Global dynamic portrait of LRRK2_RCKW is revealed by HDX-MS and GaMD simulations

To identify the solvent exposed regions of LRRK2_RCKW, we mapped the HDX-MS data onto the cryo-EM structure of LRRK2_RCKW (PDB: 6VNO) [10] (Fig 1A). Our overall HDX-MS coverage of LRRK2_RCKW, which was >98% (S1 Fig), shows the relative fractional deuterium uptake of each peptide. We previously mapped the HDX-MS profile onto a model of the kinase domain [12] while here we mapped the HDX-MS exchange pattern onto the entire LRRK2_RCKW cryo-EM structure. This allows us to capture the interactions between and within the C-terminal domains. As shown in S2A Fig, the C-lobe of the kinase domain around the activation loop (A-loop), a part of the surface region of the ROC domain, and parts of the COR-B domain show the highest deuterium uptake. These regions are either highly flexible or likely to be unfolded, and some are not resolved in the static cryo-EM structure [10]. In contrast, the core of the ROC domain and the core of the COR-B domain show low deuterium uptake suggesting that they are well folded and form rigid domains (S2B Fig). The WD40 domain, except for several loops, also has less deuterium uptake, which indicates that its core is less dynamic in solution.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. The overall dynamic of LRRK2_RCKW.

(a) The relative deuterium uptake after 2-min deuterium exposure is color-coded mapped on the LRRK2_RCKW model. Gray color indicates no deuterium uptake information. The surface of the kinase domain is shown in gray. The highly protected COR-B Helix and the Ct-Helix located at the back of the kinase domain are labeled. (b) Snapshot of LRRK2_RCKW in the MD simulation. The surface of each domain is shown in different colors. Left represents the compact architecture of LRRK2_RCKW. The middle is the structure from cryo-EM (PDB: 6VNO) [10]. The right represents one of the extended states of LRRK2_RCKW. cryo-EM, cryogenic electron microscopy; Ct-Helix, C-terminal helix; MD, molecular dynamics; ROC, ras-of-complex.

https://doi.org/10.1371/journal.pbio.3001427.g001

There are also regions on the surface of each domain that are highly protected from solvent, suggesting that these are domain–domain interfaces. The N-lobe of the kinase domain, for example, is relatively well shielded from solvent in contrast to the highly exposed C-lobe (S2B Fig). The 2 helices (residues 1,567 to 1,578 and 1,600 to 1,612, respectively) of the COR-A domain at the interface contacting the ROC domain are mostly shielded, indicating that the interaction between COR-A and the ROC domain is stable and persistent. One of the ROC domain helices (residues 1,425 to 1,441) and the adjacent loops of the COR-B domain all have low deuterium uptake, indicating that the ROC:COR-B interface is also shielded from solvent in this inactive conformation. Parts of the kinase domain N-lobe with low deuterium uptake are mostly shielded by the COR-B-Kinase linker and the COR-B domain. Another interface that is well shielded from solvent lies between the kinase and WD40 domain. The beginning of the WD40 domain, the N-terminal end of the αE helix in the kinase domain and the N-terminus of the Ct-Helix, which extends from the WD40 domain, all have very low deuterium uptake, suggesting that the WD40 domain interacts persistently with the C-lobe of the kinase in solution (S2B Fig). This could be one of the reasons why a stable and active isolated kinase domain of LRRK2 has not been expressed yet.

