Abstract

The LRR (leucine-rich repeat)–Roc (Ras of complex proteins)–COR (C-terminal of Roc) domains are central to the action of nearly all Roco proteins, including the Parkinson's disease-associated protein LRRK2 (leucine-rich repeat kinase 2). We previously demonstrated that the Roco protein from Chlorobium tepidum (CtRoco) undergoes a dimer–monomer cycle during the GTPase reaction, with the protein being mainly dimeric in the nucleotide-free and GDP (guanosine-5′-diphosphate)-bound states and monomeric in the GTP (guanosine-5′-triphosphate)-bound state. Here, we report a crystal structure of CtRoco in the nucleotide-free state showing for the first time the arrangement of the LRR–Roc–COR. This structure reveals a compact dimeric arrangement and shows an unanticipated intimate interaction between the Roc GTPase domains in the dimer interface, involving residues from the P-loop, the switch II loop, the G4 region and a loop which we named the ‘Roc dimerization loop’. Hydrogen–deuterium exchange coupled to mass spectrometry (HDX-MS) is subsequently used to highlight structural alterations induced by individual steps along the GTPase cycle. The structure and HDX-MS data propose a pathway linking nucleotide binding to monomerization and relaying the conformational changes via the Roc switch II to the LRR and COR domains. Together, this work provides important new insights in the regulation of the Roco proteins.

Introduction

Roco proteins are a class of G proteins that occur in all three domains of life and are characterized by the presence of a Roc (Ras of complex proteins) and a COR (C-terminal of Roc) domain that invariably occur together and are therefore believed to form one functional supradomain [14]. While the Roc domain belongs to the superfamily of the Ras-like small GTPases, the function of the COR domain is not entirely known although it has been shown that this domain contributes to the homodimerization of the Roco proteins [5,6]. Apart from the Roc and COR domains, Roco proteins can contain several other domains, and based on the domain architecture, a classification in three groups has been proposed [1,7]. Group 1 Roco proteins display the simplest domain architecture, consisting of the typical Roc–COR domains preceded by a leucine-rich repeat (LRR) domain. Group 2 Roco proteins are characterized by the minimal presence of the LRR, Roc and COR domains, invariably followed by a protein kinase domain plus optional additional domains. Group 3 Roco proteins are characterized by a protein kinase domain that precedes the Roc–COR supradomain and the absence of the LRR domain. In human, four Roco protein members are found, distributed over the three groups: MFHAS1 (malignant fibrous histiocytoma amplified sequence 1, previously MASL1) belonging to group 1, LRRK1 and LRRK2 (leucine-rich repeat kinases 1 and 2) belonging to group 2, and DAPK1 (death-associated protein kinase 1) belonging to group 3 [3,813].

Research on Roco proteins significantly intensified since 2004, when single-point mutations in LRRK2 were linked to autosomal familial Parkinson's disease (PD) [14,15]. PD-linked mutations are mainly concentrated in the catalytic Roc–COR and kinase domains of LRRK2. While the clinical PD mutations in the Roc domain decrease the GTPase activity, the most prevalent PD mutation in the kinase domain (G2019S) results in an enhanced kinase activity [1622].

Despite the huge interest in LRRK2 and related Roco proteins, large gaps are remaining in our understanding of the detailed molecular mechanism of the LRR–Roc–COR domains, common to group 1 and 2 Roco proteins. In 2008, Deng et al. [23] reported a crystal structure of the isolated Roc domain of LRRK2, showing a very unusual domain swapped dimer. However, subsequent studies suggest that the domain swapped arrangement likely does not reflect the true conformation of the Roc domain in solution [19]. At about the same time, Gotthardt et al. [5] reported the structure of the Roc–COR supradomain of the group 1 Roco protein from the bacterium Chlorobium tepidum. This structure showed that homodimerization occurs through the COR domains, but resolved only one of the two Roc domains in the dimer. More recently, the crystal structure of the C. tepidum Roc–COR supradomain, together with cross-linking data and low-resolution small-angle X-ray scattering (SAXS) and electron microscopy (EM) data, was used to generate a computational model of the three-dimensional arrangement of a LRRK2 dimer [24].

We discovered that rather than acting as permanent dimers, the Roco proteins undergo a complex dimer–monomer cycle tuned by GTP (guanosine-5′-triphosphate) binding and hydrolysis [25]. Concomitant with the GTPase reaction, the protein cycles between a compact dimeric state and a seemingly more elongated monomeric state. Moreover, mutations analogous to PD-linked mutations decrease the GTPase activity by interfering with this dimer–monomer cycle. However, the exact nature and mechanism of the nucleotide-induced changes in conformation and oligomerization remained unknown.

Here, we report the first crystal structure showing the arrangement of the LRR–Roc–COR domains of the C. tepidum Roco protein in its nucleotide-free form. This structure reveals a compact dimeric arrangement where the LRR domains fold back over the COR domains and, for the first time, shows the intimate interactions between the Roc domains in the dimer interface involving residues from the P-loop, the switch II loop, the G4 region and a dedicated loop which we named the ‘Roc dimerization loop’. The structure, moreover, proposes a mechanism for linking nucleotide binding to Roco monomerization and intra-subunit domain movements. Hydrogen–deuterium exchange coupled to mass spectrometry (HDX-MS) was then used to map the conformational changes upon GTP binding, GTP hydrolysis and GDP (guanosine-5′-diphosphate) release and reveal the major events underlying the complex conformational cycle of the Roco proteins. In conclusion, this work provides the first detailed insights in the regulation of the LRR–Roc–COR module underlying the action of Roco proteins.

Experimental procedures

Protein production, crystallization and data collection

The protocols used for the production of the Roco protein from Chlorobium tepidum (CtRoco), the CtRoc–COR and the CtLRR–Roc–COR proteins, and the crystallization and the data collection of the CtLRR–Roc–COR protein have been described previously [5,25,26]. In brief, recombinant protein expression was performed starting from the open reading frames cloned in pProEX plasmids and transformed in Escherichia coli BL21 (DE3) cells. All proteins were purified using a three-step protocol consisting of metal chelate affinity chromatography on Ni-NTA, anion exchange chromatography on a source Q30 column and size-exclusion chromatography on a Superdex 200 column. Before setting up the crystals, the protein was made nucleotide-free by adding an excess of EDTA and loading it on a S200 26/60 gel filtration column equilibrated with 20 mM Hepes (pH 7.5), 150 mM NaCl, 1 mM DTT and 5% glycerol. Afterwards, 5 mM MgCl2 was added to the protein and the nucleotide-load was determined using reversed-phase chromatography coupled to HPLC. Crystals were obtained in 0.8 M sodium formate, 0.1 M sodium acetate (pH 5.5), 10% PEG8000 and 10% PEG1000, and data collection was performed at the PROXIMA2 beamline of the SOLEIL synchrotron (Paris, France) using an EIGER detector (Dectris) [26].

X-ray crystallography data processing and refinement

Data indexing was done using the XDS suite [27] followed by scaling and merging with the AIMLESS program [28]. The phase problem was solved by molecular replacement using the program Phaser and the structures of the individual CtRoc, CtCOR and CtLRR domains as search models (PDB entries: 3dpu and 5il7) [5,24,29]. The initial model was completed and refined by cycles of manual building in coot and automated refinement using phenix.refine [30].

Size-exclusion chromatography-SAXS experiment

The protocol used for the size-exclusion chromatography (SEC)-SAXS analysis of the CtLRR–Roc–COR protein has been described previously [24,25]. In brief, data were collected on the BM29 beamline at the ESRF in Grenoble (France) using an inline HPLC set-up. About 50 µl of an 8 mg/ml sample was injected on a Bio SEC-3 HPLC column (Agilent) equilibrated with 20 mM Hepes (pH 7.5), 150 mM NaCl, 5 mM MgCl2, 5% glycerol and 1 mM DTT. Initial processing of the data was done with DATASW [31]. The crystallographic model was compared with the experimental SAXS data using CRYSOL [32].

SEC-multi-angle light scattering experiment

SEC-MALS (multi-angle light scattering) analysis was performed by injecting 10 µl of an 8 mg ml−1 protein sample on a Bio SEC-3 HPLC column (Agilent, 3 µm 300 Å) that was equilibrated with 20 mM Hepes (pH 7.5), 150 mM NaCl, 5 mM MgCl2, 5% glycerol and 1 mM DTT. For data collection in the presence of nucleotides, the nucleotide-free proteins were pre-incubated with 1 mM of the nucleotide and 200 µM (molar excess) of the nucleotide was added to the running buffer. A Dawn Heleos detector (using 9 angles) and an Optilab T-rEX detector (Wyatt Technology) were attached to the HPLC system (Shimadzu). The molar masses were calculated with the ASTRA 5.3.4.20 software.

