Structural and Biophysical Characterization of Rab5a from Leishmania Donovani

Abstract

Leishmania donovani possess two isoforms of Rab5 (Rab5a and Rab5b), which are involved in fluid phase and receptor-mediated endocytosis, respectively. We have characterized the solution structure and dynamics of a stabilized truncated LdRab5a mutant. For the purpose of NMR structure determination, protein stability was enhanced by systematically introducing various deletions and mutations. Deletion of hypervariable C-terminal and the 20 residues LdRab5a specific insert slightly enhanced the stability, which was further improved by C107S mutation. The final construct, truncated LdRab5a with C107S mutation, was found to be stable for longer durations at higher concentration, with an increase in melting temperature by 10°C. Solution structure of truncated LdRab5a shows the characteristic GTPase fold having nucleotide and effector binding sites. Orientation of switch I and switch II regions match well with that of guanosine 5'-(β, γ-imido)triphosphate (GppNHp)-bound human Rab5a, indicating that the truncated LdRab5a attains the canonical GTP bound state. However, the backbone dynamics of the P-loop, switch I, and switch II regions were slower than that observed for guanosine 5'-(β, γ-imido)triphosphate (GMPPNP)-bound H-Ras. This dynamic profile may further complement the residue-specific complementarity in determining the specificity of interaction with the effectors. In parallel, biophysical investigations revealed the urea induced unfolding of truncated LdRab5a to be a four-state process that involved two intermediates, I1 and I2. The maximal 4,4'-dianilino-1,1'-binaphthyl-5,5'-disulfonic acid (Bis-ANS) binding was observed for I2 state, which was inferred to have molten globule like characteristics. Overall, the strategy presented would have significant impact for studying other Rab and small GTPase proteins by NMR spectroscopy.

Introduction

Endocytosis is a fundamental process by which cell acquires essential nutrients and other molecules from the extracellular milieu (1). The cargos carrying nutrients and molecules get fused with early endosomes, once they are recognized by cell surface receptors, where they are sorted, and targeted to their appropriate intracellular location (2). This transportation of cargos follows a series of coordinated and specific vesicle fusion events which are tightly regulated by small GTP-binding members of Rab family (3). To perform their functions, Rab proteins work as binary molecular switches, and cycle between two alternate conformations depending upon the GTP or GDP binding, in a manner that is similar to Ras proteins (4). In the GTP bound active conformation (5), Rab proteins get attached to specific target membrane and interact with downstream effectors, activate them, and execute signaling responses such as intracellular vesicle trafficking, cell proliferation, etc (3, 6). Upon stimulation by GTPase activating protein (GAP), which usually increases the intrinsic hydrolytic activity of Rab proteins (7), the GTP is hydrolyzed, leading the Rab proteins into an inactive conformation (3). In the inactive and GDP-bound conformation, Rab proteins return to cytosol, primarily in combination with the GDP dissociation inhibitor (GDI), and are unable to interact with downstream effector proteins (8). In order to reactivate the Rab proteins in response to extracellular signal, guanine nucleotide exchange factors (GEF) facilitate the release of GDP from Rab proteins (7). The high concentration of GTP in the cytosol ensures that GTP binds with Rab as soon as GDP dissociates (9).

Rab proteins are evolutionary conserved with 55–75% identity between members from yeast to humans (10). Along with other organisms, endocytic Rabs have been identified in trypanosomatids also (11) (e.g., in Trypanosoma brucei, Rab4, Rab5, Rab7 and Rab11), and in Dictyostelium discoideum, Rab7 and Rab4 have been reported so far (12, 13). Similarly, in Leishmania, many endocytic Rab proteins have been identified, including Rab5 isoforms, Rab7, Rab6, etc. Three isoforms of Rab5 are present in higher eukaryotes, Rab5a, Rab5b, and Rab5c, which share more than 90% similarity and regulate the dynamics of early endosome fusion and rate of endocytosis in eukaryotic cells (14). Rab5 isoforms might have different roles in different cell types. Unlike mammalian cells, Leishmania possess only two isoforms of Rab5, Rab5a and Rab5b, which share ∼60% similarity, and both of them are located on the early endosomal compartment (15). Null mutants of Rab5 isoforms in L. donovani were shown to be lethal. Hence, these proteins are essential for the parasites (15). Trypanosomatids have developed receptor system that mediate uptake of low-density lipoproteins and transferrin, which are involved in efficient supply of nutrients required for rapid growth, such as cholesterol and metal ions (16). It is reported that Rab5a in L. donovani specifically stimulates HRP uptake via fluid phase endocytosis as well as enhances the kinetics of lysosomal transport. In contrast to Rab5a, Rab5b in L. donovani induces hemoglobin uptake by receptor-mediated endocytosis as well as enhances the targeting of internalized hemoglobin to the lysosomes (15, 17).

Rab proteins, like other Ras superfamily proteins, consist of a GTPase domain and a hypervariable C-terminal, which is the site for prenylation and membrane anchoring (8, 10). A canonical GTPase domain is ∼20–25 kDa in size, and it consists of a central six stranded mixed β-sheet surrounded by five α-helices (18). The highly conserved elements of the G-domain are five polypeptide loops that connect the secondary structure elements. These loops are named as G1–G5 and form guanine nucleotide binding site, active site for the GTP hydrolysis, as well as binding site for various effector and regulatory molecules. The G1 loop/P-loop/di-phosphate binding loop, which has a consensus sequence GXXXXGKS, interacts with α and β phosphate of the nucleotide. G2 loop/switch I contains a conserved Thr residue and G3 loop/switch II has the consensus sequence DXXG (18). Conserved Thr residue of switch I and Gly residue of switch II form hydrogen bond with the γ phosphate of GTP and they also interact with Mg⁺². Additionally, the glutamine residue present after DXXG sequence is also conserved in all Rab proteins, and is well known for playing a crucial role for GTP hydrolysis and GDP to GTP exchange. In Rab/Ras proteins, interaction of this glutamine residue with conserved Lys residue of G1/P-loop is promoted by GAP proteins for the release of phosphate. After release of phosphate, Gln side chain gets oriented in a different direction, which promotes the interaction of Lys residue of P-loop with Asp residue of GEF proteins for GDP to GTP exchange (19). Beside these loops, G4 and G5 loops, which have conserved sequences NKXD and (T/G)(C/S)A, respectively, play a role in guanine base recognition and binding. Although the catalytic machinery is highly conserved among most Rab proteins, the intrinsic GTP hydrolysis rates in the Rab family vary by more than an order of magnitude (20).

Upon hydrolysis of γ phosphate from GTP, switch I and switch II become relatively disordered and the protein becomes inactive. This is because in monomeric GTPases, switch I and switch II regions are also critical for effector binding (21). Crystal structures of Rab5 in complex with EEA1 C2H2 zinc-finger domain and with Rabaptin5, and structures of other Rab-effector complexes, have shown that the switch I and II regions, the interswitch region (β2–β3), and the α3–β5 loop contribute to the specificity of interactions with various effectors. Further, these regions are perceived to be typically mobile (22, 23, 24). The dynamics of these regions are also indicated by NMR solution studies on closely related protein H-Ras in complex with GMPPNP (25). For the latter, it was shown that binding of c-Raf Ras binding domain decreased the flexibility of the P-loop and switch regions (26). The dynamic regions of Rab5 have also been recently explored through molecular dynamics (MD) simulations (27). Although the study was primarily undertaken to explain the differences in the GTPase activities of mutants of the P-loop residue A30 of Rab5a (A30R and A30P), control MD simulations of wt-Rab5a showed higher root mean square fluctuation (RMSF) values for the switch I, switch II, and the interswitch regions (27). Another very interesting MD simulations study has been reported for a prenylated Rab5a embedded into a lipid bilayer. This study showed that the GTP to GDP conversion was accompanied by a rotation around the protein axis with reference to the bilayer and a reorientation of the switch regions. This study again showed higher than average RMSF values for the switch I and II regions in both the GDP- and GTP-bound states, with the fluctuations being more pronounced in the GDP-bound state (28).

