STI1 domains coordinate partitioning of UBQLN2 into stress-induced condensates

This study reveals that while the STI1-II domain is essential for both baseline and stress-induced UBQLN2 condensate partitioning, the STI1-I domain plays a distinct, context-dependent role that is dispensable for baseline puncta formation but critical for robust stress-induced phase separation.

The Gist

Imagine your cell is a bustling, high-tech city. Inside this city, there's a crucial worker named UBQLN2. Think of UBQLN2 as a universal delivery truck that picks up broken or damaged packages (misfolded proteins) and drives them to the recycling plant (the proteasome) or the compost heap (autophagy) to be cleaned up.

Sometimes, the city gets overwhelmed by a heatwave (heat stress). When this happens, the city needs to build emergency shelters to protect important documents (mRNA) and stop traffic. These shelters are called stress granules.

This paper is a detective story about how UBQLN2 decides where to park its trucks. Does it park at the emergency shelters? Does it hang out at the recycling centers? And most importantly, what parts of the truck are needed to make it park in the right place?

Here is the breakdown of the findings using simple analogies:

1. The Truck Has Special "Hooks" (Domains)

UBQLN2 isn't just a plain box; it has different parts with specific jobs:

• The Front (UBL): The driver's seat. It talks to the recycling plant.
• The Back (UBA): The cargo hook. It grabs the broken packages.
• The Middle (STI1-I and STI1-II): These are the engine and the transmission. They are the most important parts for this story. They help the trucks stick together to form a convoy.

2. The "Normal Day" vs. The "Heatwave"

The researchers wanted to see what happens to these trucks on a normal day versus during a heatwave.

• On a Normal Day (Baseline):

  • The trucks naturally form small groups (puncta) in the city.
  • They love hanging out near the Recycling Centers (p62). Even without a crisis, if there are recycling centers, UBQLN2 trucks park there.
  • The Engine (STI1-II) is essential: If you remove the STI1-II engine, the trucks fall apart. They become a single, lonely vehicle that can't form a convoy. They just drive around aimlessly and never park.
  • The Transmission (STI1-I): On a normal day, this part isn't strictly necessary. The truck can still form a convoy without it, thanks to the engine.

• During a Heatwave (Heat Stress):

  • The city gets hot, and the emergency shelters (stress granules) open up.
  • The Front (UBL) is actually a brake: Surprisingly, the front part of the truck (the UBL domain) usually stops the trucks from forming too many groups. When the researchers removed this "brake," the trucks went wild and formed massive convoys everywhere, even inside the nucleus (the city hall).
  • Both Engines are needed for the Heatwave: When the heat hits, the trucks need both the STI1-II engine and the STI1-II transmission to react quickly and form the emergency shelters.
    • If you remove STI1-II, the trucks are paralyzed. They can't react to the heat at all.
    • If you remove STI1-I, the trucks can still move, but they are slow and weak. They form very few emergency shelters compared to a full truck.

3. The "Recycling Center" Connection

One of the coolest discoveries is that UBQLN2 trucks have a special relationship with the Recycling Centers (p62).

• Even when the city is calm, the trucks park at the recycling centers.
• When the heatwave hits, the trucks don't build new recycling centers; instead, they swarm the existing ones, packing them tighter.
• This is important because in diseases like ALS (a condition where the city's cleanup system fails), these trucks and recycling centers get stuck together in giant, solid piles that the city can't break down. This paper shows that the STI1-II engine is the glue holding these dangerous piles together.

4. The "In Vitro" Test (The Lab Simulation)

The researchers also tested these trucks in a test tube (outside the cell). They found a perfect match:

• If a truck part could form a liquid drop in a test tube, it would also form a parking group in the cell.
• If a truck part couldn't form a drop in the test tube (like the one missing the STI1-II engine), it wouldn't form a group in the cell either.
• This proves that the rules of physics (how proteins stick together) inside the cell are the same as in a test tube.

The Big Takeaway

Think of UBQLN2 as a smart delivery truck with two engines:

1. Engine 2 (STI1-II) is the Main Engine. It is always needed. Without it, the truck is broken and can't do its job, whether it's a calm day or a crisis.
2. Engine 1 (STI1-I) is the Turbo Charger. It's not needed for driving around on a calm day, but when the heatwave hits, you need the turbo to get the job done fast.

Why does this matter?
In diseases like ALS, mutations often break these engines. If the "Main Engine" (STI1-II) is broken, the trucks can't clean up garbage, leading to toxic piles. If the "Turbo Charger" (STI1-I) is broken, the city can't handle stress effectively. Understanding exactly which part does what helps scientists design better medicines to fix these specific broken parts.