To investigate the dynamic features of LRRK2_RCKW and the interactions within its domains, we performed GaMD simulations to recapitulate the behavior of LRRK2_RCKW in solution. To capture a more accurate representative model of LRRK2_RCKW breathing dynamics, we applied enhanced sampling to broadly sample the conformational changes that take place during the simulations. Both extended and compact conformations of LRRK2_RCKW are captured by the simulations (Figs 1B and S3), and the kinase domain is at the center of the breathing dynamics of LRRK2_RCKW. The COR-B domain persistently interacts with the N-lobe of the kinase during the simulation, while the WD40 domain interacts stably with the C-lobe of the kinase domain. The COR-A and ROC domains move as a single rigid body and fluctuate between far and near states relative to the C-lobe of the kinase domain. When the kinase domain is in a closed conformation, the ROC domain and the COR-A domain are brought closer to the C-lobe of the kinase domain (Fig 1B, left), while in an open conformation (Fig 1B, right), the COR-A domain and ROC domain move further away from the C-lobe of the kinase domain, thereby creating a more extended conformation. The dynamic features that bring the ROC domain and the C-lobe of the kinase domain into close proximity correlate with the intramolecular interactions between the kinase and GTPase domains. As described below, our GaMD simulations also revealed many specific interactions in this space that could potentially be involved in mediating allosteric conformational changes, and these interactions might also be influenced by PD mutations.

Domain interfaces with the kinase domain

To better understand how the kinase domain is shielded from solvent, we focused on 2 dominant helices that embrace the N- and C-lobes of the kinase domain (S4 Fig). One lies in the COR-B domain (residues 1,771 to 1,791) and is buttressed up against the αC helix in the N-lobe, while the other helix (residues 2,500 to 2,527) lies at the C-terminus and is anchored mainly to the C-lobe of the kinase domain. We refer to these as the COR-B Helix and the Ct-Helix, respectively. In contrast to the well-shielded N-lobe, much of the C-lobe is disordered in the cryo-EM LRRK2_RCKW structure (PDB: 6VNO) [10], which represents an inactive conformation. In the kinase domain, we focus on the extended AS, which is typically well ordered in active kinases and poised to interact with substrates and inhibitors [13]. In the LRRK2_RCKW structure, the disordered A-loop at the beginning of the AS faces the COR-B Helix, while the region at the end of the AS faces ROC and COR-A (S4 Fig).

Capturing interactions between the kinase and COR-B domains

The COR-B domain plays a critical role in coordinating the communication that takes place between the kinase domain and the ROC domain. Based on our HDX-MS results (Fig 2 and S1 Table), the interface between the N-lobe of the kinase domain and the COR-B domain is mostly shielded from solvent. This surface is dominated by a long amphipathic helix, the COR-B Helix, and the αC helix of the kinase domain (Fig 2). The COR-B Helix can be divided into 3 segments based on HDX-MS. While the middle region (residues 1,777 to 1,788) shows almost no uptake, the N-terminus shows a slow deuterium exchange. In contrast, the C-terminus of the COR-B Helix is more exposed to solvent but also close to both the disordered A-loop of the kinase domain and to the ROC domain. It has approximately 50% deuterium uptake at 2 min, which is reduced to 35% when the type I LRRK2 kinase inhibitor MLi-2 is bound. The N-terminus of the αC helix in the kinase domain that binds to the COR-B Helix shows an unusual uptake pattern. The uptake increases at a linear rate signifying a slow exchange without reaching a plateau within 2 min, while binding of MLi-2 significantly protects against uptake. Together, HDX-MS shows that the COR-B:kinase domain interface is mostly shielded from solvent and that the COR-B Helix, in particular, is stably anchored to the N-lobe of the kinase while the N-terminus of the COR-B Helix is dynamic and interacts with both the kinase and ROC domains. To dissect these interactions more rigorously, we carefully analyzed the 3 segments of the COR-B Helix that face the αC helix in the kinase domain, the disordered AS and the Ct-Helix (Fig 3A).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Deuterium uptake around the COR-B Helix.