Structural biology of Rabs and their related regulators and effectors is largely unexplored for Leishmania. Although a number of crystal structures of Rab proteins from different organisms have been reported (29, 30, 31, 32, 33, 34, 35), no structure for any of the Rab proteins from Leishmania has been determined so far. Although conventional Rab5 effectors like EEA1, Rabex5, Rabaptin5, and Rabenosyn5 have been well characterized in mammalian cells, their homologs have not been identified in Leishmania (15). Therefore, Leishmania is an interesting system for the discovery of Rab effectors and their roles in membrane trafficking and sorting. Several Leishmania Rabs, like Rab1, Rab5a, Rab5b, and Rab7, have been functionally characterized (15, 36, 37). LdRab5a is a GTPase that has been shown to bind to GTP and hydrolyze it to GDP. It specifically regulates fluid-phase endocytosis. Although the G1–G5 loop sequences are well conserved in LdRab5, it has unique interswitch region with substitutions and a 20-residue insertion with respect to the human Rab5 (15). The unique insertion feature of LdRab5a prompted us to choose NMR spectroscopy as the most suitable technique for determining its structure. It is pertinent to mention that NMR solution structure has not been determined for any Rab protein from any organism. This would also provide the opportunity for exploring the residue-specific dynamics for LdRab5a in particular and reveal the general implications for the Rab family of proteins. Hence, we have undertaken the determination of solution structure and characterization of backbone relaxation dynamics for LdRab5a by using NMR spectroscopy.

However, stabilization of LdRab5a through systematic and specific mutations and deletions was necessary to arrive at a construct that was stable enough for recording long multidimensional NMR experiments. The NMR solution structure could finally be determined for LdRab5a mutant, labeled as Rab5aMΔCΔ60-79C107S, having Q93L, P58D, P59G, and C107S mutations, and Δ60–79 and Δ196–235 deletions. In addition to NMR solution structure and backbone relaxation dynamics, the stability of Rab5aMΔCΔ60-79C107S was characterized through thermal unfolding and equilibrium unfolding studies. Such studies, which have been reported for p21-HRas, provide insight into the role of the nucleotide binding domains in the folding and stability of the proteins (38). Therefore, we have also characterized the unfolding pathway of Rab5aMΔCΔ60-79C107S. The results show that the truncated LdRab5a mutant retains the characteristic GTPase fold with all the conserved catalytic sites. It exhibits higher thermal stability and it undergoes unfolding via a four-state unfolding pathway.

Materials and Methods

Sub-cloning of LdRab5a and its deletion mutants

A number of variations have been done in Rab5a gene by polymerase chain reaction (PCR) to attain the stability using specific primers. The gene corresponding to the LdRab5aM (M = Q93L mutation) (15) was subcloned in pQE30 vector at BamHI and HindIII restriction sites by PCR, restriction digestion and ligation. The native LdRab5a protein consists of 235 residues. After comparison with the canonical folding of small GTPases, 40 residues from the C-terminal (Rab5aMΔC) were deleted by PCR. We also tried to express LdRab5a by deleting only 37 residues from C-terminal (Rab5aMΔC₃₇), but it could not be expressed. Furthermore, 20 nonconserved residues from P60 to M79 were deleted by PCR (Rab5aMΔCΔ60-79). Upon sequence alignment, we found that at 58 and 59 positions, there are two proline residues in LdRab5a, which were also replaced by Asp58 and Gly59 by site-directed mutagenesis. Thus, in P60–M79 deletion construct, P58D and P59G mutations are also present.

The cloning procedure added an extra 12 residues at the N-terminus, including a noncleavable 6xHis-tag in all the recombinant constructs in the pQE30 vector. A point mutation, C107S, was generated to further stabilize the protein by using specific primers. However, this clone did not express properly and could not be purified with His-tag affinity chromatography. Therefore, the gene corresponding to Rab5aMΔCΔ60-79 was cloned in glutathione S-transferase (GST)-TAG vector (pGEX-4T1). The cloning procedure added a GST tag and a cleavable thrombin site at N-terminal. Only two residues (Glycine-Serine) were left at the N-terminal after thrombin digestion of the GST tag. C107S mutation was introduced in this construct and the resultant protein could be expressed well and purified (Rab5aMΔCΔ60-79C107S). All of the constructs were confirmed by double restriction digestion and DNA sequencing. The final construct is as follows: Rab5a-M(Q93L); ΔC(L196-C235); Δ60–79(P58D; P59G; ΔP60-M79); C107S.

Overexpression and purification of LdRab5a mutants

The recombinant proteins (Rab5aMΔC and Rab5aMΔCΔ60-79) were expressed in Escherichia coli TG1 strain and purified as soluble proteins from the cytoplasmic fraction. The protein was purified with Ni–NTA superflow matrix (Qiagen, Hilden, Germany) packed in a column by gravity method. The column was pre-equilibrated with lysis buffer (50 mM NaH₂PO₄, pH 8.0, 300 mM NaCl, 10 mM Imidazole, 5 mM MgCl₂, 1 mM GTP). The lysate was applied to the column, the flow through was collected, and the column was washed with lysis buffer and subsequently with increasing concentrations of imidazole from 20 to 50 mM. The protein was eluted by elution buffer (50 mM NaH₂PO₄, pH 7.5, 300 mM NaCl, 250 mM imidazole, 5 mM MgCl₂, and 1 mM GTP). 2 mM DTT (Dithiotheritol) and a protease inhibitor cocktail were added to protein after elution.

The gene for Rab5aMΔCΔ60-79C107S was cloned in pGEX-4T1 vector, and the recombinant protein was expressed in E. coli BL-21 (λDE3) strain. Glutathione agarose 4B (MACHEREY-NAGEL, Düren, Germany) beads were used for purification. Beads were equilibrated with buffer A (50 mM NaH₂PO₄, pH 8.0, 100 mM NaCl, 5 mM MgCl₂, 1 mM GTP) and after binding of lysate, the beads were washed with buffer A by varying NaCl concentration (200–300 mM) to remove nonspecific binding. The GST tag from fusion protein was digested with thrombin protease (Calbiochem, San Diego, CA) at 25°C for 12 h. After nickel-nitrilotriacetic acid (Ni-NTA) and GST-based affinity chromatography, the proteins were further purified by size-exclusion chromatography over Superdex-75 10/300GL column (GE Healthcare, Chicago, IL) using a fast performance liquid chromatography system (Biorad Biologic Duo Flow; Biorad, Hercules, CA). Untagged proteins thus obtained were analyzed by SDS-PAGE for purity assessments.

For isotopic labeling, overexpression of different constructs were standardized in M9 minimal media containing ¹⁵N ammonium sulfate and ¹³C glucose as the sole nitrogen and carbon sources, respectively. The NMR sample was prepared in an NMR buffer (20 mM NaH₂PO₄, pH 6.5, 50 mM NaCl, 5 mM MgCl₂, 1 mM GTP, 2 mM DTT, 3% glycerol, 0.1% NaN₃) containing 93% H₂O and 7% D₂O.