Hydrogen deuterium exchange mass spectrometry

HDX-MS was performed essentially as previously described [3335]. Proteins in 20 mM Tris (pH 7.5), 150 mM NaCl, 5 mM MgCl2 and 10% glycerol were diluted 6 : 4 with 8 M urea, 1% trifluoroacetic acid, and passed over an immobilized pepsin column (2.1 mm × 30 mm, ThermoFisher Scientific) in 0.1% trifluoroacetic acid, 5% methanol at 15°C. Peptides were captured on a reversed-phase C8 cartridge, desalted and separated by a Zorbax 300SB-C18 column (Agilent) at 1°C using a 5–40% acetonitrile gradient containing 0.1% formic acid over 12 min and electrosprayed directly into an Orbitrap mass spectrometer (LTQ-Orbitrap XL, ThermoFisher Scientific) with a T-piece split flow set-up (1 : 400). Data were collected in profile mode with source parameters: spray voltage 3.4 kV, capillary voltage 40 V, tube lens 170 V and capillary temperature 170°C. Where applicable, MS/MS CID fragment ions were detected in centroid mode with an AGC target value of 104. CID fragmentation was 35% normalized collision energy (NCE) for 30 ms at Q of 0.25. HCD fragmentation NCE was 35 eV. Peptides were identified using Mascot (Matrix Science) and manually verified to remove ambiguous peptides. For measurement of deuterium uptake, protein was diluted 1 : 9 in 20 mM Tris (pH 7.5), 150 mM NaCl and 5 mM MgCl2 prepared in deuterated solvent. Where applicable, 1 mM nucleotide was added to the buffer to ensure saturation of the binding site after dilution into deuterated buffer. Samples were incubated for 0, 60 or 900 s at 22°C followed by the aforementioned digestion, desalting, separation and mass spectrometry steps. The intensity weighted average m/z value of a peptide's isotopic envelope is compared plus and minus deuteration using the HDX workbench software platform [36]. Individual peptides are verified by manual inspection. Sequence repetition within the LRR was unexpectedly high. In total, 16 peptides were found in repetition. The HDX workbench software was unable to distinguish between them and thus assigned each repeated peptide to the first available slot rather than all possible slots and peptide coverage within this region appeared to suffer. However, once taking this into account, the actual peptide coverage was calculated to be 95%. The apparent structure for each isolated peptide and thus deuterium uptake was similar to the other peptides within each repeated ‘family’. In addition, the deuterium uptake in this region was unaltered by nucleotide binding. Thus, conclusions regarding structural alterations upon nucleotide binding were unaffected by the sequence repetition. Data are visualized using Pymol. Where applicable, deuterium uptake was normalized for back-exchange by comparing deuterium uptake to a sample incubated in 6 M urea, 1.2 µM pepsin, 10 mM NaH2PO4 prepared in deuterated buffer for 12–18 h at room temperature and processed as indicated above.

Results

The structure of nucleotide-free C. tepidum LRR–Roc–COR shows a compact dimeric arrangement

In the nucleotide-free state, the Roco protein from the bacterium C. tepidum (CtRoco) occurs as a homodimer in solution [25], where each subunit is 1102 amino acids long and consists of an LRR, Roc and COR domain followed by a C-terminal stretch of 156 amino acids (Supplementary Figure S1). While the presence of the LRR, Roc and COR domains is generally conserved among groups 1 and 2 Roco proteins, the C-terminal domain is highly variable among different orthologues. To gain insights into the structure and function of the hallmark LRR–Roc–COR domains, we previously reported the purification and crystallization of a domain construct of CtRoco consisting of amino acids 1–946 (CtLRR–Roc–COR) and collected X-ray diffraction data to 3.3 Å resolution [26]. Here, the structure was solved using molecular replacement and was refined to 3.3 Å and a Rwork of 0.23 and a Rfree of 0.28 (Table 1).

Table 1
Data collection and refinement statistics1,2
Data collection 
 X-ray source PROXIMA 2, SOLEIL 
 Wavelength (Å) 0.9801 
 Detector EIGER X 9M 
 Rotation range per image (°) 0.1 
 Total rotation range (°) 120 
 Space group P212121 
a, b, c (Å) 95.6, 129.8, 179.5 
α, β, γ (°) 90°, 90°, 90° 
 Resolution range (Å) 47.8–3.3 (3.4–3.3) 
 Total number of reflections 155 879(13 110) 
 No. of unique reflections 34 302(3210) 
 Completeness 0.99 (0.96) 
 Multiplicity 4.5 (4.1) 
 Mean I/σ(I9.3 (1.5) 
 Rmeas 0.158 (1.103) 
 CC ½ 0.995 (0.539) 
Refinement 
 Rwork/Rfree 0.2305/0.2813 
 No. of atoms 13 680 
 Average B-factor, all atoms (Ų) 87 
 Wilson B-factor (Ų) 85 
 Bond length RMS (deviation) (Å) 0.011 
 Bond angle RMS (deviation) (°) 0.901 
 Ramachandran favoured/allowed/outliers (%) 89/10/1 
PDB Code 6HLU 
Data collection 
 X-ray source PROXIMA 2, SOLEIL 
 Wavelength (Å) 0.9801 
 Detector EIGER X 9M 
 Rotation range per image (°) 0.1 
 Total rotation range (°) 120 
 Space group P212121 
a, b, c (Å) 95.6, 129.8, 179.5 
α, β, γ (°) 90°, 90°, 90° 
 Resolution range (Å) 47.8–3.3 (3.4–3.3) 
 Total number of reflections 155 879(13 110) 
 No. of unique reflections 34 302(3210) 
 Completeness 0.99 (0.96) 
 Multiplicity 4.5 (4.1) 
 Mean I/σ(I9.3 (1.5) 
 Rmeas 0.158 (1.103) 
 CC ½ 0.995 (0.539) 
Refinement 
 Rwork/Rfree 0.2305/0.2813 
 No. of atoms 13 680 
 Average B-factor, all atoms (Ų) 87 
 Wilson B-factor (Ų) 85 
 Bond length RMS (deviation) (Å) 0.011 
 Bond angle RMS (deviation) (°) 0.901 
 Ramachandran favoured/allowed/outliers (%) 89/10/1 
PDB Code 6HLU 
1

Data collection parameters are also published in ref. [26].

2

Values in parentheses are for the highest resolution shell.

Two peptide chains are found in the asymmetric unit, revealing in both copies the presence of the LRR domain (residues 2–410), a linker region between the LRR and the Roc domains (residues 411–450), the Roc domain (residues 451–618), a short linker between the Roc and the COR domains (residues 619–629) and the COR domain (residues 630–940). As previously suggested, the COR domain is composed of two subdomains, which we will further refer to as N-COR (residues 630–775) and C-COR (residues 796–940), linked by a flexible peptide stretch (residues 776–795). Both peptide chains could be nearly completely traced and modeled except for four regions in each chain that could not be modeled due to lack of electron density: residues 476–485, 601–605, 733–744 and 786–789 in chain A and residues 474–486, 601–605, 729–747 and 792–797 in chain B (Supplementary Figure S2).

The two protomers present in the asymmetric unit correspond to the presumptive biological dimer that is also present in solution in the nucleotide-free state [5,25]. Overall, the protein dimer adopts a compact arrangement, where the LRR domains fold back upon the Roc–COR domains within the same protomer (Figure 1A). Comparison of the theoretical SAXS curve calculated from our structure with the experimental SAXS data of the CtLRR–Roc–COR protein in solution in the absence of any nucleotides [24] yields a χ2-value of 1.16 (Figure 1B) [32]. This very good match between the experimental and theoretical scattering curves indicates that the compact arrangement shown by our crystal structure is a good representation of the major protein conformation in solution.

Structure of the LRR–Roc–COR domain construct of CtRoco.

Figure 1.
Structure of the LRR–Roc–COR domain construct of CtRoco.

(A) Crystal structure of the CtLRR–Roc–COR homodimer in its nucleotide-free state, shown in two orientations. The LRR domain is colored yellow, the Roc domain red, the N-COR cyan, the C-COR blue and the connecting regions orange. The corresponding domains of the B-subunit are colored in a paler shade. (B) Comparison using CRYSOL of the theoretical scattering curve derived from the dimeric CtLRR–Roc–COR crystal structure (red line) with the experimental solution scattering curve of CtLRR–Roc–COR in the absence of nucleotides (black dots). (C) Zoom in on the Roc domains of the CtLRR–Roc–COR dimer crystal structure. The Roc domains from both subunits are colored dark and light gray respectively, with the most important regions colored as follows: P-loop: yellow, switch I: cyan, switch II: green, Roc dimerization loop: magenta, G4 loop: orange, G5 loop: salmon. The GppNHp molecules, that were placed in the active site by superposition on the structure of Ras in complex with GppNHp (PDB 5P21), are shown in transparent stick representation with carbon atoms colored green. Placement of GppNHp in the nucleotide-binding site would lead to steric clashes with the P-loop due to the restrained conformation of the P-loop in the dimer interface.

Figure 1.
Structure of the LRR–Roc–COR domain construct of CtRoco.