(a) The deuterium uptake of selected peptides is plotted and mapped on the LRRK2_RCKW structure (PDB: 6VNO). The charts are color coded to the corresponding regions shown. The middle of the COR-B Helix has almost no deuterium uptake, suggesting that it is shielded from the solvent. Other peptides that are located at the surface and the A-loop all demonstrate high deuterium uptake. Binding of MLi-2 reduces the deuterium uptake of COR-B-Kinase linker, the αC helix, and the A-loop. The uptake is also reduced in peptides that are located around the N-terminal or C-terminal ends of the COR-B Helix. Peptide 1,426–1,449 in the ROC domain is the only peptide that its uptake increases when binding to MLi-2. (b) The relative deuterium exchange for each peptide detected from the N-terminus to the C-terminus of LRRK2_RKCW in apo state the kinase (black) and MLi-2 bound (red) conditions at 2 min. The arrows indicate the peptides shown in (a). The data underlying the graphs shown can be found in S1 Data. A-loop, activation loop; ROC, ras-of-complex.

https://doi.org/10.1371/journal.pbio.3001427.g002

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Characterization of COR-B Helix.

(a) The COR-B Helix, the αC helix and Ct-Helix, and the ROC domain that near the C-terminal end of the COR-B Helix are colored based on the relative fractional uptake (PDB: 6VNO). (b) The surface electrostatic potential of the COR-B Helix. The positively charged N-terminal end of the COR-B Helix is interacting with the Ct-Helix, while the C-terminal end is negatively charged and interacts with the ROC domain. (c) All hydrophobic residues of the COR-B Helix are located on the same side and buried in the COR-B domain, while the charged residues that are forming multiple salt bridges with the αC helix are located on the other side. A-loop, activation loop; Ct-Helix, C-terminal helix; ROC, ras-of-complex.

https://doi.org/10.1371/journal.pbio.3001427.g003

The COR-B Helix has charged residues at both its N- and C-termini while the middle part of the COR-B Helix is amphipathic (Fig 3B). Multiple hydrophobic residues face the core of the COR-B domain making this a very stable interface (Fig 3C). On the opposite side, multiple charged residues interact strongly with the αC helix of the kinase domain through electrostatic interactions, locking the αC helix into an inactive conformation (Fig 3C) [14]. Typically, in an active kinase, the basic residue that lies at the beginning of the αC helix (R1915 in LRRK2) interacts with a phosphate on the A-loop, which is missing in this structure [15,16]. Three phosphorylation sites on the A-loop, pT2031, pS2032, and pT2035, have been reported and discussed by others, but how these phosphorylation sites regulate kinase activity is still unclear [17]. Our hypothesis is that the interactions involving the N-terminus of the αC helix will be altered depending on whether the kinase domain is in an active or inactive conformation [14].

The N-terminus of the COR-B Helix interacts with the COR-B-Kinase linker that is more solvent exposed and wraps around the N-lobe of the kinase domain (Fig 3A). This region serves as a “cap” for the N-lobe of the kinase domain, indicating that the conformation and flexibility in particular of the N-lobe appears to be tightly controlled by the linker. The N-terminus of the COR-B Helix, which contains 2 basic residues (R1771 and K1772), also approaches the end of the Ct-Helix that follows the WD40 domain (Fig 4A) and contains the last few residues of LRRK2 that are thought to be crucial for kinase activity. It was previously shown, for example, that deletion of C-terminal residues reduces the kinase activity of LRRK2_RCKW [18–20].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Capturing interactions at the N-and C-termini of the COR-B Helix.

(a) The N-terminal end of the COR-B Helix is in close proximity to the C-terminal residues of the Ct-Helix, which are undiscernible in the cryo-EM structure (PDB: 6VNO). (b) GaMD simulations capture the interactions between the COR-B Helix with the COR-B-Kinase linker and with the C-terminal end of the Ct-Helix. (c) In LRRK2_RCKW, the COR-B Helix is stably anchored to the αC helix of the kinase N-lobe. Its N-terminal interacts with the Ct-Helix and the N-lobe of the kinase, while its C-terminal is tethered to the ROC domain and is in close proximity to the A-loop. (d) The C-terminus of the COR-B Helix is anchored to the side chain of R1441 in the ROC domain through W1791. (e) The interaction of E1780 and W1791 with the pathogenic mutation sites R1441 and N1437 on the ROC domain could be captured using GaMD simulations. A-loop, activation loop; cryo-EM, cryogenic electron microscopy; Ct-Helix, C-terminal helix; GaMD, Gaussian accelerated molecular dynamics; MD, molecular dynamics; ROC, ras-of-complex.