NMR assignment and structure calculation

In order to assign the backbone and side chain chemical shifts, various two-dimensional spectra (¹⁵N-¹H-HSQC, ¹³C-¹H-HSQC (aliphatic), ¹³C-¹H-HSQC (aromatic), 2D-(HB)CB(CGCE)HE, and 2D-(HB)CB(CGCD)HD) and three-dimensional (3D) spectra (HNCACB, CBCA(CO)NH, HNCO, HN(CA)CO, H(CCO)NH-TOCSY, (H)C(CO)NH-TOCSY, and HCCH-TOCSY) were recorded. Distance restraints were obtained from 3D ¹⁵N-edited NOESY-HSQC (τ_mix 100 ms), ¹³C (aliphatic)-edited NOESY-HSQC (τ_mix 150 ms), and ¹³C (aromatic)-edited NOESY-HSQC (τ_mix 150 ms) experiments. All spectra were collected at 298 K on either Varian Inova 600 MHz or Bruker (AVANCE III) 800 MHz spectrometer equipped with actively shielded Z-gradient triple resonance cold probe. Spectra were processed by using the software NMRPIPE (39) and Topspin 3.0, and all NMR data were analyzed by CARA-1.8.4 (40). The NMR data were referenced for ¹H chemical shifts by using 2, 2-dimethyl-2-silapentane-5-sulphonic acid at 298 K as a standard. The ¹³C and ¹⁵N chemical shifts were referenced indirectly (41). All NOEs were assigned manually and integrated nuclear Overhauser effect (NOE) peaks were calibrated and converted into distance restraints with the program CALIBA (42). In the final structural determination, the program CYANA-3.0 (43) was used. The torsion angle restraints were obtained from the assigned backbone chemical shifts by using the program TALOS+ (44). In total, 3406 NOE distance restraints were identified, to which 33 hydrogen bond restrains were added for determination of LdRab5a structure. In total, 200 randomized conformers were generated and 10 conformers with lowest target function with no distance violation >0.5 Å (1 Å = 0.1 nm) and no angle violations were selected for each. These 10 structures with the lowest target functions were further subjected to MD simulation in explicit water with NMR-derived distance restraints and angle restraints by using the Crystallography and NMR system 1.21 program and the standard water shell refinement protocol (45, 46, 47). This step improved the Ramachandran plot statistics and also the Z-score for the Procheck (ϕ-ψ) and Procheck (all) for the ordered residues. The program PSVS v1.4 (http://www.psvs-1_4.nesg.org) was used to analyze the quality of the structures.

Relaxation measurements

To study the backbone dynamics of Rab5aMΔCΔ60-79C107S, experiments for the measurement of ¹⁵N longitudinal and transverse relaxation rate constants and {¹H}-¹⁵N steady-state NOE were recorded (48) and analyzed by model-free formalism, as described previously (see Supporting Materials and Methods for details).

Protein unfolding studies

The thermal and urea denaturation studies for Rab5a mutants were carried out by differential scanning calorimetry (DSC), fluorescence and far-UV circular dichroism spectroscopy, and size exclusion chromatography (see Supporting Materials and Methods for details).

MD simulations

MD simulations were carried out for Rab5aMΔC model using the GROMACS code (version 5.1.2) and Gromos43a1 force field for 100 ns. Model of Rab5aMΔC (with the insert) was generated by using Swiss Model homology modeling server (https://swissmodel.expasy.org/) by giving solution structure of Rab5aMΔCΔ60-7C107S as a template. The model was energy minimized before the MD simulations and no constraints were imposed. GROMACS utilities for system preparation were used thoroughly. The system was then solvated by using cubic simple point-charge water box followed by the charge neutralization of the system. Subsequently the entire system was minimized by using the steepest descent algorithm. The system was then equilibrated for 100 ps at 300 K using the constant number, volume, and temperature ensemble and for 100 ps using NPT ensemble. Production MD simulation was run for 100 ns. After the simulation root-mean-square-deviation (RMSD) and residue wise RMSF profiles were computed and analyzed.

Results

Sequence comparison of LdRab5a with other Rab proteins from different organisms

A multiple sequence alignment of LdRab5a, LdRab5b, and five other Rab5 isoforms is shown in Fig. S1 (see Supporting Materials and Methods for details). LdRab5a sequence displays the five Rab family motifs (RabF1–F5) and four Rab subfamily motifs (RabSF1–SF4). Residues which are specific for GTP hydrolysis are also found to be conserved in LdRab5a along with other Rab proteins. Besides this, 40 residues at the C terminus align with the hypervariable C-terminals of other Rab proteins, whereas an extra region of 20 residues (P60–M79) is specifically present in LdRab5a, the function of which is not characterized yet.

C107S mutation enhances the stability of protein (Rab5aMΔCΔ60-79C107S)

Cys107 is not conserved in all the Rab members. In LdRab5a, this is the only Cys that has been left in the protein after C-terminal deletion. The presence of this Cys could be the cause of protein precipitation because it may be involved in disulphide bond formation upon protein unfolding. C107S mutation was done in Rab5aMΔCΔ60-79 using site-directed mutagenesis (Rab5aMΔCΔ60-79C107S). This mutation provides enhanced stability to the protein for up to 6–7 days at 0.8 mM concentration, which made it possible to record the long NMR spectra to achieve the aim of the study (i.e., characterization of solution structure and dynamics). Preliminary NMR characterization of LdRab5aMΔC and LdRab5aMΔCΔ60-79 is shown in Fig. S2, A–C (see also Supporting Materials and Methods).

Further, we have characterized the conformational stability of the LdRab5a deletion constructs (Rab5aMΔCΔ60-79 and Rab5aMΔCΔ60-79C107S) by using intrinsic Trp fluorescence and CD spectroscopy. Equilibrium unfolding studies of a protein using chaotropic agents can provide the measure of its conformational stability (49, 50). LdRab5a has two Trp residues according to its primary amino acid sequence at positions 67 and 107 (after deletion of P60–M79). Emission wavelength for both of the constructs of LdRab5a was observed at 337 nm for Trp fluorescence. In a folded protein, buried Trp show emission maxima at 330–335 nm, whereas upon unfolding the exposed Trp, emission maxima shifts to ∼350 nm (51). Thus, Trp residues in both of the LdRab5a constructs are partially exposed to the solvent. Far-ultraviolet (UV) CD studies on urea-induced unfolding of LdRab5a constructs (Rab5aMΔCΔ60-79 and Rab5aMΔCΔ60-79C107S) were also carried out to study the effect of urea on the secondary structure. The effect of increasing concentration of urea on the ellipticity at 216 nm for the native constructs is shown in Fig. 1 B. No significant change was observed up to 2 M urea for Rab5aMΔCΔ60-79. However, between 2 and 6 M urea, we found a complete loss of the secondary structure for this mutant. On the other hand, Rab5aMΔCΔ60-79C107S was able to maintain its secondary structure up to 3.5 M urea without any substantial change in the ellipticity at 216 nm, but from 3.5 M urea to 6 M urea, almost complete loss of signal for secondary structure was observed (Fig. 1 B). The results obtained from the intrinsic Trp fluorescence were also found comparable to the results from CD spectroscopy (Fig. 1 A).

Figure 1.

