For the first time, researchers have solved the molecular structure of LRRK2, a major risk factor for Parkinson’s and autoimmune diseases. Structural biologists at the University of California, San Diego, used innovative methods to provide two complementary views of the molecule. In the August 7 Cell, researchers led by Elizabeth Villa described the three-dimensional architecture of full-length, pathogenic LRRK2 inside cells. The protein formed filaments that corkscrewed around microtubules. Researchers led by Andres Leschziner isolated a fragment of wild-type LRRK2 suitable for cryoEM. In the August 19 Nature, their high-resolution molecular maps revealed that this portion of the protein folds to bring its kinase and GTPase domains into close proximity. The finding explains previous experimental data suggesting the two domains interact.
- By cryoimaging LRRK2, researchers solve its three-dimensional structure.
- Kinase and GTPase domains lie close together.
- Pathogenic mutations enhance microtubule binding, which blocks cargo transport.
By combining data from both studies, the research groups determined that for LRRK2 to bind microtubules, its kinase has to be in a closed, or active, conformation. Pathogenic mutations, some of which are known to increase microtubule binding, appear to bias the molecule toward this shape. Leschziner and colleagues’ data hint that microtubule binding could contribute to toxicity, because LRRK2 filaments impede the passage of motor proteins down these highways, potentially causing traffic jams.
LRRK2 kinase inhibitors currently being tested as PD therapeutics may also trap the protein in this closed conformation.
“The new structures described in these two elegant studies represent a significant step forward in our understanding of LRRK2,” Dario Alessi at the University of Dundee, Scotland, wrote to Alzforum (full comment below). “The data provide new insights into how pathogenic mutations might exert their effects by promoting LRRK2 to adopt a closed conformation that is able to bind microtubule filaments.”
A Peek at 3-D Structures in Their Native Habitat
Mutations in LRRK2 account for up to 10 percent of familial PD cases. In addition, the concentration of this protein is elevated in many cases of sporadic disease, hinting at a broad role in pathology (Di Maio et al., 2018). Yet how LRRK2 contributes to Parkinson’s has remained murky. Besides being a kinase and GTPase, this large protein contains many other protein interaction domains (see diagram below). LRRK2 participates in numerous cellular processes, including vesicle trafficking, cell signaling, and autophagy, prompting a plethora of hypotheses about how it might cause harm (Oct 2012 news; Mar 2013 conference news). A structural map of the protein could help researchers decipher its function, but despite years of effort, LRRK2 has stubbornly resisted crystallization.
Villa and colleagues decided to flip the challenge of mapping LRRK2 on its head. Instead of trying to purify and crystallize the protein, they reasoned, why not simply image it in place? They took advantage of the fact that pathogenic LRRK2 is known to decorate microtubules, making it easy to find and visualize within the cell. Joint first authors Reika Watanabe, Robert Buschauer, and Jan Böhning expressed fluorescently tagged LRRK2 bearing the PD mutation I2020T in a kidney cell line, froze the cells, diced them into thin sections, and located LRRK2 on microtubules by correlating light and electron microscopy. Then they tilted the sections at different angles and imaged these decorated microtubules with an electron microscope in a process known as cryo-electron tomography. CryoET constructs three-dimensional structures from sequential two-dimensional images of surfaces (Lucić et al., 2008). With this technology, the authors mapped microtubule-bound LRRK2 filaments to a resolution of 14 angstroms.
Notorious Multitasker. In addition to a Ras of complex (ROC) GTPase (green), and a kinase domain (pink), LRRK2 contains many protein-protein interaction domains. The C-terminal of ROC (COR) N and C domains regulate the GTPase. These are commonly called COR-A (yellow) and COR-B (orange). [Courtesy of Watanabe et al., Cell.]
“To my knowledge, this is the first time someone has solved a structure inside a cell before it could be solved with biochemistry,” Villa said. “We took a technologically fancy but a biochemically lazy approach.”
CryoET revealed LRRK2 molecules forming a double helix around each microtubule, like the spiraling stripes of a candy cane (see image at top). In these long daisy chains of LRRK2 proteins, each bound to the one behind it through their respective WD40 domains (yes, here WD40 is a glue, not a lubricant), and to the one in front of it through their respective COR-B domains. The proteins were oriented such that their C-terminal halves, containing both catalytic domains, were located near the microtubule surface, while their N-terminal portions floated off into the cytoplasm and could not be resolved by cryoET. This orientation left the ROC GTPase domain facing the microtubule surface, and the kinase exposed to cytosol (see image below). Structural modeling suggested that the kinase was in the closed conformation, although this detail could not be resolved visually.
Dual Dimerization. Two LRRK2 monomers (left) sitting on a microtubule (outline) link via their COR domains (yellow), leaving their kinases (pink) exposed to cytosol. Their WD40 domains (red) link to adjoining monomers (gray). Rotation to show the view along the microtubule axis (right) exposes the GTPase domain (green) nestled against the microtubule surface. [Courtesy of Watanabe et al., Cell.]