https://doi.org/10.1371/journal.pbio.3001427.g004

In the LRRK2_RCKW cryo-EM structure (PDB: 6VNO), the terminal 3 residues (residues 2,525 to 2,527) are disordered [10], while these terminal residues form another turn of the Ct helix in the full-length LRRK2 structure (PDB:7LHW) and in the AlphaFold model [21]. In addition, there is a highly flexible loop (residues 1,715 to 1,730) in the COR-B domain that lies close to the N-lobe of the kinase domain (Fig 5A). This loop is solvent exposed and also disordered in the LRRK2_RCKW structure. Our MD simulations show a dominant interaction of the C-terminal residue, E2527, both the side chain and the α-carboxyl group, with R1771 and K1772 at the beginning of the COR-B Helix (Fig 4B).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. The dynamic of Ct-Helix.

(a) The Ct-Helix spans across both the N- and the C-lobe of the kinase domain with the C-terminus being located in close proximity to the COR-B Helix and the COR-B loop. The Ct-Helix and the according interaction sites are shown and colored by their relative fractional uptake (PDB: 6VNO). (b) Showing the surface electrostatic potential of the Ct-Helix. The Ct-Helix docks on the kinase domain through the side that is positively charged while the other side is negatively charged and involved in interactions with the NTDs. (c) The deuterium uptake of selected peptides is plotted and mapped on the LRRK2_RCKW structure. The COR-B-kinase loop and the loop in COR-B domain both show high deuterium uptake (70%–90%), indicating that they are solvent exposed. And their uptake is reduced in the presence of MLi-2. The data underlying the graphs shown can be found in S1 Data. Ct-Helix, C-terminal helix; NTD, N-terminal domain.

https://doi.org/10.1371/journal.pbio.3001427.g005

Interactions between the COR-B domain, the ROC domain, and the activation loop in the C-lobe of the kinase domain

The C-terminus of the COR-B Helix is close to both the A-loop of the kinase domain and the ROC domain, and in the inactive conformation that is captured by the cryo-EM structure (PDB: 6VNO), the tip of this helix is anchored to the ROC domain by the side chain of R1441, which binds to the backbone carbonyl of W1791 and helps to “cap” the COR-B Helix (Fig 4C and 4D). The C-terminus of the COR-B Helix also faces the A-loop of the kinase domain, which is likewise disordered in the LRRK2_RCKW cryo-EM structure (PDB: 6VNO) (Fig 3A). We know that this is a critical region because exchange of R1441 to either C, G, or H is one of the well-documented PD mutations that leads to activation of LRRK2 [22]. R1441 as well as N1437, another PD mutation at this interface, are also thought to impair the monomer–dimer cycle of LRRK2 and affect GTPase activity [23].

While a single static conformational state is trapped in the cryo-EM structure, GaMD simulations capture additional potential domain:domain interactions that can occur in this region. The simulations suggest, for example, that the side chain of R1441 can also interact with E1790, a residue that is anchored to R1915 in the αC helix of the kinase domain (Figs 3C and 4E and S2 File). This is the residue that would be predicted to interact with the phosphorylation site (P-site) in the A-loop when the kinase is in an active conformation [15,16]. The dynamics of the COR-B Helix would likely be significantly influenced by PD mutations of R1441 and N1437, as well as Y1699C, which would all, in principle, uncouple the ROC domain from the COR-B domain and thus enhance potential interactions with the kinase domain that may facilitate activation of LRRK2. Each of these pathogenic mutations would alter the dynamics of the COR-B domain by distinct mechanisms, leaving it free to interact with the A-loop of the kinase domain. Based on our GaMD simulations, multiple residues, such as Q2022, Y2023, and R2026 in the A-loop, can potentially interact with the C-terminus of the COR-B Helix (S5 Fig) once the ROC:COR-B interface is altered in response to PD mutations. These interactions, as discussed later, could be important for stabilizing the A-loop in an extended conformation, which would obviously affect LRRK2 kinase phosphorylation and activation.