Effect of C107S mutation on the chemical and thermal stabilities, as studied by fluorescence, CD spectroscopy, and DSC. (A) The plot of the fractional change in the wavelength maxima of Trp fluorescence of Rab5aMΔCΔ60-79 (black circles) and Rab5aMΔCΔ60-79C107S (grey circles) with increasing concentration of urea. (B) The plot of fractional change in the ellipticity at 216 nm for Rab5aMΔCΔ60-79 (black squares) and Rab5aMΔCΔ60-79C107S (grey squares) by CD spectroscopy with the increasing concentration of urea. (C) DSC profile of Rab5aMΔCΔ60-79 having Tm 52.14°C. (D) DSC profile of Rab5aMΔCΔ60-79C107S having Tm 62.44°C.

Changes in the CD ellipticity at 216 nm (θ₂₁₆), and tryptophan fluorescence emission λ_max, of Rab5aMΔCΔ60-79 and Rab5aMΔCΔ60-79C107S with increasing urea concentrations showed a cooperative process (Fig. 1, A and B). Data of both fluorescence and CD were converted into the fraction of denaturation (see Supporting Materials and Methods for details). The denaturation curves from both optical methods show a sigmoidal dependence and a significant difference was observed in the stability between the two constructs of LdRab5a (Fig. 1, A and B). Rab5aMΔCΔ60-79 shows the C_m value (midpoint transition) at ∼3.5 M urea, whereas Rab5aMΔCΔ60-79C107S shows C_m value at ∼4.5 M urea (Fig. 1, A and B). The higher C_m for Rab5aMΔCΔ60-79C107S protein suggests that it is significantly more stable than the Rab5aMΔCΔ60-79 protein.

Thermal stability of Rab5a mutants Rab5aMΔCΔ60-79 and Rab5aMΔCΔ60-79C107S

DSC studies were carried out to assess the thermal stabilities of the LdRab5a mutants (Rab5aMΔCΔ60-79 and Rab5aMΔCΔ60-79C107S) and to observe the effect of C107S mutation on the thermal stability of the protein. The protein samples showed irreversible transition and precipitated after the DSC run, so the profiles were mainly analyzed in terms of the peak temperature, melting temperature (Tm). A DSC thermogram displaying thermal unfolding of the mutants at a concentration of 0.02 mM in phosphate buffer is shown by the solid line curve in Fig. 1, C–D. Thermogram indicates that the Rab5aMΔCΔ60-79 unfolding peak is characterized by a Tm of ∼52°C, whereas the C107S mutant was found to be more stable with a Tm of ∼62°C. It is clearly seen from the DSC curve that the C107S mutation of Rab5aMΔCΔ60-79 has a strong influence on the thermal stability of LdRab5a.

NMR assignments and data deposition

The ¹⁵N-¹H-HSQC spectrum of ¹⁵N labeled-Rab5aMΔCΔ60-79C107S showed well-dispersed peaks with a less crowded central region, which implies that the protein was well folded and was suitable for its NMR characterization studies. Backbone resonance assignments were made for 165 of the 171 possible nonproline residue HN/N crosspeaks in the ¹⁵N-¹H HSQC spectrum. The ¹⁵N-¹H HSQC spectrum of ¹⁵N labeled-Rab5aMΔCΔ60-79C107S, with assignments of residues name and number, is shown in Fig. S3. Main chain amide proton assignments were not made for the residues G47, L72, E73, R74, S77, and A79. In Rab5aMΔCΔ60-79C107S, 97.27% assignment for Cα, Cβ, and C’, and 92.44% for nonaromatic and noncarbonyl side-chain carbons, 74.5% for aromatic sidechain carbon, and 95% of Hα was completed. The chemical shifts of Rab5aMΔCΔ60-79C107S have been deposited in the BioMagResBank (http://www.bmrb.wisc.edu) under the accession number 27192. Coordinates of Rab5aMΔCΔ60-79C107S have been deposited in the Protein Data Bank (PDB) with accession number 5YOZ.

Solution structure of truncated Rab5a (Rab5aMΔCΔ60-79C107S)

The ribbon representation of lowest energy structure and the wire representations of the final ensemble of 10 structures of Rab5aMΔCΔ60-79C107S are shown in Fig. 2, A and B, respectively. The pairwise RMSD for the secondary structure regions of the final ensemble of 10 lowest energy structures was 0.6 Å. The atomic coordinates for all 10 structures have been deposited in the PDB. Structural parameters for the solution structure of Rab5aMΔCΔ60-79C107S are summarized in Table S1.

Figure 2.

Solution structure of Rab5aMΔCΔ60-79C107S. (A) Superimposition of backbone traces from final ensemble of 10 structures with lowest target function with labeled β-strands and α-helices. (B) Ribbon representation of final ensemble of 10 structures.

The solution structure of Rab5aMΔCΔ60-79C107S possesses a typical GTPase fold. The core of the Rab5aMΔCΔ60-79C107S consists of a central six-stranded mixed β-sheet. The strand ordering is β2-β3-β1-β4-β5-β6, in which the four central strands are parallel, whereas the strands β2–β3 run antiparallel to each other. Helices α1–α5 are placed in such a way that they surround the β-sheet core. The β-strands are β1 (T12–L19), β2 (A48–I57), β3 (R60–T69), β4 (G88–D94), β5 (I121–N126), and β6 (F153–A156); the α-helices are α1 (K26–R35), α2 (S77–Y82), α3 (S98–A115), α4 (F138–E148), and α5 (V164–K175). Like other GTPase, in LdRab5a, G1 loop/P-loop corresponds to ₂₀GESGAGKS₂₇ and connects the β1 strand to α1 helix. G2 loop/switch I loop connects α1 helix to β2 strand, and it contains the conserved T45 residue. NH of threonine residue present in the switch I loop (T35 in Ras and T45 in LdRab5a) forms a hydrogen bond with oxygen of the γ phosphate and its hydroxyl group coordinates with Mg⁺², as described previously (4). Since we have determined the solution structure in the presence of GTP, in Rab5aMΔCΔ60-79C107S position of the switch I loop matches with the guanosine 5'-(β, γ-imido)triphosphate (GppNHp)-bound structure of human Rab5a (PDB: 1R2Q), although it has no match with the GDP bound structure of human Rab5a (PDB: 1TU4) (Fig. 3). Side chain orientation of the T45 residue also matches well for Rab5aMΔCΔ60-79C107S and GppNHp-bound human Rab5a. ₆₈DTAG₇₁ (₈₈DTAG₉₁ before truncation) sequence at the N-terminus of the α2 helix corresponds to the switch II/G3 loop. This loop was defined previously for the interaction with Mg⁺² as well as with γ-phosphate. Glycine residue of the switch II, like T45 of switch I, forms a hydrogen bond with the oxygen of the γ phosphate (52).

Figure 3.

Comparison of human Rab5a and Rab5aMΔCΔ60-79C107S in the nucleotide bound state. (A) HsRab5a in GDP-bound form showing open conformation of switch I and switch II loops. (B) HsRab5a in GppNHp-bound form showing closed conformation of switch I and switch II loops. (C) Overlay of HsRab5a in GppNHp-bound (light gray) and GDP-bound (dark gray) forms to differentiate the closed and open conformation. (D) Overlay of Rab5aMΔCΔ60-79C107S (black) with GppNHp-bound HsRab5a (light gray) to display the similarity in orientation of switch I and switch II loops.