Villa believes her group’s approach of using cryoET to scan molecules inside cells might help crack other recalcitrant structures, as well as provide clues to what proteins are doing in their native environment and how that changes during disease. “It’s the beginning of a new era of bridging structural and cellular biology,” Villa said.
Close Contact. The C-terminal half of LRRK2 folds to bring its kinase (orange) and GTPase (green) into proximity. Common Parkinson’s mutations are well-placed to modify this contact (right). [Courtesy of Deniston et al., Nature.]
A High-Resolution Glimpse of LRRK2’s Business End
For their part, Leschziner, co-corresponding author Samara Reck-Peterson, and colleagues took a different approach. Co-first author Sebastian Mathea in Stefan Knapp’s lab at Goethe University in Frankfurt had expressed the C-terminal half of wild-type human LRRK2 in insect cells, and found it was amenable to purification. Co-first author Colin Deniston imaged these molecules to 3.5 angstrom resolution by cryoEM. They determined that the molecule folded into a J shape that brought the ROC GTPase into close contact with the kinase domain (see image above). Previous studies had found that GTPase activity was essential for the kinase to function, but it was unclear how these domains interacted (Ito et al., 2007; West et al., 2007). In this protein fragment, the kinase assumed its open, catalytically inactive shape.
Intriguingly, the C-terminal tail of the WD40 domain formed a long α-helix that extended along the backbone of the kinase, interacting with it at several points. Noting that this α-helix contains at least one phosphorylation site, Leschziner speculated that modification of this tail might help regulate the shape of the kinase, perhaps switching it on and off.
Finally, the researchers overlaid their model onto the LRRK2 filaments described by Villa and colleagues to see if the structures matched. The monomer fit relatively well, but not perfectly—the COR domains clashed against those of the neighboring LRRK2s. When Leschziner and colleagues altered their structure to model a closed kinase domain, however, these steric clashes resolved (see image below). This finding suggested that an open conformation of the kinase would prevent microtubule binding, Leschziner said.
How Do Microtubules Fit In?
Whether LRRK2 gloms onto microtubules under physiological conditions is unclear. In cultured cells containing endogenous, wild-type LRRK2, the protein is not apparent on microtubules, Villa noted. However, when wild-type LRRK2 is overexpressed, it forms filaments on microtubules. In addition, five of the six most common PD mutations—I2020T, N1437H, R1441G, R1441C, and Y1699C—promote LRRK2 filament formation.
All of these mutations supercharge kinase activity, which would compel the kinase into its closed shape. I2020T sits in the activation loop of the kinase domain, right after G2019S, the most common pathogenic LRRK2 mutation, and the other three are at the interface between the GTPase and the COR-B domain, where they would be positioned to alter communication between the kinase and GTPase (see image above). “It seems that if you force LRRK2 into an active state, it binds microtubules,” Villa said.
Reck-Peterson’s data suggest that microtubule binding could cause problems. Co-first author John Salogiannis combined LRRK2, microtubules, and the motor proteins kinesin and dynein in cell-free assays. The motor proteins normally “walk” along microtubules, ferrying cargo toward the strands’ plus and minus ends, respectively. However, even low nanomolar amounts of LRRK2 shortened the distance the motors were able to walk. At 25 nM LRRK2, the motors ground to a halt, unable to step over the helical LRRK2 strands in their path.
No one knows if this roadblock serves a purpose, but Leschziner noted that LRRK2 is known to phosphorylate a subset of Rab GTPases that adorn vesicles transported along microtubules by motor proteins. Possibly, transient binding of LRRK2 oligomers to microtubules could pause motors long enough for the kinase to phosphorylate Rabs and change what cargoes get transported.
What Could This Mean for Therapy?
Questions about microtubule binding may be pertinent for PD therapy development. Type I kinase inhibitors trap the enzyme in its closed state, while keeping it inactive by preventing it from binding ATP. Do Type I LRRK2 inhibitors enhance microtubule binding? Deniston et al.’s data suggest as much, at least in cell-free assays. The researchers added the Type I inhibitor MLi-2 to their assay along with LRRK2, and found that the inhibitor further hampered motor protein movement along microtubules. MLi-2 is a pharmaceutical tool, not a drug in development (Fell et al., 2015; Scott et al, 2017). Conversely, Type II inhibitors, including the Bcr-Abl kinase inhibitor GZD-824, which stabilize an open kinase conformation, freed the motors to move again.
Mark Cookson at the National Institute on Aging, Bethesda, Maryland, noted that this finding might help explain the apparent paradox that LRRK2 kinase inhibitors enhance the formation of filaments inside cells, just as pathogenic PD mutations do. “This was particularly puzzling when considering that mutations in LRRK2 are gain-of-function, and we have thought of kinase inhibitors as potentially therapeutic,” Cookson wrote (full comment below).
“These data provide further insight into another potential pathological mechanism of LRRK2, and may explain some on-target toxicity of certain LRRK2 inhibitors, though more research is needed,” Andrew Koemeter-Cox at the Michael J. Fox Foundation wrote to Alzforum (full comment below). Likewise, Alessi suggested investigating whether Type II inhibitors would have fewer side effects.