(A) Crystal structure of the CtLRR–Roc–COR homodimer in its nucleotide-free state, shown in two orientations. The LRR domain is colored yellow, the Roc domain red, the N-COR cyan, the C-COR blue and the connecting regions orange. The corresponding domains of the B-subunit are colored in a paler shade. (B) Comparison using CRYSOL of the theoretical scattering curve derived from the dimeric CtLRR–Roc–COR crystal structure (red line) with the experimental solution scattering curve of CtLRR–Roc–COR in the absence of nucleotides (black dots). (C) Zoom in on the Roc domains of the CtLRR–Roc–COR dimer crystal structure. The Roc domains from both subunits are colored dark and light gray respectively, with the most important regions colored as follows: P-loop: yellow, switch I: cyan, switch II: green, Roc dimerization loop: magenta, G4 loop: orange, G5 loop: salmon. The GppNHp molecules, that were placed in the active site by superposition on the structure of Ras in complex with GppNHp (PDB 5P21), are shown in transparent stick representation with carbon atoms colored green. Placement of GppNHp in the nucleotide-binding site would lead to steric clashes with the P-loop due to the restrained conformation of the P-loop in the dimer interface.

Within the homodimer, the two protomers are nearly identical with an overall r.m.s.d. of 0.589 Å upon superposition, and with only a very small difference in the relative orientation of the C-COR domains with respect to the other domains. The LRR domains adopt the typical horseshoe shape consisting of 17 successive structural repeat motifs [24]. This LRR domain is connected to the Roc GTPase domain via a stretch of 40 amino acids (residues 411–450). The first residues of this connecting region are rich in proline and consist of a 414PLxxPPPE421 motif, of which Pro414, Leu415, Pro418 and Glu421 are highly conserved among the prokaryotic Roco proteins. The following residues of the linker region are folded in one short and one longer helix. The latter helix interacts with the first three helices of the N-COR domain and was previously shown to be important for the stable expression of the Roc–COR domain construct of the CtRoco protein [5].

The Roc GTPase domains adopt a very similar conformation as previously observed for the Roc domain in the structure of the CtRoc–COR construct (PDB: 3DPU) [5]. However, in contrast with the latter structure, where only one of the two Roc domains in the homodimer was resolved; both Roc domains are clearly resolved and could be unambiguously built in the current structure. Similar to other small G proteins, the Roc domain adopts a canonical arrangement with a central six-stranded β-sheet flanked on both sides by α-helices and displays the highly conserved G1–G5 motifs [37,38]. While the P-loop (G1, residues 457–466), the switch II region (G3, residues 513–532) and the loop spanning the G4 motif (residues 569–580) could be entirely traced, electron density is missing due to flexibility for a part of the switch I region (G2, residues 470–489) and the loop spanning the G5 motif (residues 597–607) (Figure 1C). Since the latter two regions typically interact with the γ-phosphate and guanine base of the nucleotide, respectively, it is expected that these regions will become structured upon GTP/GDP binding. One marked difference in the conformation of the Roc domain of the current CtLRR–Roc–COR structure in comparison with the previously determined CtRoc–COR structure (3DPU) is found in the P-loop. An additional turn is added at the N-terminus of the first α-helix of the Roc domain, corresponding to residues 461GMAG464 of the conserved G1 motif 459GxxxxGKT466. As a consequence of this conformational difference, fitting of a nucleotide in the active site of the CtLRR–Roc–COR structure by superposition with GppNHp (guanosine-5′-[(β,γ)-imido]triphosphate)-bound Ras [39] would result in severe steric clashes between these residues of the P-loop and the phosphates of the nucleotide (Figure 1C). This means that conformational changes of the P-loop residues would be required to allow nucleotide binding. Intriguingly, this same region of the P-loop is also implicated in the Roc–Roc dimer interface by forming interactions with the same P-loop residues and with residue Arg543 of the 541–551 loop (‘Roc dimerization loop’) of the adjacent subunit (see further).

The Roc GTPase domain is connected to the COR domain via a short stretch of 10 amino acids that also contains two Pro residues (Pro620 and Pro627) separated by one turn of α-helix. The COR domain adopts the same overall fold as previously observed for the CtRoc–COR structure [5] and consists of an N-terminal N-COR subdomain and a C-terminal C-COR subdomain connected by a 20 amino acid long linker that is partially flexible. N-COR is mainly α-helical and consists of two 3-helix bundles and a three-stranded β-sheet. One relatively long loop (spanning residues 733–744 and 729–747 in the A chain and B chain, respectively) within the second 3-helix bundle could not be modeled due to flexibility. The C-COR domain consists of a mixed six-stranded β-sheet flanked by two α-helices, followed by a C-terminal part consisting of two α-helices and a β-hairpin that is involved in dimer formation (see further).

The Roc domains contribute significantly to the Roco dimer interface

While, in the previously available structure of the CtRoc–COR dimer, only one of the two Roc domains was visible [5], the current structure of CtLRR–Roc–COR reveals both Roc domains, allowing us to study the complete dimer interface in detail and revealing contacts not previously visible.

Analysis with the PISA server shows that the two protein chains in the asymmetric unit of the crystal structure, forming the presumptive biological dimer, bury a total surface interface of 2085 Ų per subunit (Figure 2A) [40]. While it was previously proposed that the COR–COR interaction would be the main determinant for protein dimerization [5], this interaction only buries an interface area of 630 Ų per subunit, accounting for 30% of the total interface area. Our structure shows more elaborate interactions between the Roc domain of one subunit and the Roc and COR domains of the adjacent subunit. These interactions represent an area of 1455 Ų per subunit, accounting for 70% of the total interface area, and many of these interactions prove critical for intersubunit and interdomain communication as highlighted by alterations in deuterium uptake during the GTPase cycle (see further).

Dimer interface of CtLRR–Roc–COR.

Figure 2.
Dimer interface of CtLRR–Roc–COR.

(A) CtLRR–Roc–COR dimer interface mapped on an accessible surface representation of the A subunit of the crystal structure, with the B-subunit shown in coil representation (the LRR of the B-subunit is not shown for clarity). The major interaction surfaces are colored according to the participating structural elements, with residues from the N-COR colored cyan, the C-COR colored blue and, within the Roc domain, residues from the P-loop yellow, from the switch II green, from the Roc dimerization loop magenta and from the G4 loop orange. (B) Schematic representation of the CtLRR–Roc–COR dimer interactions. Polar interactions (salt bridges and hydrogen bonds) are indicated by green lines and non-polar interactions by brown lines. (C) Zoom in on the Roc–Roc interface emphasizing the interactions between the Roc dimerization loop (magenta) of the A subunit and the P-loop (yellow), switch II (green) and Roc dimerization loop (magenta) of the B-subunit. Polar interactions together with the interatomic distances are shown by sand-colored dotted lines.

Figure 2.
Dimer interface of CtLRR–Roc–COR.

(A) CtLRR–Roc–COR dimer interface mapped on an accessible surface representation of the A subunit of the crystal structure, with the B-subunit shown in coil representation (the LRR of the B-subunit is not shown for clarity). The major interaction surfaces are colored according to the participating structural elements, with residues from the N-COR colored cyan, the C-COR colored blue and, within the Roc domain, residues from the P-loop yellow, from the switch II green, from the Roc dimerization loop magenta and from the G4 loop orange. (B) Schematic representation of the CtLRR–Roc–COR dimer interactions. Polar interactions (salt bridges and hydrogen bonds) are indicated by green lines and non-polar interactions by brown lines. (C) Zoom in on the Roc–Roc interface emphasizing the interactions between the Roc dimerization loop (magenta) of the A subunit and the P-loop (yellow), switch II (green) and Roc dimerization loop (magenta) of the B-subunit. Polar interactions together with the interatomic distances are shown by sand-colored dotted lines.

A first important interface consists of Roc–Roc interactions (Figure 2A,B). Contacts within this interaction interface involve residues from the P-loop, the switch II region and a long loop spanning residues 541–551 between the fourth β-strand and third α-helix of the Roc domain. Considering its central localization in the dimer interface, we dubbed this region the ‘Roc dimerization loop’ (Figure 2B,C). A central residue from this dimerization loop is Arg543. Although the relatively weak electron density associated with its side chain suggests some flexibility (Supplementary Figure S2), Arg543 from subunit A forms a salt bridge with Asp515 and is within van der Waals distance to Gly517 of the switch II loop of the adjacent subunit, and it is located close to the region Gly459–Ala463 of the P-loop of subunit B. Other residues of the dimerization loop involved in Roc–Roc dimer interactions are Ser546, Asn547, Tyr550 and Trp551, which interact with the same stretch of residues of the dimerization loop of the adjacent subunit. In accordance with its central role in dimer formation, conservation analysis using the Consurf server shows that many residues of the dimerization loop are highly conserved among the Roco proteins [41]. Finally, also reciprocal interactions are made between residues from the adjacent P-loops.

A second dimerization interface involves interactions between the Roc and the COR domains of the adjacent subunits, involving the N-COR, C-COR and the linker between the latter two subdomains (Figure 2A,B). On the side of the Roc domain, these interactions are mainly formed by residues of the loop spanning the G4 motif (Asn574, Pro575, Ser576 and Asn578).