Capturing interactions between the Ct-Helix and the kinase domain

The structures of LRRK2 reveal a unique helix at the C-terminus that extends from the WD40 domain and spans both N- and C-lobes of the kinase domain. This Ct-Helix is present in the inactive LRRK2_RCKW cryo-EM structure (PDB:6VNO) and is also docked onto the kinase domain in the full-length cryo-EM structure (PDB:7LHW), which corresponds to an inactive dimer, indicating that it is a very stable helix [9–11]. The combined HDX-MS data captures the dynamic features of the interface of the Ct-Helix with the kinase domain (Fig 5A). The short segment connecting the kinase and the WD40 domain is embedded between the N-terminus of the Ct-Helix and the C-terminus of the αE helix in the kinase domain and all show low deuterium uptake, suggesting a stable interaction between the WD40 domain and the C-lobe of the kinase domain that is shielded from solvent. The Ct-Helix also interacts with the β7-β8 loop in the kinase domain, which is larger in LRRK2 when compared to most kinases (Fig 5A). The αC-β4 loop of the kinase domain, which is an allosteric docking surface for some kinases such as BRAF [24], is almost completely shielded from solvent (S6 Fig and S1 Table). The Ct-Helix interacts with the kinase domain through both hydrophobic and positively charged residues on one side (Fig 5B). The mainly negatively charged residues on the other surface could potentially create a binding interface with the N-terminal domains (NTDs) of LRRK2 when the kinase activity of LRRK2 in the cell is quenched. The recently solved cryo-EM structure of full-length LRRK2 revealed how the ANK and LRR domains interact with the Ct-Helix in the inactive state [11]. The charged surface could also be involved in binding or tethering to substrates or activators when the N-terminal noncatalytic domains are not restrained by the C-terminal domains and in particular by the kinase domain.

The C-terminal portion of the Ct-Helix directly interacts with the COR-B domain and the N-lobe of the kinase domain

The last 3 residues at the end of the Ct-Helix are at the junction between the COR-B domain and the N-lobe of the kinase domain. This region forms a “cap” for the N-lobe of the kinase domain (Fig 6A). The 2 peptides that cover the COR-B-Kinase linker (residues 1,840 to 1,861 and residues 1,862 to 1,876, respectively) have 70% and 75% uptake of deuterium at 2 min, respectively, and both show a noticeable decrease in deuterium uptake upon binding of MLi-2. The loop in the COR-B domain, which interacts with the C-terminal tail (residues 1,713 to 1,724) also shows reduced uptake (Fig 5C and S1 Table). These changes in deuterium uptake in response to MLi-2 binding indicate correlated changes in dynamics that include both the kinase and COR-B domains and highlights that long-distance communication between the GTPase and kinase domains takes place. Our GaMD simulations capture some of the potential interactions that could occur in this region (Figs 4B and 6B). C-terminal residue E2527 and the free C-terminal carboxyl moiety are of particular interest as the simulations show how the last 5 residues can fluctuate between different structural states and form different interactions with the COR-B domain, the N-lobe of the kinase domain, and the linker between the COR-B domain and the kinase domain. Two arginine residues, R1723 and R1725, in the long loop of the COR-B domain that is disordered in the cryo-EM structures, form H-bonds with E2527 (Fig 6B and S3 File). Another 2 positively charged residues, R1771 and K1772, at the N-terminal end of the COR-B Helix could stabilize the C-terminal tail (residues 2,522 to 2,527) based on the simulations (Fig 4B and S1 File). R1866 on the COR-B-Kinase linker can also bind to E2527. Other residues, such as E1899 in the loop that connects the β2 and β3 strands of the kinase domain, can also interact with R2523 (Fig 6B). In addition, the 2 hydroxyl groups from T2524 and S2525 are also capable of forming H-bonds with either the COR-B or the COR-B-Kinase linker. These interactions that appear in different states during the simulations show how the C-terminal tail can potentially bridge to the COR-B domain and the kinase domain and thereby contribute to the intramolecular network that regulates the kinase and GTPase domains.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 6. Capping of the N-lobe of the kinase domain.