Other residues present in switch I and switch II loops facilitate the interaction between these two loops via contributing a hydrophobic interface (4, 21), which is consistent with Rab5aMΔCΔ60-79C107S also because of the presence of hydrophobic residues at the corresponding positions. Also, NOEs present between the residues of switch I (I46 and G47) and residue of switch II (A70) further corroborate the fact that switch I and switch II are stabilized by hydrophobic interactions. Upon superposition of the structure of Rab5aMΔCΔ60-79C107S with GppNHp- and GDP-bound crystal structures of human Rab5a, orientation of switch I and switch II loops matches with the GppNHp-bound human Rab5a (Fig. 3). The G4 and G5 loops that correspond to ₁₂₆NKKD₁₂₉ (₁₄₆NKKD₁₄₉ before truncation) and ₁₅₆ASA₁₅₈ (₁₇₆ASA₁₇₈ before truncation) consensus sequences, which were defined previously for the recognition of guanine ring, are also well formed in the structure. Positions of the P loop/G1 loop, G4 loop, and G5 loop are similar in both GppNHp- and GDP-bound forms of human Rab5a and Ras, whereas the position of switch I/G2 loop and switch II/G3 loop is different (53). In Rab5aMΔCΔ60-79C107S, the position of all three regions P loop/G1 loop, G4 loop, and G5 loop matches well with both the GppNHp- and GDP-bound structure of human Rab5a (Fig. 3). In the GppNHp-bound structure, the α2 is longer in length in comparison to the GDP-bound structure. However, in the GTP-bound LdRab5a structure, α2 is relatively small because the residues of this region are not assigned (absent in the ¹⁵N-HSQC due to exchange broadening). Therefore, this region is not converged in the solution structure of Rab5aMΔCΔ60-79C107S and shows high RMSD.

A structural homology search using the DALI server showed that the deletion construct of LdRab5a (Rab5aMΔCΔ60-79C107S) shares 51% identity with HsRab5c (PDB: 1Z07) with 1.8 Å RMSD and 23.5 Z-score. Similarly, Rab5aMΔCΔ60-79C107S shows structural similarity with several different classes of Ras superfamily of small GTPase (Table S2). The top seven members that match with Rab5aMΔCΔ60-79C107S, having Z-scores from 23.5 to 22.1 and RMSD 1.8 to 2.0 Å, were Homo sapiens Rab5c (PDB: 1Z07), H. sapiens Rab5a (PDB: 1TU3), H. sapiens Rab22a (PDB: 1Z0J), H. sapiens Rab31 (PDB: 2FG5), H. sapiens Rab11a (PDB: 2D7C), Saccharomyces cerevisiae YPT51 (PDB: 1EK0), and Plasmodium falciparum Rab6 (PDB: 1D5C).

Backbone amide ¹⁵N relaxation parameters for Rab5aMΔCΔ60-79C107S

After the exclusion of missing/unassigned and overlapped resonances, ¹⁵N spin-lattice relaxation rate constant (R₁), spin-spin relaxation rate constant (R₂), and steady-state heteronuclear {¹H}- ¹⁵N NOEs were determined for 147 out of 171 nonproline residues of Rab5aMΔCΔ60-79C107S at 600 MHz by using program Curvefit. The average values of relaxation parameters for residues of Rab5aMΔCΔ60-79C107S are R₁= 1.29 s^-1, R₂ = 18.42 s^-1, {¹H}-¹⁵N NOE = 0.77. The residue specific R₁, R₂, and steady-state heteronuclear {¹H}-¹⁵N NOE values obtained for Rab5aMΔCΔ60-79C107S at 600 MHz are shown in Fig. 4. Analysis of the R₁, R₂, and {¹H}-¹⁵N NOE provides information on the dynamics on both the picosecond-to-nanosecond as well as the microsecond-to-millisecond time scales. According to R₁, R₂, and {¹H}-¹⁵N NOE data, the dynamic flexibility was observed mainly for the residues of N-terminal, switch I, and switch II.

Figure 4.

Sequence dependent variations of backbone ¹⁵N relaxation parameters of Rab5aMΔCΔ60-79C107S. (A)–(D) are the plots of R₁, R₂, R₂/R₁ ratios, and steady-state {¹H}-¹⁵N NOE values, respectively, versus amino acid sequence of Rab5aMΔCΔ60-79C107S. All experiments of Rab5aMΔCΔ60-79C107S were performed on 0.6 mM ¹⁵N-labeled sample at 25°C on a Varian Inova spectrometer at 600 MHz.

R₁, R₂, and steady state heteronuclear {¹H}-¹⁵N NOE were subjected to model-free analysis (54, 55, 56, 57). The fitting of the residues to different models of model-free was done with the isotropic diffusion tensor. Model-free analysis yielded a rotational correlation time (τ_m) of 10.91± 0.03 ns for the protein molecule, which confirms that the molecule exists as monomer in solution. The generalized order parameter (S²), the effective internal correlation time (τ_e), and a conformational exchange broadening parameter (R_ex) for each backbone amide NH vector that could be determined using the model-free formalism for Rab5aMΔCΔ60-79C107S are shown in Fig. S4. For Rab5aMΔCΔ60-79C107S out of 147 residues, only 139 residues could be fitted, whereas eight residues, T3, H4, T34, I95, L101, W107, K127, and S157 (numbering after truncation) could not be fitted to any of the five models. The ribbon representations of the Rab5aMΔCΔ60-79C107S structure, shaded according to residue-specific generalized order parameters (S²), effective correlation time (τ_e), and conformational exchange term (R_ex), are shown in Fig. S4, D and E.

Overall, residues in secondary structure elements display high ¹H-¹⁵N order parameters (S²), with an average of 0.88, as shown in Fig. S4 A. In total, 96 out of 141 residues were fitted to model 1, 4 residues were described by model 2, 32 residues were described by model 3, 2 residues were described by model 4, and 4 residues were described by model 5. In Rab5aMΔCΔ60-79C107S residues showing motion on nanosecond and picosecond (τ_e) are clustered in four regions, N-terminal (N2, L8–T12), residues around switch I (Q42–T44, A48), β2–β3 loop (D58), and C-terminal (L175), whereas those displaying motion on millisecond to microsecond timescale (R_ex) are found mainly in the secondary structure regions, including α1 helix (L31, R35), α2 helix (L78, I81, Y82), which is part of the switch II region, α3 helix (Q105, I108–E110, A113, A115), β5 strand (L123, G125) close to G4 loop, as well as loop regions including residues of switch I loop (Q43–T44), switch II loop (D68), and G4 loop (K128).

MD simulations of Rab5aMΔC

MD simulations provide the conformational dynamics of protein molecules by computationally probing the conformational energy landscape. MD simulations were performed for the Rab5aMΔC model for 100 ns to understand the structural state of P60–M79 insert. The RMSD profile for simulations is shown in Fig. S5 A, and it shows that after 5 ns the simulations get stabilized. RMSF is a measure of the deviation between the position of particle and some reference point. Residue-wise, RMSF was calculated for Rab5aMΔC, and its plot is shown in Fig. S5 B. The average value of RMSF is 0.16 nm. The MD simulations result shows that residues G62–P77, which map to the P60–M79 insert, do not form any secondary structure and, therefore, possess higher flexibility. Thus, the P60–M79 insert may be forming an unstructured long loop, which consequently reduces the stability of the Rab5aMΔC construct. Besides this, other regions that show flexibility corresponds to N-terminal (M1–E10), switch I (L39–I46), switch II (G91₇₁–R96₇₆), and G4 loop (L149₁₂₉–S156₁₃₆), which are found to be comparable with NMR relaxation data (residue numbers written in subscript correspond to the residue numbers after P60–M79 deletion). Few residues of α4 helix (F157–Q161) also showed high fluctuations, which were not observed in the NMR relaxation study.

