mechanism and to show the same interactions to Rabex-5 when in nucleotide-free form.

As suggested by the reviewers, we have clarified in the text that the Rab5 in the Rab5-Rabex-5Δ-Rabaptin-5C212 complex is a nucleotide-free form as follows: “No nucleotide and/or metal ion are found at the active site of Rab5 and thus the bound Rab5 is in nucleotide-free form.”

We have also cited the works of Delprato and Lambright (2007) and Langemeyer (2014) when discussing the similar interactions of Rab5 and Rab21 with Rabex-5 in the text as follows: “Nevertheless, the interactions between Rabex-5 and Rab5 are similar to those between Rabex-5 and Rab21, suggesting that Rabex-5 may activate Rab5 via a similar mechanism as for Rab21 (Figure 2D) (Delprato and Lambright, 2007; Langemeyer et al, 2014).”

3) Figure 2 should show only one complex and not both copies in the overall structure for clarity. The text indicates that one Rab5 is not well defined, so this could be shown in the supplement rather than the main figure.

As suggested by the reviewers, we have now shown only one Rab5-Rabex-5Δ-Rabaptin-5C212 complex in the overall structure in Figure 2A but both copies in Figure 2–figure supplement 1A.

4) The Discussion and model should better explain how the complex might interact with and sit at a membrane surface given the known interactions of Rabex-5 via its UIM and of rabaptin with GGAs and Rabs. 3. Also, the coiled coil of Rabaptin seems to exist as an either bent form or elongated form in solution and can form a 3-helix bundle with RABEX5 to open the enzyme. Not clear is whether this coiled coil is freely available to behave that way within the context of the full length Rabaptin-5 protein and this needs to be discussed and included in Figure 5. Also not clearly specified is why Lambright saw more dramatic inhibition and relief thereof, by the Rabaptin domain (please add this to the Discussion).

As suggested by the reviewers, we have re-generated the schematic model to show the activation mechanism of Rabex-5 by Rabaptin-5 in the context of the early endosomal membrane localization of Rabex-5 and Rabaptin-5 in Figure 6 and discussed the model in more details in the text as follows: “In the context of the early endosomal membrane, Rabex-5 can directly target to the early endosomes either through the interaction of the N-terminal ubiquitin binding domain (UBD) with the ubiquitinated cargoes or through the early endosomal targeting domain (EET) (Mattera and Bonifacino, 2008; Zhu et al, 2007). In this case, as the GEF activity of Rabex-5 is autoinhibited by Rabex-5CC, Rabex-5 can only activate Rab5 at the basal level. In the cells, most Rabex-5 forms a stable complex with Rabaptin-5 which can be recruited to early endosomes via the binding of Rabaptin-5 through the C-terminal region to GTP-bound Rab5 (Lippe et al, 2001; Zhu et al, 2010). In this case, Rabaptin-5 not only assists the recruitment of Rabex-5 to the early endosomal membrane, but also activates the GEF activity of Rabex-5. The positive feedback loop among Rab5, its effector Rabaptin-5, and its GEF Rabex-5 can lead to a robust activation of Rab5, which then promotes the fusion of early endosomes efficiently.”

As a scaffold protein, Rabaptin5 comprises several coiled-coil regions that mediate interactions with different proteins to exert different functions. Rabaptin-5C21 always forms a dimer, which can assume either a V-shaped conformation or a linear conformation and can bind other proteins via different regions. In addition, it prefers to further dimerize to form a dimer of dimers via the regions that are not involved in the interactions with other proteins. In the previous version of the manuscript, we did not discuss these issues due to space limitation. As suggested by the reviewers, in the revision, we have added a brief discussion on the oligomerization and conformational flexibility of Rabaptin-5 within the context of the full-length protein as follows: “In the GAT-Rabaptin-5C212 structure, Rabaptin-5C212 assumes a linear conformation, the N-terminal regions of Rabaptin-5C212 bind one GAT, and two symmetry-related Rabaptin-5C212 dimerize through the C-terminal regions (Zhu et al, 2004a) which are the binding site for Rabex-5CC. Interestingly, Rabaptin-5C212 assumes a linear conformation in the Rabex-5CC-Rabaptin-5C212 structure but a V-shaped conformation in the R2Δ structure, and in both structures, the C-terminal regions of Rabaptin-5C212 bind one Rabex-5CC and two symmetry-related Rabaptin-5C212 dimerize through the N-terminal regions which are the binding site for GAT. Moreover, in the R3Δ structure, Rabaptin-5C212 also assumes a linear conformation but two Rabaptin-5C212 dimerize through the middle regions (residues 590-600) between the GAT-binding site and the Rabex-5CC-binding site. As the GAT-binding site is unoccupied, Rabaptin5 may bind Rabex-5 and GGA1 simultaneously. These results demonstrate that Rabaptin-5C21 always forms a dimer which can assume either a V-shaped or a linear conformation and can bind other proteins via different regions. In addition, the dimeric Rabaptin-5C212 prefers to further dimerize to form a dimer of dimers via the regions that are not involved in the interactions with other proteins. In the context of full-length Rabaptin-5, it is possible that Rabaptin-5C212 might exist in either the V-shaped or the linear conformation depending on the functional state. The dimerization of Rabaptin-5C212 might not only avoid exposure of the hydrophobic surface and minimize the overall energy of the protein in aqueous environment, but also play some functional roles in the tethering and fusion of early endosomes and/or with other vesicles.”