The last dimerization interface is formed by interactions between residues of the C-COR domains of both subunits, as previously described [5]. This interface contains several residues from the C-terminal β-hairpin and the preceding α-helix of both C-COR domains.

Thus, the current structure shows that the Roc domain is more important in dimer interactions than previously anticipated. The Roc domain forms extensive interactions with the Roc and COR domains of the adjacent subunit, using residues from the P-loop, switch II loop, dimerization loop and the region spanning the G4 motif. The latter regions all surround the nucleotide-binding pocket and undergo conformational changes upon nucleotide binding. Moreover, in our structure, the conformation of the P-loop, which is stabilized by dimer interactions between the P-loops of both subunits, would hamper nucleotide binding, hence requiring conformational changes upon binding of the nucleotide that, in turn, could reduce the stability of the dimer interface. These observations could be the basis for the observed shift toward the monomeric state upon nucleotide binding [25].

Switch II is ideally positioned to couple nucleotide binding to changes in domain organization

Previous SAXS and negative stain EM analyses revealed that the CtRoco protein undergoes large intra-subunit conformational changes coupled to nucleotide-induced monomerization. These analyses suggest that upon GTP (analog) binding, the CtRoco protein changes from a compact dimer to a more elongated monomeric species. Although a compact dimeric arrangement in the nucleotide-free state agrees with the current crystal structure, no structural information is available regarding the conformational changes that might take place upon nucleotide binding. Studying the CtLRR–Roc–COR structure and individual steps of the GTPase cycle by HDX-MS (see further) now allows us to look in more detail at the interdomain interactions within each subunit that are affected upon binding of a nucleotide to the Roc domain.

The LRR domain seems to interact only weakly with the Roc–COR domains, although the middle part of the LRR solenoid (ninth and 10th repeat) makes some interactions with the C-terminal part of the C-COR domain (Figure 1A). The LRR domain is connected to the Roc domain by a region of ∼40 residues, containing the aforementioned 414PLxxPPPE421 motif followed by two helices that could potentially serve as a hinge for conformational changes of the LRR with respect to the Roc domain. Moreover, the C-terminal part of the LRR and the LRR–Roc hinge region also make direct interaction with the Roc domain, mainly involving residues of the C-terminal end of the switch II region of Roc (Figure 3A). Gln527 and Phe528 of the switch II region are in close contact to Trp399 of the LRR domain, while Phe528 is also within van der Waals distance to Pro420 of the 414PLxxPPPE421 motif. Residues from this latter motif, moreover, interact with Arg532 of the switch II region, with the side chain of Glu421 forming a salt bridge and the main chain carbonyl of Leu415 forming a H-bond with the side chain of Arg532. As such these interactions between the switch II region of Roc and the hinge region linking the LRR and the Roc domain could be responsible for relaying conformational changes upon nucleotide binding to the LRR domain.

Switch II takes a central position at the domain interfaces.

Figure 3.
Switch II takes a central position at the domain interfaces.

(A) Zoom in on the interaction interface between the C-terminal part of switch II (green), the LRR domain (yellow), the hinge region connecting the LRR to the Roc domain (orange) and the linker region connecting the Roc domain with the N-COR domain (salmon). Polar interactions together with the interatomic distances are shown by sand-colored dotted lines. The LRR–Roc hinge region spans the conserved 414PLxxPPPE421 motif. Residue Arg532 links the switch II loop to the regions connecting the LRR to the Roc and the Roc to the N-COR domains. (B) Zoom in on the interactions of the N-terminal part of switch II (green) and the α3 helix of the Roc domain (gray) with the N-COR (cyan) and C-COR (blue) domains. The boxed residues His554, Tyr558 and Tyr804 align with Asn1437, Arg1441 and Tyr1699 in human LRRK2, which have been found mutated in PD.

Figure 3.
Switch II takes a central position at the domain interfaces.

(A) Zoom in on the interaction interface between the C-terminal part of switch II (green), the LRR domain (yellow), the hinge region connecting the LRR to the Roc domain (orange) and the linker region connecting the Roc domain with the N-COR domain (salmon). Polar interactions together with the interatomic distances are shown by sand-colored dotted lines. The LRR–Roc hinge region spans the conserved 414PLxxPPPE421 motif. Residue Arg532 links the switch II loop to the regions connecting the LRR to the Roc and the Roc to the N-COR domains. (B) Zoom in on the interactions of the N-terminal part of switch II (green) and the α3 helix of the Roc domain (gray) with the N-COR (cyan) and C-COR (blue) domains. The boxed residues His554, Tyr558 and Tyr804 align with Asn1437, Arg1441 and Tyr1699 in human LRRK2, which have been found mutated in PD.

The LRR domain and the hinge region connecting the LRR and Roc domains are also in close proximity to the N-COR domain as well as the linker connecting the Roc and N-COR domains, suggesting that a co-ordinated conformational change of the LRR and COR domains with respect to the Roc domain could take place. Lys384 of the LRR is involved in a salt bridge with the side chain of Asp621 of the Roc–COR linker region, and the main chain carbonyl of Asp621 forms a H-bond with Arg532 of switch II. This places Arg532 of switch II in the center of an interaction network with the Roc–COR linker (Asp621), the LRR–Roc hinge region (Leu415 and Glu421) and the LRR (Lys384) (Figure 3A). Glu416 of the 414PLxxPPPE421 motif is also within the interaction distance of the main chain carbonyl of Pro620 and of Thr626 within the Roc–N-COR linker. Finally, residues of the 414PLxxPPPE421 motif and the following residues also make direct interactions with the N-COR domain, with Pro419, Glu421, Ile422, Gln425 and Gln432 of the LRR–Roc hinge, interacting with Asp665, Thr671, Tyr675 and Asn678 of the third α-helix of N-COR.

Apart from the central position of the Gln527–Arg532 region of the switch II loop at the junction of the LRR, Roc and N-COR domains, the switch II loop also makes extensive interactions with residues that are located at the interface of the Roc, N-COR and C-COR domains, as previously described (Figure 3B) [5,6]. Phe516, Gln519, Ile521 and His526 of switch II point toward the interaction interface with the N-COR domain and interact with Asn677, Tyr706, Ser778 and Leu765 of the N-COR. This centralized position of the switch II loop at the dimer interface (Gly515 and Asp517, see before) and at the interface between the Roc, LRR and N-COR domains, make this region ideally suited to couple nucleotide binding to monomerization and interdomain conformational changes. Moreover, also switch I interacts with the N-COR domain, with Gly486 forming an H-bond with Asn710, and Leu487 being in van der Waals contact distance to Tyr706. Apart from switch I and switch II, the Roc domain also directly interacts via its α3 helix (residues Tyr550, Arg553, His554, Glu556, Lys557, Tyr558 and Gly560) with the C-COR domain (residues Asp803, Tyr804, Phe883, Thr884 and Asn885) (Figure 3B).

HDX-MS experiments map the nucleotide-induced conformational changes

We next sought to study the transitions between the nucleotide-bound and nucleotide-free (NF) forms by HDX-MS. Full-length CtRoco (1–1102) was incubated with saturating amounts of either GppNHp (as a proxy for GTP) or GDP, and the deuterium uptake profiles were compared with those of the NF protein. In this way, we could monitor the changes occurring in the nucleotide exchange and hydrolysis processes.

First, we will detail the structural changes upon binding of GppNHp by NF CtRoco. Upon binding, GppNHp the protein undergoes monomerization and further relaxation into a more extended conformation [25]. The associated changes observed with HDX-MS can be classified into four categories: (1) stabilization due to direct contacts with the GTP analog, (2) destabilization due to monomerization (loss of contacts at the dimer interface), (3) destabilization caused by relaxing into the extended conformation and (4) unexpected alterations in deuterium uptake (Supplementary Figure S3). Starting with category 1: the P-loop (G1), switch I (G2), switch II (G3) and G4 region, which encompass most of the nucleotide contact sites, were all protected upon binding GppNHp (Figure 4A,B). The G5 region, which also presumably contacts the nucleotide, is not visible in the HDX data, but the peptides surrounding G5 show mild protection upon binding GppNHp. Parts of these regions also make up the dimer interface, but the dramatic protection from deuterium exchange seen by nucleotide binding obscures the more mild enhanced exchange as one would expect from monomerization. However, the remaining two main dimerization interfaces show enhanced exchange as one would expect from the loss of direct dimer contacts. This can be seen in the Roc dimerization loop and the adjacent α3 helix (538–565) and the C-COR interface (901–940) (Figure 4A,B). The more interesting results come from categories 3 and 4. It was seen from SAXS and EM data that activation of CtRoco involves not only monomerization, but also a relaxation into a more extended conformation [25]. This can be seen in Figure 4A,C as enhanced deuterium exchange in the LRR (371–402) and the hinge region between the LRR and Roc domains (411–451) encompassing the 414PLxxPPPE421 motif. HDX-MS allows us to study the full-length protein and illustrates interdomain communication not fully appreciated by crystallography alone. One can see from the crystal structure that these regions are interfacing with each other as well as with the switch II loop of the Roc domain (Figure 3), allowing us to speculate how structural alterations in the Roc domain could affect neighboring domains as a sort of relay switch. Our HDX-MS results show that these regions in the LRR domain and the linker between the LRR and Roc domain with the 414PLxxPPPE421 motif change conformation and can act as a hinge, allowing CtRoco to shift into a more extended conformation upon monomerization. There are three distinct regions falling into category 4, showing unexpected structural alterations upon nucleotide binding. Two of these appear to play a role in transmitting structural alterations between domains. One encompasses 608–655 and 656–674 within the Roc–COR linker and N-COR regions, respectively (Figure 4). Their position at a domain interface and interacting with the LRR–Roc hinge region suggests that changes within these regions contribute to the structural communication between LRR, Roc and N-COR domains. Next is 856–883 forming a long α-helix in C-COR, which shows enhanced deuterium uptake and appears to connect the region around switch II and α3 of Roc to the C-COR–C-COR dimerization surface. The last region in category 4 is 706–730 of the N-COR domain, which shows protection upon binding GppNHp (Figure 4). This region is adjacent to the G4 loop (566–578) in the opposite subunit and might be expected to undergo enhanced exchange upon monomerization. Instead, it shows protection from exchange, suggesting that it could be generating new contact sites within the same subunit as a means of stabilizing the Roc and/or COR domains.