Characterization of unfolding pathway of Rab5a

Far-UV CD and steady-state fluorescence spectroscopic changes in the presence of different concentrations of urea for Rab5aMΔCΔ60-79C107S show that concentrations of ≥6 M result in its complete unfolding (Fig. 1, A and B). The data obtained from these two independent probes (Fig. S6, A and B) were superimposable upon each other (Fig. S6 C). This seems to suggest that unfolding of Rab5aMΔCΔ60-79C107S is a two-state process.

We have also monitored the changes in the binding of the hydrophobic dye 4,4'-dianilino-1,1'-binaphthyl-5,5'-disulfonic acid (Bis-ANS) upon change in urea concentration for the characterization of unfolding pathway of Rab5aMΔCΔ60-79C107S (Fig. S6 D). Bis-ANS is proven to be a sensitive probe for the identification and characterization of the partially folded intermediates in protein-folding pathways. As can be seen from Fig. S6 D, in case of Rab5aMΔCΔ60-79C107S, a gradual increase in Bis-ANS/protein complex fluorescence was observed with increasing urea concentrations, reaching a maximum at ∼4.0 M urea. A further increase of urea concentration above 4.0 M led to a gradual decrease in the Bis-ANS fluorescence, which tailed off at 6 M urea concentration. Change in the Bis-ANS binding is the result of the changes in the hydrophobic patches of the protein. Increase in the Bis-ANS binding up to ∼4.0 M urea suggests the increase in solvent-exposed hydrophobic patches in the protein, whereas decrease in the Bis-ANS binding above ∼4.0 M urea results from the decrease in the hydrophobic patches due to unfolding. Maximal Bis-ANS binding at ∼4.0 M urea suggests the maximal exposure of hydrophobic patches in comparison to native and unfolded state of the protein. In Far-UV CD spectroscopic studies, change in the secondary structure of Rab5aMΔCΔ60-79C107S was initiated at 4 M urea, which suggests that at 3.5 M urea, Rab5aMΔCΔ60-79C107S retained its entire secondary structure, whereas Bis-ANS binding study clearly indicates change in the tertiary structure (change in hydrophobic patch) was initiated at ∼1 M urea and maximal hydrophobic patches were exposed due to change in tertiary structure at ∼4 M urea. The enhanced Bis-ANS binding and native-like secondary structure at ∼4 M urea indicate the existence of molten globule-like intermediate state around ∼4 M urea. Thus, a comparison of the data from Far-UV CD spectroscopy and Bis-ANS binding suggests the stabilization of at least one intermediate having a characteristic of the molten globule in unfolding pathway of Rab5aMΔCΔ60-79C107S.

To confirm the presence of folding intermediate, we further characterized the unfolding pathway of Rab5aMΔCΔ60-79C107S by size-exclusion chromatography, which is a well-established technique to identify the intermediates in unfolding pathway (58, 59), as well as to compare the hydrodynamic radii of the native, intermediate, and unfolded states of the protein (58, 59). In case of LdRab5a, “all or none transition” was observed from a compact (at higher elution volume with smaller molecular dimension) to a less compact state (lower elution volume with larger molecular dimension), which is suggested by bimodal distribution of protein elution volume in the transition region (3.5–4.5 M urea).

Elution profile of Rab5aMΔCΔ60-79C107S at different urea concentrations and the plots of urea dependence of the elution volume of compact and less compact state are shown in Figs. S7 and 5 A, respectively. The maximal elution volume of the compact state corresponds to the native state and is denoted by N. Elution volume of compact state continues to decrease up to a certain urea concentration at which all compact molecules disappear (i.e., 5 M urea).

Figure 5.

Urea-induced unfolding study of Rab5aMΔCΔ60-79C107S by SEC. (A) Dependence of elution volumes of compact, V_C, and less compact, V_LC, states with increasing urea concentration, shown by black circles and black squares, respectively. The urea dependence of the average elution volume, V_el (grey squares) was calculated from V_C and V_LC using the equation V_el = [1−F_LC] × V_C + F_LC × V_LC. The urea dependence of the fraction of less compact molecules (F_LC) is shown in black stars. (B) Fraction unfolded curve of Rab5aMΔCΔ60-79C107S obtained from average elution volume. (C) The fraction of the unfolded state (F_U) obtained from the fluorescence study, and the fractions of the less compact state (F_LC) and the denatured state (F_D) obtained from SEC, are shown in black circles, black stars, and black squares, respectively. (D) Population curves for the I1 and I2 states in terms of the fractions in the N-U transition.

The fraction (F_LC) of molecules that undergoes the transition from a compact (C) to a less compact (LC) state can be determined from the areas (S) of corresponding peaks:

$$F_{LC}=S_{LC}/\left(S_{LC}+S_C\right)$$ (1)

The urea dependence of the averaged elution volume V_el is calculated from V_C and V_LC by using the following equation:

$$V_{el}=\left[1-F_{LC}\right]\times V_C+F_{LC}\times V_{LC}$$ (2)

Further, we have converted the average elution volume into fraction unfolded (F_U). The curve obtained from the average elution volume seems to be a three-state curve (Fig. 5 B). Therefore, we have fitted the following three-state model equation:

$$S_{obs}=\left[S_N+S_I{e}^{-}\left(\frac{\mathit{\Delta }{G}_{NI}}{RT}\right)+S_U{e}^{-}\left(\frac{\mathit{\Delta }{G}_{NU}}{RT}\right)\right]/\left[1+e^{-}\left(\frac{\mathit{\Delta }{G}_{NI}}{RT}\right)+e^{-}\left(\frac{\mathit{\Delta }{G}_{NU}}{RT}\right)\right]$$ (3)

where S_obs is the observed signal at a given concentration of denaturant, whereas S_N, S_I, and S_U are the signals due to the native state, the intermediate state, and the unfolded state, respectively. ΔG_NI and ΔG_NU are the corresponding free energies of the transitions, N↔I and N↔U, where ΔG_NI and ΔG_NU are assumed to have linear dependence on denaturant concentration [D], resulting in the following equations, where m_NI[D] and m_NU[D] are the corresponding slopes of transitions from the native to the intermediate and native to the unfolded states, respectively.

$$\mathit{\Delta }{\mathrm{G}}_{\mathrm{NI}}=\mathit{\Delta }{\mathrm{G}}_{\mathrm{NI}}\left({\mathrm{H}}_2\mathrm{O}\right)+{\mathrm{m}}_{\mathrm{NI}}\left[\mathrm{D}\right]$$ (4)

$$\mathit{\Delta }{\mathrm{G}}_{\mathrm{NU}}=\mathit{\Delta }{\mathrm{G}}_{\mathrm{NU}}\left({\mathrm{H}}_2\mathrm{O}\right)+{\mathrm{m}}_{\mathrm{NU}}\left[\mathrm{D}\right]$$ (5)

The urea dependence of fraction unfolded (F_U), obtained from average elution volume of size-exclusion chromatography, with that for F_D, obtained from Far-UV CD spectroscopy and F_LC, is plotted in Fig. 5 C. Clearly, the curves for F_D and F_U are not superimposable with each other or with the curve of fraction of less compact state (Fig. 5 C). It suggests the existence of three different stages of unfolding of Rab5aMΔCΔ60-79C107S, with two intermediates. The size exclusion chromatography (SEC) data for Rab5aMΔCΔ60-79C107S is comparable to previously reported SEC data for carbonic anhydrase, β-lactamase, and CPR3, which have two intermediates in the unfolding pathway (60, 61, 62).