We think that the discrepancy between Lambright et al. and ours on the magnitude of the autoinhibition and the relief by the Rabaptin-5C21 binding might be caused by differences of the assay systems, for example, different Rab5 and Rabex-5 constructs, different concentration of the proteins, and different sensitivity of the instruments. In the revision, we have added these conjectures as follows: “However, the magnitude of the autoinhibition by Rabex-5CC is smaller than that reported by Delprato et al (Delprato and Lambright, 2007) which may be caused by differences of the assay systems, for examples, different Rab5 and Rabex-5 constructs, different concentration of the proteins, and different sensitivity of the instruments.”

5) The Title is not a correct description of the work. The authors uncover details of the way in which rabaptin binding relieves auto inhibition of Rabex-5. It does not provide the mechanism of Rab5 activation in detail or look at endocytosis.

We have changed the title to “Molecular mechanism of the activation of the Rabex-5 GEF activity by Rabaptin-5 in endocytosis”.

6) The statement that the “the substrate-binding site is largely exposed to the solvent and accessible by the substrate” appears to be contradicted by the previous sentence, which states that ”the binding sites for switch II and a small portion of the interswitch region of Rab21 are occupied by the three-helix bundle” and by Figure 1A, 1B and Figure 1–figure supplement 4 in which the substrate binding site appears to be blocked by the three helix bundle.

We thank the reviewers for the kindly reminder. In the Rabex-5Δ-Rabaptin-5C212 complex, the substrate-binding site of Rabex-5 is partially occupied by the three-helix bundle. The interaction interface buries a total of solvent accessible surface area of 1040 Å2 which is much smaller than that between Rabex-5 and Rab21 (2400 Å2) (Delprato & Lambright, 2007). Rab21 uses switch I, switch II, and the interswitch region to interact with the substrate-binding site of Rabex-5. Although the binding sites for switch II and a small portion of the interswitch region of Rab21 are occupied by the three-helix bundle, the binding sites for switch I and a large portion of the interswitch region of Rab21 are exposed (Figure 1–figure supplement 3C). Hence, we consider that the substrate-binding site of Rabex-5 is largely exposed to the solvent and partially accessible by the substrate.

In the previous Figure 1–figure supplement 3B, we colored switch I, switch II, and the interswitch region of Rab21 in orange, yellow, and green, respectively, and the other regions in magenta and labeled Rab21 in magenta as well. That might lead the reviewers to consider the magenta colored region as the whole of Rab21. To avoid confusion, we have regenerated it as Figure 1–figure supplement 3C in a slightly different orientation and with different colorings. In this new figure, switch II and the interswitch region of Rab21 remain to be colored in dark yellow and green, respectively, but switch I is colored in red and the other regions in gray. Meanwhile, we color Rabex-5CC and Rabaptin-5C21 in violet and pink, respectively. In this new figure, it is clearer that switch I and a large portion of the interswitch region are not occupied by the three-helix bundle.

7) Abstract: “Rabaptin-5C212” should be defined on first usage. In particular, clarify what the “2” after “Rabaptin-5C21” signifies.

We thank the reviewers for this suggestion. In the Abstract, we have specifically spelled out the dimeric Rabaptin-5C21 as follows: “We report here the crystal structures of Rabex-5 in complex with the dimeric Rabaptin-5C21 (Rabaptin-5C212) and Rabex-5 in complex with Rabaptin-5C212 and Rab5.”

8) Several abbreviations (e.g., R2, R2delta, R3, etc) are missing from the list. The authors should decrease jargon wherever possible as there are many acronyms here. Thus, please define R2 complex? R3 complex? Constructs used in 3G? It is hard for the reader.

We introduced the abbreviations of R2, R2Δ, R3, and R3Δto avoid lengthy description of the complexes. As suggested by the reviewers, we have added the abbreviations for R2, R2Δ, R3, and R3Δ in the abbreviation list and defined these abbreviations when they are first used in the text. We have also avoided to using these abbreviations if ambiguity might exist.

The constructs used in Figure 5 (original Figure 3G) have been specified in the Materials and methods section (Immunocytochemistry) as follows: “GFP-tagged Rab5 (residues 1-215), Myc-tagged Rabaptin-5 (residues 552-862), and Flag-tagged Rabex-5 (residues 132-455), Flag-tagged Rabex-5Δ (residues 132-455Δ393-407), and Flag-tagged Rabex-5 GEF (residues 132-392) were cloned into the pcDNA3 vector (Invitrogen), respectively.”