Nucleotide-induced conformational changes assessed by HDX-MS and mapped on the CtLRR–Roc–COR crystal structure.

Figure 4.
Nucleotide-induced conformational changes assessed by HDX-MS and mapped on the CtLRR–Roc–COR crystal structure.

(A) HDX-MS data overlaid onto the structure of nucleotide-free CtLRR–Roc–COR. The left panel shows the alterations upon binding GppNHp. The middle panel approximates the changes occurring during GTP hydrolysis by comparing GppNHp-bound protein with GDP-bound protein. The right panel shows the changes occurring upon GDP release. In each case, negative values (blue and green) are indicative of reduced deuterium uptake in the region. Positive values (red, orange and yellow) correspond to enhanced deuterium uptake in a given region. (B) Zoomed in view on the left panel (GppNHp binding) of (A) highlighting important regions within the Roc–Roc dimer interface. (C) Zoomed in view on the left panel (GppNHp binding) of (A) highlighting important regions around the LRR–Roc–N-COR interface. In panels (B) and (C), a GppNHp molecule is modeled in and shown as transparent green sticks to indicate the position of the nucleotide-binding site. Note that this position of GppNHp results in steric clashes with P-loop residues.

Figure 4.
Nucleotide-induced conformational changes assessed by HDX-MS and mapped on the CtLRR–Roc–COR crystal structure.

(A) HDX-MS data overlaid onto the structure of nucleotide-free CtLRR–Roc–COR. The left panel shows the alterations upon binding GppNHp. The middle panel approximates the changes occurring during GTP hydrolysis by comparing GppNHp-bound protein with GDP-bound protein. The right panel shows the changes occurring upon GDP release. In each case, negative values (blue and green) are indicative of reduced deuterium uptake in the region. Positive values (red, orange and yellow) correspond to enhanced deuterium uptake in a given region. (B) Zoomed in view on the left panel (GppNHp binding) of (A) highlighting important regions within the Roc–Roc dimer interface. (C) Zoomed in view on the left panel (GppNHp binding) of (A) highlighting important regions around the LRR–Roc–N-COR interface. In panels (B) and (C), a GppNHp molecule is modeled in and shown as transparent green sticks to indicate the position of the nucleotide-binding site. Note that this position of GppNHp results in steric clashes with P-loop residues.

Moving on to the next phase of the GTPase cycle, we studied the differences in deuterium uptake in GppNHp-bound and GDP-bound CtRoco reflecting changes occurring upon GTP hydrolysis. In many ways, the structure of GDP-bound CtRoco is an intermediate between that of NF and GppNHp-bound protein as much of the nucleotide-binding pocket is still occupied and there is not the full dimerization one sees in the NF protein [25]. The HDX-MS data reflect this intermediate position in both nucleotide binding and dimerization (Supplementary Figure S4). One sees the expected alterations in G1–G4, encompassing the known nucleotide interaction sites. Upon loss of interactions with the γ-phosphate, each of these regions is destabilized and shows enhanced deuterium uptake (Figure 4). Also, the shift toward dimerization is apparent as reduced deuterium uptake or protection of the C-COR dimerization site 901–940 (Figure 4). One can also see the accompanying return to the compact structure as stabilization in 371–402 and 411–432 of the LRR domain and hinge between LRR and Roc domains (Figure 4). Many of the regions involved in relaying conformational change between domains, which were discussed above in the context of GppNHp binding, show alterations in deuterium uptake when comparing GppNHp-bound and GDP-bound protein. One example is 608–632, 642–655 and 656–674 of the Roc–N-COR linker region and the third α-helix of N-COR, which together seem to make an interface connecting the LRR hinge, Roc and N-COR domains, thereby relaying conformational alterations between these domains. These interfaces, created by long-range contacts in the protein, allow communication between distant regions within the protein and are revealed by the HDX-MS data. Another site altered by GDP binding compared with GppNHp binding is 706–724 and 725–730 of the N-COR domain. This region is destabilized by the replacement of GppNHp by GDP (Figure 4) and is positioned adjacent to switch II. Since it is adjacent to the Roc domain, it could be stabilizing the Roc domain in the monomer state, but is released in the dimer state. Such a release would be coupled to enhanced deuterium uptake as is seen here (Figure 4). The last region altered by GDP binding is interestingly not altered by GppNHp binding. It is composed of 798–814 and 886–900 within the C-COR domain. This appears to make up an interface connecting the C-COR dimerization surface with switch II (G3) of the Roc domain and likely helps convey structural alterations between these regions.

Proceeding to the last stage in the GTPase cycle, release of GDP, we studied the differences in deuterium uptake between GDP-bound and NF CtRoco. All of the changes seen here are the logical extension of the aforementioned HDX-MS results one would expect from GDP release and the ensuing shift toward dimerization (Supplementary Figure S5). One sees destabilization of 453–506 and 566–592 of the Roc domain, encompassing P-loop (G1), switch I (G2) and G4, as expected from the release of GDP (Figure 4). There is no change in switch II (G3), but none was expected since it largely contacts the γ-phosphate. The dimerization loop of Roc (530–565 peptides) and the C-COR dimer interface (911–934 peptides) are stabilized, consistent with a shift towards complete dimerization. The regions of 396–402 (LRR), 411–432 and 439–452 (both hinge between LRR and Roc) all show stabilization or decreased deuterium uptake, which would be caused by the compaction of the structure correlating with dimerization (Figure 4). Peptides covering the region of 633–673, of the N-COR domain, are stabilized and, as mentioned above, appear to create an interface between the LRR domain, its hinge and the Roc domain illustrating communication between these regions (Figure 4). Lastly, 876–883 of the C-COR domain is stabilized and is apparently interfacing with 551–564 of the Roc domain further illustrating the numerous ‘relay switches’ between domains, which are critical for the dramatic conformational alterations occurring in this protein when it undergoes the multiple steps needed for the GTPase cycle.

A role of Arg543 in Roco dimerization

The current CtLRR–Roc–COR structure, in combination with the HDX data, revealed an important role of the region spanning residues 541–548 in dimerization. Considering its dedicated role in dimerization, we called this loop the ‘Roc dimerization loop’, although it should be emphasized that it is only one of the many structural regions contributing to the dimer interface. Moreover, the Roc α3-helix (residues 550–560) just C-terminal of the Roc dimerization loop interacts with the C-COR. Deuterium uptake shows that both regions undergo conformational changes upon nucleotide binding. One central residue of the Roc dimerization loop is Arg543, which makes interactions with residues of the P-loop and switch II regions of the adjacent subunit (Figure 3C). Interestingly, Arg543 was previously identified as a residue contributing to GTP hydrolysis, and mutation of this residue in the Roco proteins of C. tepidum or Methanosarceri barkeri was reported to result in a reduced GTPase activity [5,6]. Based on this observation, it has been proposed that Arg543 could act as an ‘arginine finger’ to complement the catalytic machinery of the adjacent subunit in the GTPase transition state. Our structure suggests an important role of Arg543 in dimer formation. Thus, to evaluate the contribution of the Arg543 residue to dimerization, we performed SEC-MALS experiments with the R543A mutant and the wild-type (wt) CtRoc–COR protein (amino acid 412–946) in the presence and absence of GDP or GppNHp (Figure 5). As we reported previously, the wt CtRoc–COR protein [theoretical MMmonomer (molecular mass of monomer) = 64.5 kDa] behaves as a stable dimer in the nucleotide-free (MMexp. = 115 kDa) and GDP-bound states (MMexp. = 112 kDa) at the conditions and concentrations used in the experiment. Upon the addition of GppNHp, the dimer–monomer equilibrium is shifted toward the monomeric form as observed by a larger elution volume on SEC. The MALS data indicate a gradual decrease in the molecular mass throughout the elution peak indicative of a dimer–monomer equilibrium that is mainly shifted toward the monomeric species, with an observed molecular mass of 72 kDa averaged over the elution peak (Figure 5A). The R543A mutant still behaves as a stable dimer in the nucleotide-free form (MMexp = 116 kDa), while, in the GppNHp-bound state, an MMexp averaged over the entire peak of 81 kDa is found, again indicative of mainly monomeric protein. However, in contrast with the wt GDP-bound protein, the elution profile of the GDP-bound R543A CtRoc–COR protein is broader and shifted toward the monomeric species (MMexp = 90 kDa), showing that the R543A mutation destabilizes the dimer interface and shifts the dimer–monomer equilibrium of the protein in the GDP state toward the monomeric form. This disruption of the dimer–monomer cycle, and the concomitant altered domain interactions, could subsequently contribute to the reported reduction in GTPase activity.