(a) In the LRRK2_RCKW cryo-EM structure, the linker from COR-B to the kinase domain lies over the N-lobe of the kinase domain. Nearby is a disordered loop from COR-B and the disordered 3 terminal residues (PDB: 6VNO). The loops, COR-B Helix, and the Ct-Helix are colored based on the relative fractional uptake. (b) MD simulations capture potential interactions between the C-terminal residues, the COR-B-kinase loop, and basic residues at the N-terminus of the COR-B Helix. cryo-EM, cryogenic electron microscopy; Ct-Helix, C-terminal helix; MD, molecular dynamics.

https://doi.org/10.1371/journal.pbio.3001427.g006

T2524 near the C-terminus is a known autophosphorylation site that can be recognized by 14-3-3 binding proteins [25,26]. GaMD simulations with phosphorylated T2524 (pT2524) suggest several new interactions, for example, with R1771 and R1886, when the C-terminal tail is closer to the COR-B-Kinase linker (S7A Fig). When the C-terminal tail is more distant from the N-lobe of the kinase, pT2524 binds to K1772 and R1723 (S7B Fig). The different networks that can be mediated upon phosphorylation can clearly affect the breathing dynamics of LRRK2 and could in turn affect LRRK2 activity. Also, yet to be resolved is whether 14-3-3 binding would stabilize an active or an inactive dimer.

Capturing the dynamics of the activation loop

As a frame of reference for the LRRK2 AS, we use the AS of the cAMP-dependent protein kinase (PKA) when the A-loop is phosphorylated and the kinase is in a fully closed conformation (S8 Fig). In PKA, the AS begins with the DFGψ motif and ends with the APE motif, 2 of the most highly conserved motifs in the protein kinase superfamily [27]. In between these 2 motifs are the A-loop and the P+1 loop. The APE-αF linker that follows the AS typically plays an important role in docking of substrates and other proteins and should be considered an extended part of the AS [13]. Our HDX-MS results show that the A-loop and part of the P+1 loop in the LRRK2 kinase domain are highly dynamic and likely unfolded (Fig 2 and S1 Movie), which is consistent with the fact that the A-loop and most of the P+1 loop are not resolved in the LRRK2_RCKW cryo-EM structure (PDB: 6VNO) (Fig 7A). The rest of the P-loop including the region extending from the APE motif through to the αF helix is, however, folded, and overlays well with the corresponding region of PKA. We consider first the disorder region in the LRRK2_RCKW structure. Several key residues in this disorder region face out toward the solvent, with the corresponding residues in PKA (R194) serving as a docking site for the regulatory subunits (R) (S9 Fig). The rest of the P+1 loop through to the αF helix is ordered and in a conformation that resembles PKA, and this conformation is conserved in the full-length LRRK2 structure and in the AlphaFold model. Several conserved residues/motif are embedded in this segment. In PKA, one of these highly conserved residues, Y215 (Y2050 in LRRK2), bridges to the backbone residues of the P-site (pT197) in the AL when the active kinase is phosphorylated on its A-loop [28,29] (S10 Fig). In the cryo-EM structure of LRRK2 (PDB: 6VNO), the AL is not phosphorylated and this Tyr is not in an active-like conformation.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 7. Comparison of AS of LRRK2.