The scientists are pursuing other LRRK2 riddles. Leschziner and Reck-Peterson have set their sights on the structure of other mutants. Curiously, so far G2019S has not been shown to increase LRRK2 binding to microtubules in cells. Like other pathogenic PD mutations, it turns on the kinase. However, unlike them, G2019S does not increase Rab phosphorylation in cells. “G2019S may contribute to disease in a different mechanistic way than the others,” Leschziner suggested.
To probe LRRK2’s physiological role, Villa will study endogenous LRRK2 in PD-relevant cell types, such as dopaminergic neurons and glia. In cells, the protein is more often found associated with membranes than microtubules. Does it assume a different shape when it binds membranes? Villa will recruit LRRK2 to membranes in cell culture, and combine cryoET with mass spectrometry to identify its structure and interaction partners.
Huaibin Cai at NIA believes that finding these interaction partners is crucial to deciphering what the protein does. “Future studies will be needed to determine the signaling cascades that regulate the conformational changes of LRRK2 kinase domain in different subcellular compartments, as well as to identify any particular cargoes stopped and modified by LRRK2,” he wrote to Alzforum (full comment below).—Madolyn Bowman Rogers
- Di Maio R, Hoffman EK, Rocha EM, Keeney MT, Sanders LH, De Miranda BR, Zharikov A, Van Laar A, Stepan AF, Lanz TA, Kofler JK, Burton EA, Alessi DR, Hastings TG, Greenamyre JT. LRRK2 activation in idiopathic Parkinson's disease. Sci Transl Med. 2018 Jul 25;10(451) PubMed.
- Lucić V, Leis A, Baumeister W. Cryo-electron tomography of cells: connecting structure and function. Histochem Cell Biol. 2008 Aug;130(2):185-96. Epub 2008 Jun 20 PubMed.
- Ito G, Okai T, Fujino G, Takeda K, Ichijo H, Katada T, Iwatsubo T. GTP binding is essential to the protein kinase activity of LRRK2, a causative gene product for familial Parkinson's disease. Biochemistry. 2007 Feb 6;46(5):1380-8. PubMed.
- West AB, Moore DJ, Choi C, Andrabi SA, Li X, Dikeman D, Biskup S, Zhang Z, Lim KL, Dawson VL, Dawson TM. Parkinson's disease-associated mutations in LRRK2 link enhanced GTP-binding and kinase activities to neuronal toxicity. Hum Mol Genet. 2007 Jan 15;16(2):223-32. PubMed.
- Fell MJ, Mirescu C, Basu K, Cheewatrakoolpong B, DeMong DE, Ellis JM, Hyde LA, Lin Y, Markgraf CG, Mei H, Miller M, Poulet FM, Scott JD, Smith MD, Yin Z, Zhou X, Parker EM, Kennedy ME, Morrow JA. MLi-2, a Potent, Selective, and Centrally Active Compound for Exploring the Therapeutic Potential and Safety of LRRK2 Kinase Inhibition. J Pharmacol Exp Ther. 2015 Dec;355(3):397-409. Epub 2015 Sep 25 PubMed.
- Scott JD, DeMong DE, Greshock TJ, Basu K, Dai X, Harris J, Hruza A, Li SW, Lin SI, Liu H, Macala MK, Hu Z, Mei H, Zhang H, Walsh P, Poirier M, Shi ZC, Xiao L, Agnihotri G, Baptista MA, Columbus J, Fell MJ, Hyde LA, Kuvelkar R, Lin Y, Mirescu C, Morrow JA, Yin Z, Zhang X, Zhou X, Chang RK, Embrey MW, Sanders JM, Tiscia HE, Drolet RE, Kern JT, Sur SM, Renger JJ, Bilodeau MT, Kennedy ME, Parker EM, Stamford AW, Nargund R, McCauley JA, Miller MW. Discovery of a 3-(4-Pyrimidinyl) Indazole (MLi-2), an Orally Available and Selective Leucine-Rich Repeat Kinase 2 (LRRK2) Inhibitor that Reduces Brain Kinase Activity. J Med Chem. 2017 Apr 13;60(7):2983-2992. Epub 2017 Mar 16 PubMed.
- Watanabe R, Buschauer R, Böhning J, Audagnotto M, Lasker K, Lu TW, Boassa D, Taylor S, Villa E. The In Situ Structure of Parkinson's Disease-Linked LRRK2. Cell. 2020 Sep 17;182(6):1508-1518.e16. Epub 2020 Aug 11 PubMed.
- Deniston CK, Salogiannis J, Mathea S, Snead DM, Lahiri I, Matyszewski M, Donosa O, Watanabe R, Böhning J, Shiau AK, Knapp S, Villa E, Reck-Peterson SL, Leschziner AE. Structure of LRRK2 in Parkinson's disease and model for microtubule interaction. Nature. 2020 Dec;588(7837):344-349. Epub 2020 Aug 19 PubMed.