SEC-MALS analysis of the role of the Roc dimerization loop residue Arg543 in nucleotide-dependent CtRoc–COR homodimerization.

Figure 5.
SEC-MALS analysis of the role of the Roc dimerization loop residue Arg543 in nucleotide-dependent CtRoc–COR homodimerization.

(A) SEC-MALS analysis of the CtRoc–COR WT protein in the absence (black) and presence of GDP (green) and GppNHp (red). (B) SEC-MALS analysis of the CtRoc–COR R543A mutant protein in the absence (black) and presence of GDP (green) and GppNHp (red).

Figure 5.
SEC-MALS analysis of the role of the Roc dimerization loop residue Arg543 in nucleotide-dependent CtRoc–COR homodimerization.

(A) SEC-MALS analysis of the CtRoc–COR WT protein in the absence (black) and presence of GDP (green) and GppNHp (red). (B) SEC-MALS analysis of the CtRoc–COR R543A mutant protein in the absence (black) and presence of GDP (green) and GppNHp (red).

Discussion

In the present paper, we report the first crystal structure of the LRR–Roc–COR module that is central to the action of all group 1 and group 2 Roco proteins, including the PD-associated LRRK2 [1,7]. This structure, in combination with HDX-MS results highlighting structural alterations induced by individual steps of the GTPase cycle, gives us much greater insights into the regulation of this class of proteins.

We recently showed that the CtRoco protein undergoes a dimer–monomer cycle during the GTPase reaction, with the protein being mainly dimeric in the nucleotide-free state and monomeric in the GTP-bound state while an intermediate situation occurs in the GDP-bound state [25]. The current structure of CtLRR–Roc–COR in the nucleotide-free state shows a compact homodimer in agreement with our previous EM and SAXS data [25]. In contrast with the previously reported structure of the CtRoc–COR construct [5], our structure reveals both Roc domains within the homodimer, demonstrating unanticipated major contributions of Roc–Roc and Roc–COR interactions to the total dimer interface (Figure 2). One of the most extensive contributions to the Roc–Roc dimerization interface is provided by a loop (residues 541–551) preceding the α3 helix of Roc, which we therefore named the ‘Roc dimerization loop’. One residue of this loop, Arg543, was initially proposed to be a catalytic arginine finger triggering GTP hydrolysis upon dimerization because an R543A mutant shows decreased GTPase activity [5,6]. However, we show here that the R543A mutation shifts the equilibrium of GDP-bound protein to the monomeric form compared with the GDP-bound WT protein. This offers an alternative explanation for the reduced activity of R543A within the framework of our model that Roco proteins alternate between a monomeric and dimeric state during the GTPase cycle.

Intriguingly, the CtLRR–Roc–COR structure shows the P-loop in an ‘occluded’ conformation which is not compatible with nucleotide binding (Figure 1C). A similar conformation of the P-loop was not observed in the structure of the monomeric M. barkeri Roc–CORΔC (lacking the C-COR domain) protein, nor in the CtRoc–COR structure where only one of the two Roc domains was observed [5,6]. This suggests that the occluded P-loop conformation is stabilized by the Roc–Roc interface. These observations provide a potential mechanistic trigger for the nucleotide-induced Roco monomerization. Within this scenario, binding of GTP in the nucleotide-binding site would require a conformational change of the P-loop and the adjacent switch II loop and Roc dimerization loop that together form the Roc–Roc dimer interface (Figures 1C and 2C), thus disrupting this interface and promoting monomerization. Moreover, the switch II loop is ideally positioned to relay these conformational changes to other parts of the protein, as testified by our HDX-MS results, ultimately coupling protein monomerization to large-scale conformational changes within each subunit. Indeed, the switch II loop (513–532), together with the Roc α3 helix (547–559), interacts with the N-COR and C-COR domains of the same subunit (Figure 3B). Additionally, the C-terminal part of the switch II forms an interaction network with residues of the conserved 414PLxxPPPE421 hinge motif that links the LRR domain to the Roc domain, as well as with the peptide segment linking the Roc domain to the N-COR domain (Figure 3A). One can easily envision how GTP binding to the nucleotide-binding site could transmit information via switch II to the LRR, N-COR and C-COR domains, generating dramatic structural alterations such as monomerization and elongation of the monomeric structure. Our HDX-MS experiments confirm the conformational changes in these regions and thus propose a pathway for relaying conformational changes from the nucleotide-binding pocket to the LRR and N-COR and ultimately to the C-COR–C-COR dimer interface (Figure 4).

Mapping of LRRK2 pathogenic and risk variant PD mutations onto the CtLRR–Roc–COR structure reveals that several of these mutations cluster together in regions that are identified as being important for conformational information transfer and are affected by nucleotide binding as assessed by HDX-MS (Supplementary Figure S6). His554, Tyr558 and Gly559 of the α3 helix of the Roc domain of CtRoco align with Asn1437, Arg1441 and Ala1442 of LRRK2, respectively, while Ser711, Thr721, Tyr804 and Val888 of the CtRoco COR domains align with Val1613, Arg1628, Tyr1699 and Leu1795 of LRRK2. LRRK2 R1441C/G/H and Y1699C are the most common PD mutations within the Roc–COR domains, while the N1437H, A1442P, V1613A, R1628P and L1795F mutations have been associated with PD [42]. In our CtLRR–Roc–COR structure, His554, Tyr558, Gly559 and Tyr804 cluster together at the interface of the C-COR and the α3 helix and switch II regions of the Roc domain, and all but Tyr804 display altered deuterium uptake upon binding GppNHp (Figure 4). The previously reported LRRK2 R1398H protective mutant that shows increased GTPase activity aligns with CtRoco Gln519 [43]. Interestingly, Gln519 is located within the Switch II region, relatively close to the Roc α3 helix of the same subunit and to the dimer interface (Figure 3B). The Gln519 side chain points toward Ser778 of the linker connecting the N-COR to the C-COR domain. Also, Glu580 locates close to the CtRoco dimer interface at the C-terminal end of the G4 region, and aligns with Lys1468 in LRRK2, of which the K1468E mutation has been identified as a potentially pathogenic LRRK2 variant. Val489 is the homolog of Ile1371 in LRRK2, and a I1371V mutation has also been associated with PD. Val489 is located within the switch I region at the interface with the N-COR domain and switch II, and stacks to His699 of N-COR and Trp514 of switch II. There is also one possibly pathogenic LRRK2 mutation (R1325Q) that locates in the hinge region connecting the LRR and Roc domains, which appears to flex causing the elongation associated with monomerization, where Arg1325 corresponds to Lys443 of CtRoco. Finally, many other potential pathogenic mutations are found in the LRR domain. Two of these residues Arg1067 and Ser1096 seem to align with, respectively, Gly160 and Thr186 in CtRoco, which are located in a region of the LRR that is close to the C-terminal extreme of the C-COR involved in dimer formation. Together, this indicates that the regions that we find important here are also key to the function of LRRK2 and suggests that the overall mechanisms of Roc–COR signaling are conserved in all Roco proteins, although the details of the signal transduction as well as the final physiological outcomes could differ. Further research is required to establish whether or not the PD mutations in LRRK2 affect its nucleotide-induced conformational cycle.