The motifs that are embedded in the AS of LRRK2_RCKW (a), full-length LRRK2 (b), and the model from AlphaFold database (c) are summarized. The AS begins with the DFG motif and ends with the APE motif, 2 of the most highly conserved motifs in the protein kinase superfamily. In between these 2 motifs are the A-loop and the P+1 loop. While the AS in the inactive LRRK2_RCKW structure is mostly disordered, the AS in the inactive full-length LRRK2 is mostly ordered, only residue 2,028–2,030 are missing. The A-loop phosphate in LRRK2, T2031, is indicated as a red sphere in panel a and in the sequence alignment, which is the corresponding residue of T197 in PKA. Additional P-sites in LRRK2 are shown as black spheres in the sequence alignment. Black spheres in panels a–c correspond to highly conserved glycines that may serve as hinge points. A-loop, activation loop; AS, activation segment; LRRK2, leucine-rich repeat kinase 2.

https://doi.org/10.1371/journal.pbio.3001427.g007

Another interaction in this region, the APE motif, is a critical feature that distinguishes the eukaryotic protein kinases (EPKs) from the eukaryotic-like kinases (ELKs), which are their evolutionary precursors [28]. In the ELKs, the A-loops are short and not dynamic, and the helical domain (αG, αH, αI) is missing. The APE motif in the AS is anchored to this helical domain by a key and highly conserved electrostatic interaction between E2042 in the APE motif of LRRK2 and R2122 in the αH-αI loop. In the LRRK2_RCKW cryo-EM structure, these 2 residues are close but not within hydrogen bonding distance (PDB: 6VNO) (S11A Fig); however, the interaction between E2042 and R2122 is frequently captured in the GaMD simulations (S11B Fig) and is actually present in the full-length LRRK2 structure (PDB: 7LHW) (S11C Fig) and in the AlphaFold model (S11D Fig).

Surprisingly, when looking at the inactive cryo-EM structure of full-length LRRK2 (PDB: 7LHW), which also contains ATP, the AS is mostly ordered except for 3 residues (Fig 7B). In this structure, the complete P+1 loop is ordered in a way that agrees with the AlphadFold model (Fig 7C) and overlays well with PKA (PDB: 1ATP). However, the DFGψ motif region is ordered in a helix that is buttressed up against the N-lobe in contrast to an active conformation where the DFGψ motif would be fused to a beta strand that binds to the C-lobe (S10 Fig). Based on PKA, there are at least 2 docking motifs embedded within the AS. One docking site, discussed above, is created by the outward facing surface of the A-loop; the other is created by the outward docking surface of the APE-αF motif. In PKA, these 2 sites are integrated to create a highly dynamic allosteric site that is destroyed by the binding of cAMP (S9 Fig). In LRRK2, based on our HDX-MS/GaMD data, we predict that the A-loop will reach over to the COR-B Helix while the APE-αF linker will dock onto COR-A. Both states are captured in our GaMD simulations (S12 Fig).

Hydrogen–deuterium exchange mass spectrometry

LRRK2_RCKW proteins were expressed and purified from Sf9 cell [12]. HDX-MS was performed using a Waters Synapt G2Si equipped with nanoACQUITY UPLC system with H/DX technology and a LEAP autosampler. The LRRK2_RCKW concentration was 5 μM in LRRK2 buffer containing 20 mM HEPES/NaOH (pH 7.4), 800 mM NaCl, 0.5 mM TCEP, 5% Glycerol, 2.5 mM MgCl₂, and 20 μM GDP. The deuterium uptake was measured in LRRK2 buffer in the presence and absence of the kinase inhibitor MLi-2 (50 μM). For each deuteration time, 4 μL complex was equilibrated to 25°C for 5 min and then mixed with 56 μL D₂O LRRK2 buffer for 0, 0.5, 1, or 2 min. The exchange was quenched with an equal volume of quench solution (3 M guanidine, 0.1% formic acid (pH 2.66)). The quenched sample (50 μL) was injected into the sample loop, followed by digestion on an in-line pepsin column (immobilized pepsin; Pierce) at 15°C. The resulting peptides were captured on a BEH C18 Vanguard precolumn, separated by analytical chromatography (Acquity UPLC BEH C18, 1.7 μM, 1.0 × 50 mm, Waters Corporation) using a 7% to 85% acetonitrile gradient in 0.1% formic acid over 7.5 min, and electrosprayed into the Waters SYNAPT G2Si quadrupole time-of-flight mass spectrometer. The mass spectrometer was set to collect data in the Mobility, ESI+ mode; mass acquisition range of 200 to 2,000 (m/z); scan time 0.4 s. Continuous lock mass correction was accomplished with infusion of leu-enkephalin (m/z = 556.277) every 30 s (mass accuracy of 1 ppm for calibration standard). For peptide identification, the mass spectrometer was set to collect data in MS^E, ESI+ mode instead.