A plot of fractions of intermediate states (i.e., F_I1 and F_I2) as a function of urea concentration is shown in Fig. 5 D, which helps in visualizing the changes in the fraction of intermediates as a function of urea concentration in N↔U transition. The fraction of intermediate states was evaluated by using the following equations:

$$F_{I1}=F_U-F_{LC}$$ (6)

$$F_{I2}=F_{LC}-F_D$$ (7)

Fig. 5 D clearly demonstrates the presence of two intermediates in unfolding pathway of Rab5aMΔCΔ60-79C107S, which are maximally populated at ∼2.5 M and ∼4 M urea. Therefore, on the basis of both optical (fluorescence and CD spectroscopy) methods and size exclusion chromatography, we infer that the urea-induced unfolding pathway of Rab5aMΔCΔ60-79C107S is a four-state process, with the presence of two intermediates maximally populated at ∼2.5 and ∼4 M urea, respectively.

The presence of the two intermediates can also be observed from the Bis-ANS fluorescence assay. We observed that the curve in Fig. S8 was asymmetric, which allowed its deconvolution into two curves. This deconvolution results in the observation of two curves having different maxima at ∼2.0 and ∼4.0 M urea. This, in fact, suggests population of two intermediates along the unfolding curve, which correlate with I1 (∼2.5 M) and I2 (∼4.0 M) states observed by SEC (Fig. 5 D).

Discussion

RabGTPases are gaining importance in human diseases such as cancer because of their critical role in endocytosis. Several structures for RabGTPases have been determined in the last decade. Just like the Ras proteins, the dynamics of the loops play a very important role in the activation and interactions of Rabs. Surprisingly, none of the Rab protein structure has been determined by using NMR spectroscopy, a technique which provides atomic-level details about the structural interactions and dynamics. Further interest came from functional studies on Rab5a, Rab5b, and Rab7 proteins from L. donovani (15, 17, 63).

From our preliminary studies, we selected LdRab5a for NMR characterization. We began with Q93L GTP-locked mutant of LdRab5a. Deletion of the unfolded C-terminal region was performed. Based on sequence homology, it was found that before the unfolded C-terminal region, there was an α-helix. After this C-terminal α-helix, there were 37 residues. However, a 37-residue deletion at the C-terminus gave a construct with poor solubility and yield. Fortuitously, a 40-residue deletion at the C-terminus gave a more soluble and stable protein. The C-terminal of our construct matches with that of HsRab5c (PDB: 1Z07) (24). ¹⁵N-labeling and three-dimensional HNCA and CBCA(CO)NH experiments were carried out with this construct. However, the protein tended to precipitate soon, and very few assignments could be performed. Introduction of T45A or T45S mutations did not improve the physicochemical or conformational stability. At this stage, the deletion of the LdRab5a specific extra insert (60–79) between β2 and β3 was performed. At the same time, by comparison with other Rab sequences, the two prolines at the beginning of this stretch (P58 and P59) were mutated to D and G, respectively. This construct was significantly more stable, and it allowed for the assignment of ∼100 residues. As further improvement was needed, C107S mutation was performed, and the protein was expressed as a fusion construct downstream of GST. Almost complete NMR assignment and structure determination was amenable with this construct. Although glycerol, GTP, MgCl₂, and DTT helped in stabilizing the protein, other additives such as proline, aspartic acid, arginine, C₈E₄, and betaine did not have much effect. We also tried to mutate just the two proline P58 and P59, without deleting the insert; however, it could not be integrated into the final scheme. Now that we have obtained nearly complete assignments, we can systematically and sequentially reverse some of these mutations and truncations to get information about native structure, stability, and folding. Since the β2–β3 region, which is often termed as the interswitch region, is a site for effector binding, it is of great interest to us to assign and structurally characterize this region. We have attempted to generate P58D, P59G double mutant of LdRab5aMΔC, and also LdRab5aMΔCC107S mutant. However, the expression of both of these mutants was poor. Further standardization is needed to characterize LdRab5a with intact insert region.

After the deletion of C-terminal and 20 extra residues patch (P60–M79), we replaced the single Cys residue by Ser residue (C107S mutant). This resulted in increased stability of the protein. Upon comparison of the stability of Rab5aMΔCΔ60-79 with that of Rab5aMΔCΔ60-79C107S by using CD, fluorescence spectroscopy, DSC, and NMR spectroscopy, we found significant differences in the stabilities of both of the constructs (Figs. 1 and S2). As observed through CD and fluorescence spectroscopy, the urea Cm values for the Rab5aMΔCΔ60-79 and Rab5aMΔCΔ60-79C107S were ∼3.5 and 4.5 M, respectively (Fig. 1, A and B). Here, it is important to mention that the Rab5aMΔCΔ60-79 was expressed as a fusion protein with N-terminal His-tag by using a pQE30 vector, whereas the Rab5aMΔCΔ60-79C107S protein was expressed as a fusion protein with GST and was obtained after proteolysis with thrombin. Therefore, it is possible that Rab5aMΔCΔ60-79 has more heterogeneity, which could be the reason for its lower thermal, chemical, and physicochemical stability. Unfolding of both of the constructs seems to be a two-state process, but the C_m values obtained for the two constructs are different. It is also possible that the replacement of Cys residue with Ser residue directly contributes to increased stability of the protein, as evidenced by ∼10°C increase in the Tm value, as observed in the DSC experiment (Fig. 1, C and D). To find out the reason behind the increased stability of protein, we have compared the solution structure of Rab5aMΔCΔ60-79C107S and modeled structure of the Rab5aMΔCΔ60-79. Through this structural comparison, we found that OG of the hydroxyl group of substituted S107 is forming an additional H-bond between hydroxyl group of serine and NH of K15 present in β1. At the same time, C107S mutation also eliminates the possibility of di-sulphide bond formation, thus reducing the aggregation and subsequent precipitation of the protein. In NMR spectroscopy, we found the Rab5aMΔCΔ60-79C107S protein to be stable for a longer time (6–7 days), which facilitated the structure determination and backbone dynamics study.

In LdRab5a, main chain amide proton assignments were not made for the residues G47, L72, E73, R74, S77, and A79, which is less in number in comparison to guanosine 5'-(β, γ-imido)triphosphate (GMPPNP)-bound H-Ras (active form of Ras), in which ¹⁵N-¹H HSQC crosspeaks were missing for 22 nonproline residues (25). All nonproline residues were assigned in case of GDP-bound H-Ras (inactive form of Ras), and GMPPNP-bound H-Ras at 5°C (26). For RalB bound to GMPPNP, six residues from switch I and switch II regions could not be assigned (64). In active form of H-Ras, residues of missing crosspeaks are all from the functionally important regions involved in binding of the β- and γ-phosphate groups of GTP for GTP hydrolysis and for the interaction with downstream effectors and regulators (25). Similarly, in LdRab5a, the missing ¹H-¹⁵N HSQC crosspeaks were found for the residues present next to switch I (G47) and switch II (L72, E73, R74, S77 and A79). The signal for these residues were extremely broadened because of chemical exchange at an intermediate rate on the NMR time scale, and these residues are in close proximity with each other in the tertiary structure of protein. Therefore, these residues may concertedly function in coordination with active site for GTP hydrolysis and for the GTP-dependent interaction with the downstream effectors and regulators.