Our HDX-MS experiments also revealed that distinct regions within the full-length CtRoco protein exist in multiple conformations at various times within the GTPase cycle (Supplementary Figure S7). Two of these regions involve the P-loop and switch I (residues 447–488) and the G4 loop (residues 566–588), while the last involves the 911–933 peptide within the C-COR domain, which is a critical dimerization surface. As mentioned previously, the P-loop has the ability to occlude the nucleotide-binding site (shown in Figure 1C). Given the importance of these regions in regulating the CtRoco protein, it is tempting to hypothesize that these multiple conformations play a functional role and that the conformational isomerism displayed in these regions could be potentially critical for controlling nucleotide binding and/or release as well as the concomitant monomerization/dimerization cycle. While conformational isomerism is often found in N-mers containing >4 subunits, it has been described in dimeric and tetrameric proteins [44,45]. This conformational isomerism creates an ensemble of slowly interconverting structures, some of which are more or less resistant to nucleotide binding and/or to monomerization. Such behavior was called ‘deterministic’ [45], to differentiate it from the stochastic behavior exhibited by most proteins. While there is a precedent for deterministic behavior, it is still a relatively rare phenomenon. Proteins exhibiting deterministic behavior exist at slightly different points along a given pathway or attain an ensemble of structures, rather than existing in one single point in equilibrium, or single structure. In our case, one potential scenario could be that the observed conformational isomerism might assist in releasing GDP after GTP hydrolysis. Most GTPases require guanine nucleotide exchange factors (GEFs) to destabilize the nucleotide-binding site and release GDP [46,47]. Roco family proteins might mimic this process in the absence of external GEFs by creating a nucleotide-binding site able to achieve multiple conformations, assisting the GDP release process. The consequence and potential drawback of having a nucleotide-binding site capable of conformational isomerism would be lowered nucleotide affinities, as is observed for the Roco proteins [48]. However, the benefit is that conformational isomerism could allow Roco proteins to cycle without the requirement for GEFs. We therefore could envision a hypothetical scenario in which dimerization of the nucleotide-free protein may slow the binding of GTP. The ensembles which have lost GDP and bind GTP will become activated and shift toward a monomeric structure. After the γ-phosphate is hydrolyzed, the protein shifts toward dimerization and the conformational isomerism destabilizes the nucleotide-binding site enough that GDP is lost. This cycles the protein back to its nucleotide-free form and shifts the equilibrium further toward dimerization where the combination of high cellular GTP concentrations and flexibility in the nucleotide-binding site allow GTP to bind and the cycle to repeat.

Data Deposition

The atomic co-ordinates and structure factors have been deposited in the Protein Data Bank (PDB) with PDB: ID code 6HLU.

Abbreviations

     
  • COR

    C-terminal of Roc

  •  
  • CtRoco

    Roco protein from Chlorobium tepidum

  •  
  • EM

    electron microscopy

  •  
  • GDP

    guanosine-5′-diphosphate

  •  
  • GEFs

    guanine nucleotide exchange factors

  •  
  • GppNHp

    guanosine-5′-[(β,γ)-imido]triphosphate

  •  
  • GTP

    guanosine-5′-triphosphate

  •  
  • HDX-MS

    hydrogen–deuterium exchange coupled to mass spectrometry

  •  
  • LRR

    leucine-rich repeat

  •  
  • LRRK2

    leucine-rich repeat kinase 2

  •  
  • MALS

    multi-angle light scattering

  •  
  • MM

    molecular mass

  •  
  • NCE

    normalized collision energy

  •  
  • NF

    nucleotide-free

  •  
  • PD

    Parkinson's disease

  •  
  • Roc

    Ras of complex protein

  •  
  • SAXS

    small-angle X-ray scattering

  •  
  • SEC

    size-exclusion chromatography

Author Contribution

E.D., J.L. and W.V. designed the experiments. E.D. solved the crystal structure. E.D. and W.V. performed the SEC-SAXS analysis. E.D. and M.L. purified the proteins. R.K.S., E.D. and R.G. performed the SEC-MALS analysis. J.L. performed the HDX-MS experiments. E.D., J.S., A.K., J.L. and W.V. interpreted the data. J.L. and W.V. wrote the manuscript. All the authors read and edited the manuscript.

Funding

This work was supported by the Fonds voor Wetenschappelijk Onderzoek (M.L. and W.V.), a Strategic Research Program Financing of the VUB (W.V.), The Michael J. Fox Foundation for Parkinson's Research (R.K.S., A.K. and W.V.), the Hercules Foundation (W.V.), BioStruct-X by the European Community's Seventh Framework Programme (W.V.).

Acknowledgements

We thank the staff at the beamline Proxima 2 of Soleil (France) and BM29 at ESRF (France) for assistance during data collection and the mass spectrometry facility at MPI-CBG for their support.

Competing Interests

The Authors declare that there are no competing interests associated with the manuscript.