The peptides were identified from triplicate MS^E analyses of 10 μM LRRK2_RCKW, and data were analyzed using PLGS 3.0 (Waters Corporation). Peptide masses were identified using a minimum number of 250 ion counts for low energy peptides and 50 ion counts for their fragment ions. The peptides identified in PLGS were then analyzed in DynamX 3.0 (Waters Corporation) using a cutoff score of 6.5, error tolerance of 5 ppm, and requiring that the peptide be present in at least 2 of the 3 identification runs. The peptides reported on the coverage maps are those from which data were obtained. The relative deuterium uptake for each peptide was calculated by comparing the centroids of the mass envelopes of the deuterated samples versus the undeuterated controls [34]. For all HDX-MS data, at least 2 biological replicates were analyzed each with 3 technical replicates. Data are represented as mean values +/− SEM of 3 technical replicates due to processing software limitations; however, the LEAP robot provides highly reproducible data for biological replicates. The deuterium uptake was corrected for back exchange using a global back exchange correction factor (typically 25%) determined from the average percent exchange measured in disordered termini of various proteins [35]. Deuterium uptake plots were generated in DECA (github.com/komiveslab/DECA), and the data are fitted with an exponential curve for ease of viewing [36].

Gaussian accelerated molecular dynamics (GaMD) simulation

The LRRK2_RCKW model for simulations were prepared based on the reported LRRK2_RCKW structure (PDB: 6VP6) using Modeller to model the missing loops [37]. The Protein Preparation Wizard was used to build missing sidechains and model charge states of ionizable residues at neutral pH. Hydrogens and counter ions were added, and the models were solvated in a cubic box of TIP4P-EW water [38] and 150 mM KCl with a 10-Å buffer in AMBER tools D.A. Case, 2016 #731}. AMBER16 was used for energy minimization, heating, and equilibration steps, using the CPU code for minimization and heating and GPU code for equilibration. Parameters from the Bryce AMBER parameter database were used for phosphoserine and phosphothreonine [39]. Systems were minimized by 1,000 steps of hydrogen-only minimization, 2,000 steps of solvent minimization, 2,000 steps of ligand minimization, 2,000 steps of side-chain minimization, and 5,000 steps of all-atom minimization. Systems were heated from 0 K to 300 K linearly over 200 ps with 2 fs time-steps and 10.0 kcal/mol/Å position restraints on protein. Temperature was maintained by the Langevin thermostat. Constant pressure equilibration with an 8-Å nonbonded cutoff with particle mesh Ewald was performed with 300 ps of protein and peptide restraints followed by 900 ps of unrestrained equilibration. GaMD was used on GPU-enabled AMBER16 to enhance conformational sampling [40]. GaMD applies a Gaussian-distributed boost energy to the potential energy surface to accelerate transitions between metastable states while allowing accurate reweighting with cumulant expansion. Both dihedral and total potential acceleration were used simultaneously. Potential statistics were collected for 2 ns followed by 2 ns of GaMD during which boost parameters were updated for each simulation. Each GaMD simulation was equilibrated for 10 ns. For each construct, 10 independent replicates of 200 ns of GaMD simulation were run in the NVT ensemble, for an aggregate of 2.0 μs of accelerated MD.