Proximity proteomics reveals a co-evolved LRRK2-regulatory network linked to centrosomes

This study utilizes BioID proximity proteomics combined with evolutionary and structural bioinformatics to map a co-evolved LRRK2 regulatory network linked to centrosomes and microtubules, revealing how the protein's interactions with specific cellular sub-compartments are dynamically modulated by its kinase activity and upstream effectors.

The Gist

Imagine your body is a bustling, high-tech city. In this city, there is a very important traffic controller named LRRK2.

When LRRK2 is working correctly, it keeps the city's roads (cells) running smoothly, directing traffic (proteins) to the right places. However, when LRRK2 gets "glitched" or overactive, it causes traffic jams and accidents, leading to a disease called Parkinson's.

This paper is like a team of detectives using a high-tech "proximity camera" to figure out exactly who LRRK2 is talking to, where they are meeting, and how their conversations change when things go wrong.

Here is the story of their investigation, broken down into simple parts:

1. The "Proximity Camera" (BioID)

Usually, scientists try to find who LRRK2 talks to by trying to grab it with a net (pull-down methods). But LRRK2 is shy; it only talks to people for a split second before running away. The net misses these quick chats.

So, the scientists used a special tool called BioID. Think of this as giving LRRK2 a spray can of invisible paint.

• When LRRK2 is in the cell, it sprays a tiny dot of "biotin" (the paint) on anyone standing too close to it.
• Later, the scientists wash the cell and use a magnet to pull out everything that has the paint on it.
• This reveals a list of everyone LRRK2 was hanging out with, even if they only met for a second.

2. The "Evolutionary Family Tree" (Co-evolution)

The scientists got a huge list of names, but they needed to know which ones were real friends and which were just random people passing by.

They used a clever trick: Co-evolution.

• Imagine two people who have been best friends since kindergarten. Over 100 years, they grow up together, change their clothes together, and move to new houses together. Their life stories are perfectly synced.
• The scientists looked at the "family trees" of LRRK2 and its potential friends across thousands of different species (from fish to humans).
• They found a special group of proteins that have been "growing up" with LRRK2 for millions of years. If LRRK2 changed, these friends changed too. This proved they are a tight-knit team.

The Big Discovery: This tight-knit team is mostly made up of construction workers who build and maintain the city's scaffolding and streetlights (specifically the centrosome and microtubules). These are the structures that give cells their shape and help them move.

3. The "Shape-Shifting" (Conformations)

LRRK2 is a shape-shifter. It can lock itself in a "safe" position or unlock into an "active" position. The scientists used a super-computer (AlphaFold) to build 3D models of LRRK2 shaking hands with its friends.

They found two distinct ways LRRK2 holds hands:

• The "Locked" Hug: LRRK2 is curled up tight. In this mode, it mostly talks to the construction crew (centrosome workers).
• The "Unlocked" Hug: LRRK2 stretches out. In this mode, it talks to the delivery drivers (vesicles and lysosomes) who move trash and packages around the cell.

4. The "Drug Test" (Inhibitors)

The scientists wanted to see if drugs could change who LRRK2 talks to. They tested two types of "brakes" (inhibitors) to stop LRRK2 from going crazy.

• The "Type I" Brake (MLi-2): When they applied this drug, LRRK2 instantly curled up into its "Locked" shape. Suddenly, it stopped talking to the delivery drivers and started hanging out only with the construction crew (specifically a group called centriolar satellites). It was like the drug forced LRRK2 to go to a construction site and ignore the rest of the city.
• The "Type II" Brake (GZD-824): This drug didn't change the crowd. LRRK2 kept talking to the same people as before.

Why does this matter? It tells us that different drugs do different things. If you want to fix the construction site, you need the "Type I" drug. If you want to fix the delivery routes, you might need a different approach.

5. The "Boss" (RAB29)

They also tested what happens when a "boss" protein called RAB29 shows up. RAB29 is known to tell LRRK2 to get to work.