Solution structure of the Rab5aMΔCΔ60-79C107S shows characteristic G-domain fold, which consists of a central six stranded mixed β-sheet surrounded by five helices. In addition to the secondary structure, all five loop regions G1–G5 are also well defined in the structure of Rab5aMΔCΔ60-79C107S. Positions of both switch I and switch II regions are found in close proximity as in the GNP-bound structure of human Rab5a (Fig. 3 D) (20). In backbone dynamics study, residues of switch II region and loop connecting β2–β3 show motion on ns to ps timescale in addition to 13 residues of N-terminal and two residues of C-terminal. Surprisingly, residues of the switch II region also show motion on ms to μs timescale. In contrast to switch I region, residues of switch II region show high S² value (∼1.0) and are fairly rigid, but the residues of α2 helix (a part of switch II as mentioned by Dumas et al) (21) present after switch II region show motion on ms to μs timescale, which is consistent with the presence of smaller helix, α2, in comparison to other GTP/GppNHp-bound structures and missing ¹H-¹⁵N HSQC crosspeaks for E73-R74, S77, and A79. K128 residue of the G4 loop along with G125 residue, preceding the G4 loop, were also in conformational exchange, whereas the other three residues of the G4 loop are rigid in nature. Similarly, two residues (A156–S157) of the G5 loop were also found in conformational exchange. Besides G2–G5 loop, residues of α4, the loop connecting α4–β5, and α5 were also found in conformation exchange. In the GMPPNP-bound structure of H-Ras, only six residues (V7, V8, V14, L56, A66, and Q70) were found in motion at ms to μs timescale, and 22 residues were found in motion at an intermediate time scale (25). In contrast, in GTP-bound Rab5aMΔCΔ60-79C107S, six residues show motion at an intermediate time scale and 32 residues show motion at ms to μs timescale. Therefore, on the basis of comparison of backbone dynamics, there are noticeable differences in the dynamics of GTP-bound Rab5aMΔCΔ60-79C107S and GMPPNP-bound H-Ras. Further, based on the reported binding affinities of H-Ras (1.6 × 10¹¹ M⁻¹) and human Rab5 (3.5 × 10¹⁰ M⁻¹) proteins (65, 66), there appears to be an inverse correlation, albeit qualitative, between the dynamics of P-loop, switch I and switch II regions, and the GTP binding affinity. The dynamic profile may further complement the residue-specific complementarity in determining the specificity of interaction with the effectors.

Crystal structures of the Rab proteins from the different organisms in both the bound and unbound form have been deposited in the PDB. However, very few studies have been reported on the folding and unfolding pathways of Rab and other GTPase proteins. Biophysical characterization from the folding/unfolding point of view has not been done for any of the Rab protein from L. donovani. In this context, in addition to structure and dynamics study, we explored here the unfolding pathway of Rab5aMΔCΔ60-79C107S using the chemical denaturant urea. In our investigations, we observed that the denaturation curves obtained by the secondary (far UV-CD spectroscopy) and tertiary (fluorescence spectroscopy) structure probes were superimposable on each other, which suggests that the unfolding pathway of Rab5aMΔCΔ60-79C107S is a two-state reaction, N-U. We have further studied the unfolding pathway of Rab5aMΔCΔ60-79C107S by SEC and Bis-ANS binding. The data obtained from SEC, and its comparison with the data obtained from the intrinsic fluorescence/far UV-CD spectroscopy, revealed the presence of two intermediates, I1 and I2, which are maximally populated at ∼2.5 and ∼4.0 M urea, respectively. Bis-ANS binding studies and deconvolution of its data in two curves further verified the presence of two intermediates, I1 and I2. Thus, the unfolding pathway of LdRab5a is a four-state reaction. The I1 state of the Rab5a is characterized by 1) native-like entire secondary structure, 2) reduced tertiary structure, and 3) increased molecular dimension, whereas I2 state is characterized by 1) native-like but reduced secondary and tertiary structure and 2) increased molecular dimension in comparison to native and I1 state. In addition to this, exposure of tryptophan residues in I1 state seems to be similar to the native state, whereas in I2 state, exposure of tryptophan gets increased. In the Bis-ANS binding experiments, we found that both I1 and I2 states show strong interaction with Bis-ANS. However, Bis-ANS binding in I2 state is relatively high in comparison to I1 state. The emission intensity of Bis-ANS upon binding to the Rab5aMΔCΔ60-79C107S in 4 M urea is approximately twice that observed with the protein in its native state. This observation suggests the molten globule-like characteristic of the I2 state. The implication of the molten globules in several physiological processes (protein-ligand interaction, enzymatic activity, assisting chaperones, translocation of toxins, and genetic diseases involving misfolded proteins) has been proposed previously (67). Proteins have been also proposed to translocate across bio-membranes in their molten globule states (67). Rab subfamily consists of ∼70 members in humans, and each family member has a specific intracellular localization. In this context, the identification and characterization of a molten globule-like state in the unfolding pathway of Rab5a has some significance.

Earlier unfolding studies performed for the p21^H-ras protein, a member of Ras family, show that it unfolds via the two-state model, involving the native and unfolded protein, whereas Rab5aMΔCΔ60-79C107S, a member of Rab family, unfolds via a four-state model with the stabilization of two intermediates, and thus it is different from p21^H-ras protein. Although both proteins belong to Ras superfamily and possess a characteristic GTPase fold, the difference in their unfolding profile could also arise because of differences in their primary sequences (∼73% similar). Such differences in the protein unfolding pathway are not restricted to the GTPase proteins but is also reported for other proteins, including members of the GST family (PfGST and PvGST) (68), cyclophilins (CPR3, LdCyp, and PPiA) (49, 62, 69), etc. Earlier studies suggest that the difference in the number of hydrophobic residues at the folding initiation site may give rise to such difference in the unfolding pathway of proteins that belong to same class (70). Finally, our NMR characterization of the folded state of truncated LdRab5a would also allow for detailed atomic-level characterization of the unfolding process in the presence of urea by NMR spectroscopy.

Conclusions

In conclusion, we have presented the first, to our knowledge, NMR solution structure of Rab protein and have characterized the backbone dynamics of Rab protein. Rab5aMΔCΔ60-79C107S belong to α/β-sheet class of protein, which contains a characteristic G-domain fold with six stranded mixed β-sheet surrounded by five helices. The stabilization of LdRab5a was carried out by systematic deletions and mutations that did not affect the characteristic fold. Deletion of the extra region of LdRab5a slightly increased the stability of protein and also increased the sharpness of linewidths in NMR spectrum. Replacement of Cys107 residue by Ser and its expression as a GST-fusion protein further increased the thermal stability of protein by 10°C and also increased the stability toward the chemical denaturant urea, which might be due to formation of an additional hydrogen bond and reducing the spontaneous aggregation because of formation of disulphide bonds. The urea unfolding pathway of Rab5aMΔCΔ60-79C107S was found to be a four-state process (N, I1, I2, and U states) with stabilization of two intermediates. Further, we found molten globule-like characteristic features of the I2 state, which has native-like substantial secondary structure.

Author Contributions

Subcloning, expression, purification, NMR assignment, structure calculation of LdRab5a, protein unfolding study by intrinsic fluorescence, CD spectroscopy, and SEC were done by, D.M.; R.R. performed cloning and functional characterization of LdRab5aM (Q93L) mutant under A.M., based on which this study was planned. R.Y. cloned Rab5aΔCM. R₁, R₂ and steady-state heteronuclear {¹H}-¹⁵N NOE data analysis and DSC experiments were done by A.J. and S.T., respectively; A.A. conceived, designed, and directed the study, acquired NMR data, analyzed the results, and wrote the article.

Editor: Wendy Shaw.