9) Finally, because the assignment and orientation of side chains shown in Figure 2C and 2D are determined primarily by the molecular replacement models and only a little bit by the data, please add a qualifying note about the uncertainty of the side chain orientations due to the low resolution of the structure.

As suggested by the reviewers, we have noted the uncertainty of the side-chain orientations in the Materials and methods section (Crystallization, data collection, and structure determination) as follows: “Due to the low resolution of the diffraction data, the structure models of Rab5, Rabex-5, and Rabaptin-5C21 in the R3Δ complex were refined as rigid bodies with deformable elastic network and group B-factor restraints (Schroder et al, 2010) and thus the side-chain orientations in this complex are somewhat uncertain.”

https://doi.org/10.7554/eLife.02687.025

Article and author information

Author details

  1. Zhe Zhang

    State Key Laboratory of Molecular Biology, Institute of Biochemistry and Cell Biology, Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences, Shanghai, China
    Contribution
    ZZ, Carried out the cloning, protein purification, crystallization, structure determination and analyses, biophysical and biochemical studies, and drafted and revised the manuscript
    Competing interests
    The authors declare that no competing interests exist.
  2. Tianlong Zhang

    State Key Laboratory of Molecular Biology, Institute of Biochemistry and Cell Biology, Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences, Shanghai, China
    Contribution
    TZ, Participated in the structure determination and analyses
    Competing interests
    The authors declare that no competing interests exist.
  3. Shanshan Wang

    State Key Laboratory of Molecular Biology, Institute of Biochemistry and Cell Biology, Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences, Shanghai, China
    Contribution
    SW, Participated in the cloning and protein purification
    Competing interests
    The authors declare that no competing interests exist.
  4. Zhou Gong

    State Key Laboratory of Magnetic Resonance and Atomic and Molecular Physics, Wuhan Institute of Physics and Mathematics, Chinese Academy of Sciences, Wuhan, China
    Contribution
    ZG, Participated in the design, analysis and discussion of the SAXS study
    Competing interests
    The authors declare that no competing interests exist.
  5. Chun Tang

    State Key Laboratory of Magnetic Resonance and Atomic and Molecular Physics, Wuhan Institute of Physics and Mathematics, Chinese Academy of Sciences, Wuhan, China
    Contribution
    CT, Participated in the design, analysis and discussion of the SAXS study, and revised the manuscript
    Competing interests
    The authors declare that no competing interests exist.
  6. Jiangye Chen

    State Key Laboratory of Molecular Biology, Institute of Biochemistry and Cell Biology, Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences, Shanghai, China
    Contribution
    JC, Participated in the design, analysis and discussion of the biochemical studies
    Competing interests
    The authors declare that no competing interests exist.
  7. Jianping Ding

    State Key Laboratory of Molecular Biology, Institute of Biochemistry and Cell Biology, Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences, Shanghai, China
    Contribution
    JD, Conceived the study, participated in the designs and data analyses of all experiments, and wrote and revised the manuscript
    For correspondence
    jpding@sibcb.ac.cn
    Competing interests
    The authors declare that no competing interests exist.

Funding

National Natural Science Foundation of China (31230017)

  • Zhe Zhang
  • Tianlong Zhang
  • Shanshan Wang
  • Jianping Ding

Ministry of Science and Technology of the People’s Republic of China (2011CB966301)

  • Zhe Zhang
  • Tianlong Zhang
  • Shanshan Wang
  • Jianping Ding

Ministry of Science and Technology of the People’s Republic of China (2011CB911102)

  • Zhe Zhang
  • Tianlong Zhang
  • Shanshan Wang
  • Jianping Ding

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

We thank the staff members at beamline 17U of Shanghai Synchrotron Radiation Facility (SSRF), China for technical support in diffraction data collection. We are grateful to Lan Bao, Xu-qiao Chen, Haixiang Shi, Fang Yu, and Jing Feng for their valuable contributions to the research work. This work was supported by grants from the National Natural Science Foundation of China (31230017) and the Ministry of Science and Technology of China (2011CB966301 and 2011CB911102).

Reviewing Editor

  1. Suzanne R Pfeffer, Reviewing Editor, Stanford University, United States

Publication history

  1. Received: March 1, 2014
  2. Accepted: June 20, 2014
  3. Accepted Manuscript published: June 23, 2014 (version 1)
  4. Version of Record published: July 22, 2014 (version 2)

Copyright

© 2014, Zhang et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 2,143
    Page views
  • 237
    Downloads
  • 9
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Comments

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Download citations (links to download the citations from this article in formats compatible with various reference manager tools)

Open citations (links to open the citations from this article in various online reference manager services)

Further reading

    1. Neuroscience
    Simon Nimpf et al.
    Research Article Updated