• When RAB29 arrived, LRRK2 stopped hanging out with the construction crew and ran straight to the delivery trucks (lysosomes).
• This confirmed that RAB29 is the switch that tells LRRK2 to go manage the cell's trash and recycling system.

The Bottom Line

This paper is a map. It shows us that LRRK2 isn't just one thing; it's a chameleon.

• Depending on its shape and who is giving it orders, it hangs out with different groups of proteins.
• One group builds the cell's skeleton (centrosomes).
• Another group manages the cell's trash (lysosomes).

The Takeaway for Parkinson's:
Parkinson's happens when LRRK2 gets stuck in the wrong mode or talks to the wrong people. By understanding exactly who it talks to and when, scientists can design better drugs. Instead of just trying to shut LRRK2 down completely (which might hurt the lungs, as seen in other studies), we might be able to design drugs that gently guide LRRK2 back to the right conversation, fixing the specific problem without breaking the whole city.

In short: They found that LRRK2 has a secret life as a construction manager and a trash collector, and different drugs force it to choose one job over the other. This helps us understand how to fix the broken machinery in Parkinson's disease.

Technical Summary

1. Problem Statement

Leucine-rich repeat kinase 2 (LRRK2) is a major genetic risk factor for both familial and idiopathic Parkinson's disease (PD). While LRRK2 is known to function as a signaling hub involved in vesicle trafficking and Rab GTPase phosphorylation, its full interactome and the mechanisms governing its context-specific interactions remain poorly understood.

• Limitations of Current Methods: Traditional affinity purification methods (e.g., co-immunoprecipitation) are biased toward stable, long-lived complexes and often miss transient or weak interactions.
• Therapeutic Gap: Current LRRK2 kinase inhibitors face challenges regarding therapeutic windows (e.g., lung toxicity) and the need to understand alternative drug targets within LRRK2 pathways.
• Knowledge Gap: There is a lack of comprehensive data linking LRRK2's conformational states (active vs. inactive) and regulatory inputs (e.g., RAB29) to specific subcellular compartments and protein networks.

2. Methodology

The authors employed a multi-omics approach combining proximity labeling, evolutionary bioinformatics, and structural modeling:

• Proximity Proteomics (BioID):

  • Generated N-terminal fusion constructs of LRRK2 with three different biotin ligase tags: BioID1, BioID2, and miniTurbo.
  • Performed experiments in HEK293T cells under basal conditions, with MLi-2 (Type I inhibitor), GZD-824 (Type II inhibitor), and RAB29 co-expression.
  • Used tag-only controls to filter endogenous biotinylated proteins.
  • Identified interactors using SAINTexpress and Label-Free Quantification (LFQ) with FDR-controlled t-tests.

• Co-evolutionary Analysis:

  • Developed a pipeline to calculate Jaccard scores based on the co-occurrence of orthologous gene pairs across sequenced genomes (using the OMA database).
  • Used this metric to distinguish high-confidence, stable direct interactions from transient associations.
  • Clustered interactors based on co-evolutionary similarity to identify functional modules.

• Structural Modeling (AlphaFold-Multimer):

  • Predicted 3D binary complexes between LRRK2 and 196 identified interactors.
  • Analyzed Root Mean Squared Deviation (RMSD) to classify LRRK2 conformations into "locked" (inactive-like, compact) and "unlocked" (active-like, rearranged) states.
  • Generated interaction fingerprints (distograms) to map binding interfaces to specific LRRK2 domains (LRR, ROC, COR, Kinase, ARM, WD40).

• Experimental Validation:

  • Validated top hits (e.g., CYLD, PCM1, SSX2IP) via Western blotting and siRNA knockdown (specifically targeting PCM1 to disrupt centriolar satellites).
  • Assessed LRRK2 steady-state levels and phosphorylation status (pS935, pT73-Rab10) under various conditions.