References

References
1
Bosgraaf
,
L.
and
Van Haastert
,
P.J.M.
(
2003
)
Roc, a Ras/GTPase domain in complex proteins
.
Biochim. Biophys. Acta, Mol. Cell Res.
1643
,
5
10
2
Marín
,
I.
(
2006
)
The Parkinson disease gene LRRK2: evolutionary and structural insights
.
Mol. Biol. Evol.
23
,
2423
2433
3
Marín
,
I.
(
2008
)
Ancient origin of the Parkinson disease gene LRRK2
.
J. Mol. Evol.
67
,
41
50
4
Marín
,
I.
,
van Egmond
,
W.N.
and
van Haastert
,
P.J.M.
(
2008
)
The Roco protein family: a functional perspective
.
FASEB J.
22
,
3103
3110
5
Gotthardt
,
K.
,
Weyand
,
M.
,
Kortholt
,
A.
,
Van Haastert
,
P.J.M.
and
Wittinghofer
,
A.
(
2008
)
Structure of the Roc-COR domain tandem of C. tepidum, a prokaryotic homologue of the human LRRK2 Parkinson kinase
.
EMBO J.
27
,
2239
2249
6
Terheyden
,
S.
,
Ho
,
F.Y.
,
Gilsbach
,
B.K.
,
Wittinghofer
,
A.
and
Kortholt
,
A.
(
2015
)
Revisiting the Roco G protein cycle
.
Biochem. J.
465
,
139
147
7
Gilsbach
,
B.K.
and
Kortholt
,
A.
(
2014
)
Structural biology of the LRRK2 GTPase and kinase domains: implications for regulation
.
Front. Mol. Neurosci.
7
,
32
8
Dihanich
,
S.
(
2012
)
MASL1: a neglected ROCO protein
.
Biochem. Soc. Trans.
40
,
1090
1094
9
Dihanich
,
S.
,
Civiero
,
L.
,
Manzoni
,
C.
,
Mamais
,
A.
,
Bandopadhyay
,
R.
,
Greggio
,
E.
et al.  (
2014
)
GTP binding controls complex formation by the human ROCO protein MASL1
.
FEBS J.
281
,
261
274
10
Civiero
,
L.
,
Dihanich
,
S.
,
Lewis
,
P.A.
and
Greggio
,
E.
(
2014
)
Genetic, structural, and molecular insights into the function of RAS of complex proteins domains
.
Chem. Biol.
21
,
809
818
11
Bialik
,
S.
and
Kimchi
,
A.
(
2006
)
The death-associated protein kinases: structure, function, and beyond
.
Annu. Rev. Biochem.
75
,
189
210
12
Carlessi
,
R.
,
Levin-Salomon
,
V.
,
Ciprut
,
S.
,
Bialik
,
S.
,
Berissi
,
H.
,
Albeck
,
S.
et al.  (
2011
)
GTP binding to the ROC domain of DAP-kinase regulates its function through intramolecular signalling
.
EMBO Rep.
12
,
917
923
13
Jebelli
,
J.D.
,
Dihanich
,
S.
,
Civiero
,
L.
,
Manzoni
,
C.
,
Greggio
,
E.
and
Lewis
,
P.A.
(
2012
)
GTP binding and intramolecular regulation by the ROC domain of death associated protein kinase 1
.
Sci. Rep.
2
,
695
14
Zimprich
,
A.
,
Biskup
,
S.
,
Leitner
,
P.
,
Lichtner
,
P.
,
Farrer
,
M.
,
Lincoln
,
S.
et al.  (
2004
)
Mutations in LRRK2 cause autosomal-dominant parkinsonism with pleomorphic pathology
.
Neuron
44
,
601
607
15
Paisán-Ruíz
,
C.
,
Jain
,
S.
,
Evans
,
E.W.
,
Gilks
,
W.P.
,
Simón
,
J.
,
Van Der Brug
,
M.
et al.  (
2004
)
Cloning of the gene containing mutations that cause PARK8-linked Parkinson's disease
.
Neuron
44
,
595
600
16
Lewis
,
P.A.
,
Greggio
,
E.
,
Beilina
,
A.
,
Jain
,
S.
,
Baker
,
A.
and
Cookson
,
M.R.
(
2007
)
The R1441C mutation of LRRK2 disrupts GTP hydrolysis
.
Biochem. Biophys. Res. Commun.
357
,
668
671
17
West
,
A.B.
,
Moore
,
D.J.
,
Choi
,
C.
,
Andrabi
,
S.A.
,
Li
,
X.
,
Dikeman
,
D.
et al.  (
2007
)
Parkinson's disease-associated mutations in LRRK2 link enhanced GTP-binding and kinase activities to neuronal toxicity
.
Hum. Mol. Genet.
16
,
223
232
18
Cookson
,
M.R.
(
2010
)
The role of leucine-rich repeat kinase 2 (LRRK2) in Parkinson's disease
.
Nat. Rev. Neurosci.
11
,
791
797
19
Liao
,
J.
,
Wu
,
C.-X.
,
Burlak
,
C.
,
Zhang
,
S.
,
Sahm
,
H.
,
Wang
,
M.
et al.  (
2014
)
Parkinson disease-associated mutation R1441H in LRRK2 prolongs the ‘active state’ of its GTPase domain
.
Proc. Natl Acad. Sci. U.S.A.
111
,
4055
4060
20
West
,
A.B.
,
Moore
,
D.J.
,
Biskup
,
S.
,
Bugayenko
,
A.
,
Smith
,
W.W.
,
Ross
,
C.A.
et al.  (
2005
)
Parkinson's disease-associated mutations in leucine-rich repeat kinase 2 augment kinase activity
.
Proc. Natl Acad. Sci. U.S.A.
102
,
16842
16847
21
Greggio
,
E.
,
Jain
,
S.
,
Kingsbury
,
A.
,
Bandopadhyay
,
R.
,
Lewis
,
P.
,
Kaganovich
,
A.
et al.  (
2006
)
Kinase activity is required for the toxic effects of mutant LRRK2/dardarin
.
Neurobiol. Dis.
23
,
329
341
22
Gloeckner
,
C.J.
,
Kinkl
,
N.
,
Schumacher
,
A.
,
Braun
,
R.J.
,
O'Neill
,
E.
,
Meitinger
,
T.
et al.  (
2006
)
The Parkinson disease causing LRRK2 mutation I2020T is associated with increased kinase activity
.
Hum. Mol. Genet.
15
,
223
232
23
Deng
,
J.
,
Lewis
,
P.A.
,
Greggio
,
E.
,
Sluch
,
E.
,
Beilina
,
A.
and
Cookson
,
M.R.
(
2008
)
Structure of the ROC domain from the Parkinson's disease-associated leucine-rich repeat kinase 2 reveals a dimeric GTPase
.
Proc. Natl Acad. Sci. U.S.A.
105
,
1499
1504
24
Guaitoli
,
G.
,
Raimondi
,
F.
,
Gilsbach
,
B.K.
,
Gómez-Llorente
,
Y.
,
Deyaert
,
E.
,
Renzi
,
F.
et al.  (
2016
)
Structural model of the dimeric Parkinson's protein LRRK2 reveals a compact architecture involving distant interdomain contacts
.
Proc. Natl Acad. Sci. U.S.A.
113
,
E4357
E4366
25
Deyaert
,
E.
,
Wauters
,
L.
,
Guaitoli
,
G.
,
Konijnenberg
,
A.
,
Leemans
,
M.
,
Terheyden
,
S.
et al.  (
2017
)
A homologue of the Parkinson's disease-associated protein LRRK2 undergoes a monomer-dimer transition during GTP turnover
.
Nat. Commun.
8
,
1008
26
Deyaert
,
E.
,
Kortholt
,
A.
and
Versées
,
W.
(
2017
)
The LRR-Roc-COR module of the Chlorobium tepidum Roco protein: crystallization and X-ray crystallographic analysis
.
Acta Crystallogr. F Struct. Biol. Commun.
73
,
520
524
27
Kabsch
,
W.
(
2010
)
Xds
.
Acta Crystallogr. D Biol. Crystallogr.
66
,
125
132
28
Evans
,
P.R.
and
Murshudov
,
G.N.
(
2013
)
How good are my data and what is the resolution?
Acta Crystallogr. D Biol. Crystallogr.
69
,
1204
1214
29
McCoy
,
A.J.
,
Grosse-Kunstleve
,
R.W.
,
Adams
,
P.D.
,
Winn
,
M.D.
,
Storoni
,
L.C.
and
Read
,
R.J.
(
2007
)
Phaser crystallographic software
.
J. Appl. Crystallogr.
40
,
658
674
30
Afonine
,
P.V.
,
Grosse-Kunstleve
,
R.W.
,
Echols
,
N.
,
Headd
,
J.J.
,
Moriarty
,
N.W.
,
Mustyakimov
,
M.
et al.  (
2012
)
Towards automated crystallographic structure refinement with phenix.refine
.
Acta Crystallogr. D Biol. Crystallogr.
68
,
352
367
31
Shkumatov
,
A.V.
and
Strelkov
,
S.V.
(
2015
)
DATASW, a tool for HPLC-SAXS data analysis
.
Acta Crystallogr. D Biol. Crystallogr.
71
,
1347
1350
32
Svergun
,
D.
,
Barberato
,
C.
and
Koch
,
M.H.
(
1995
)
CRYSOL — a program to evaluate X-ray solution scattering of biological macromolecules from atomic coordinates
.
J. Appl. Crystallogr.
28
,
768
773
33
He
,
W.
,
Bai
,
G.
,
Zhou
,
H.
,
Wei
,
N.
,
White
,
N.M.
,
Lauer
,
J.
et al.  (
2015
)
CMT2D neuropathy is linked to the neomorphic binding activity of glycyl-tRNA synthetase
.
Nature
526
,
710
714
34
Mayne
,
L.
,
Kan
,
Z.Y.
,
Sevugan Chetty
,
P.
,
Ricciuti
,
A.
,
Walters
,
B.T.
and
Englander
,
S.W.
(
2011
)
Many overlapping peptides for protein hydrogen exchange experiments by the fragment separation-mass spectrometry method
.
J. Am. Soc. Mass Spectrom.
22
,
1898
1905
35
Walters
,
B.T.
,
Ricciuti
,
A.
,
Mayne
,
L.
and
Englander
,
S.W.
(
2012
)
Minimizing back exchange in the hydrogen exchange-mass spectrometry experiment
.
J. Am. Soc. Mass Spectrom.
23
,
2132
2139
36
Pascal
,
B.D.
,
Willis
,
S.
,
Lauer
,
J.L.
,
Landgraf
,
R.R.
,
West
,
G.M.
,
Marciano
,
D.
et al.  (
2012
)
HDX workbench: software for the analysis of H/D exchange MS data
.
J. Am. Soc. Mass. Spectrom.
23
,
1512
1521
37
Wittinghofer
,
A.
and
Vetter
,
I.R.
(
2011
)
Structure-function relationships of the G domain, a canonical switch motif
.
Annu. Rev. Biochem.
80
,
943
971
38
Verstraeten
,
N.
,
Fauvart
,
M.
,
Versees
,
W.
and
Michiels
,
J.
(
2011
)
The universally conserved prokaryotic GTPases
.
Microbiol. Mol. Biol. Rev.
75
,
507
542
39
Pai
,
E.F.
,
Kabsch
,
W.
,
Krengel
,
U.
,
Holmes
,
K.C.
,
John
,
J.
and
Wittinghofer
,
A.
(
1989
)
Structure of the guanine-nucleotide-binding domain of the Ha-ras oncogene product p21 in the triphosphate conformation
.
Nature
341
,
209
214
40
Krissinel
,
E.
and
Henrick
,
K.
(
2007
)
Inference of macromolecular assemblies from crystalline state
.
J. Mol. Biol.
372
,
774
797
41
Ashkenazy
,
H.
,
Abadi
,
S.
,
Martz
,
E.
,
Chay
,
O.
,
Mayrose
,
I.
,
Pupko
,
T.
et al.  (
2016
)
Consurf 2016: an improved methodology to estimate and visualize evolutionary conservation in macromolecules
.
Nucleic Acids Res.
44
,
W344
W350
42
Monfrini
,
E.
and
Di Fonzo
,
A.
(
2017
)
Leucine-rich repeat kinase (LRRK2) genetics and Parkinson's disease
.
Adv. Neurobiol.
14
,
3
30
43
Nixon-Abell
,
J.
,
Berwick
,
D.C.
,
Grannó
,
S.
,
Spain
,
V.A.
,
Blackstone
,
C.
and
Harvey
,
K.
(
2016
)
Protective LRRK2 R1398H variant enhances GTPase and Wnt signaling activity
.
Front. Mol. Neurosci.
9
,
18
44
Botelho
,
M.G.
,
Rietveld
,
A.W.M.
and
Ferreira
,
S.T.
(
2006
)
Long-lived conformational isomerism of protein dimers: the role of the free energy of subunit association
.
Biophys. J.
91
,
2826
2832
45
Erijman
,
L.
and
Weber
,
G.
(
1991
)
Oligomeric protein associations: transition from stochastic to deterministic equilibrium
.
Biochemistry
30
,
1595
1599
46
Cherfils
,
J.
and
Zeghouf
,
M.
(
2013
)
Regulation of small GTPases by GEFs, GAPs, and GDIs
.
Physiol. Rev.
93
,
269
309
47
Bos
,
J.
,
Rehmann
,
H.
and
Wittinghofer
,
A.
(
2007
)
pGEFs and GAPs : critical elements in the control of small G proteins
.
Cell
129
,
865
877
48
Wauters
,
L.
,
Terheyden
,
S.
,
Gilsbach
,
B.K.
,
Leemans
,
M.
,
Athanasopoulos
,
P.S.
,
Guaitoli
,
G.
et al.  (
2018
)
Biochemical and kinetic properties of the complex Roco G-protein cycle
.
Biol. Chem.
399
,
1447
1456

Author notes

*

Current address: Astbury Centre for Structural Molecular Biology, School of Molecular and Cellular Biology, Faculty of Biological Sciences, University of Leeds, Leeds LS2 9JT, U.K.

Supplementary data