3. Key Contributions

• Comprehensive Interactome Definition: Defined a non-redundant set of 208 high-confidence LRRK2 interactors, the majority of which were previously unreported.
• Evolutionary Prioritization: Demonstrated that co-evolutionary metrics (Jaccard scores) effectively predict direct, stable protein-protein interactions (PPIs), identifying a highly conserved module enriched in centrosomal and cytoskeletal proteins.
• Conformation-Specific Interactome: Mapped specific interactors to distinct LRRK2 conformations ("locked" vs. "unlocked"), revealing that structural states dictate functional partnerships.
• State-Dependent Rewiring: Showed that pharmacological inhibition (Type I vs. Type II) and upstream effectors (RAB29) induce distinct rewiring of the LRRK2 interactome, shifting localization between centriolar satellites and lysosomal pathways.

4. Key Results

A. The Centrosome-Linked Network

• Functional enrichment analysis revealed a massive module of interactors associated with microtubules, centrosomes, and cilia.
• Co-evolution Clustering: A specific cluster (Cluster 2) showed the highest co-evolution with LRRK2. The top interactor was CYLD (a deubiquitinase), followed by centrosomal proteins like CEP85 and CEP43.
• Validation: Co-expression of CYLD increased steady-state levels of the pathogenic LRRK2 Y1699C variant, suggesting CYLD regulates LRRK2 stability.

B. Conformational Switching and Interface Mapping

• Locked vs. Unlocked: AlphaFold modeling identified two main conformational clusters:
  • "Locked" (Inactive): Characterized by lower RMSD and higher co-evolution. Interactors here are enriched for centriolar satellite and cilium assembly functions.
  • "Unlocked" (Active): Characterized by major domain rearrangements (LRR/ROC/COR). Interactors here are enriched for cytoskeleton and microtubule organizing center (MTOC) functions.
• Binding Interfaces:
  • Core Cluster: Interacts with LRR-ROC-COR-Kinase domains (e.g., DVL1, CYLD).
  • WD40 Cluster: Binds the WD40 domain via disordered peptides.
  • ARM Cluster: Binds the Armadillo domain (e.g., AXIN1, RAB12).

C. State-Dependent Rewiring

• MLi-2 (Type I Inhibitor):
  • Induced a specific enrichment of centriolar satellite proteins (e.g., PCM1, SSX2IP, CEP131, OFD1).
  • These interactors showed the highest co-evolution scores and bound LRRK2 in the "locked" conformation.
  • Mechanism: MLi-2 binding leads to dephosphorylation of S935 and a conformational shift that favors association with the centriolar satellite network.
• GZD-824 (Type II Inhibitor): Did not induce the same specific enrichment of centriolar satellites, highlighting a mechanistic difference between inhibitor classes.
• RAB29 Co-expression:
  • Recruited LRRK2 to vesicular and lysosomal pathways (e.g., RAB8A, VAMP7, SPAG9).
  • RAB29 binding promoted the "unlocked" conformation and interaction with substrate-like partners (e.g., RAB8A) via the catalytic core.

5. Significance

• Pathophysiological Insight: The study establishes a direct mechanistic link between LRRK2, centrosomal cohesion, and primary cilia. This provides a molecular explanation for the ciliogenesis defects and centrosome cohesion issues observed in PD patient cells and models.
• New Therapeutic Targets: Identifies CYLD and the centriolar satellite network as potential alternative drug targets. Modulating this network could offer a way to control LRRK2 pathology without the lung toxicity associated with direct kinase inhibition.
• Conformational Pharmacology: Demonstrates that Type I inhibitors (like MLi-2) do not merely inhibit kinase activity but actively reorganize LRRK2 into specific subcellular compartments (centriolar satellites). This suggests that drug design must consider conformational states and subcellular localization, not just catalytic inhibition.
• Methodological Advance: The integration of proximity proteomics with co-evolutionary analysis and AlphaFold modeling provides a robust framework for dissecting complex, transient interactomes of large multi-domain proteins.

In conclusion, this paper redefines LRRK2 not just as a kinase, but as a conformationally dynamic scaffold that integrates signals from the centrosome, cilia, and vesicular trafficking systems, with distinct regulatory networks activated by specific inhibitors and upstream effectors.