Technical Summary

1. Problem Statement

Ubiquilin-2 (UBQLN2) is a critical shuttle protein in protein quality control (PQC) pathways, linking ubiquitinated substrates to the proteasome and autophagy. Mutations in UBQLN2 are a known cause of familial Amyotrophic Lateral Sclerosis (ALS) and Frontotemporal Dementia (FTD). While UBQLN2 is known to undergo liquid-liquid phase separation (LLPS) in vitro and form stress-induced condensates (such as stress granules and p62 puncta) in vivo, the specific molecular determinants governing its partitioning into these distinct condensates remain poorly characterized.

Specifically, the central region of UBQLN2 contains two STI1 domains (STI1-I and STI1-II) and disordered linkers. It is unclear how these individual domains contribute to:

1. Baseline puncta formation (in unstressed cells).
2. Stress-induced puncta formation (specifically in response to heat stress).
3. Recruitment to distinct condensate types (stress granules vs. autophagy receptor p62 puncta).
4. The relationship between in vitro phase separation propensity and in vivo cellular behavior.

2. Methodology

The authors employed a multi-faceted approach combining live-cell imaging, immunofluorescence, and in vitro biophysics:

• Cell Models: Used HeLa cells for endogenous protein analysis and HEK293 cells for overexpression studies.
• Construct Library: Generated a comprehensive library of mEos3.2-tagged UBQLN2 constructs, including:
  • Single domain deletions: $\Delta$STI1-I, $\Delta$STI1-II, $\Delta$PXX, $\Delta$UBA, and $\Delta$UBL (N-terminal truncation).
  • Sequential N-terminal truncations (e.g., 109-624, 178-624, 248-624) to dissect domain interactions.
• Stress Conditions:
  • Baseline: Imaging 24–30 hours post-transfection without stress.
  • Heat Stress: Live-cell timelapse imaging during a temperature ramp from 37°C to 42°C (30 min to 2 hours) and fixed-cell immunofluorescence after 30 min at 42°C.
• Imaging & Quantification:
  • Used machine learning (Trainable Weka Segmentation) for automated cell and puncta segmentation.
  • Quantified puncta volume, colocalization with markers (G3BP for stress granules, p62 for autophagy, FK2 for ubiquitin, HSP70), and nuclear-to-cytoplasmic (N:C) ratios.
• Biophysical Characterization:
  • SEC-MALS: Determined oligomeric states (monomer vs. dimer) of purified constructs.
  • Phase Separation Assays: Measured cloud point temperatures ($T_{cp}$) and saturation concentrations ($c_{sat}$) via turbidity and sedimentation assays to determine in vitro phase separation propensity.

3. Key Contributions

• Decoupling Domain Functions: The study systematically dissects the non-redundant roles of the two STI1 domains in UBQLN2 condensation, distinguishing between baseline and stress-induced mechanisms.
• Correlation of In Vitro and In Vivo: Establishes a strong quantitative correlation between in vitro phase separation propensity ($c_{sat}$) and cellular puncta formation under baseline conditions.
• Mechanism of Stress Response: Identifies that while STI1-II is essential for all condensation, STI1-I plays a latent but critical role specifically in heat stress responses, which is masked by N-terminal autoinhibition in the full-length protein.
• Recruitment Dynamics: Clarifies that UBQLN2 recruitment to p62 puncta is constitutive and enhanced by stress, whereas recruitment to stress granules (G3BP) is highly dependent on expression levels and is poor for overexpressed protein.

4. Key Results

A. Endogenous vs. Overexpressed UBQLN2 Behavior

• Stress Granules: Endogenous UBQLN2 robustly localizes to heat-induced stress granules (G3BP+). However, overexpressed mEos3.2-UBQLN2 shows poor recruitment to stress granules, consistent with previous reports.
• p62 Puncta: Both endogenous and overexpressed UBQLN2 colocalize significantly with p62 puncta. Heat stress increases the fraction of p62 contained within UBQLN2 puncta, suggesting stress recruits UBQLN2 to pre-existing p62 hubs rather than forming new p62 structures.
• Expression Dependence: Colocalization with p62 is negatively correlated with UBQLN2 intensity (high UBQLN2 favors homotypic condensates) but positively correlated with p62 intensity.

B. Domain Requirements for Baseline Puncta (No Stress)

• STI1-II is Essential: Deletion of STI1-II ($\Delta$STI1-II) completely abrogates baseline puncta formation. SEC-MALS confirms $\Delta$STI1-II is monomeric, whereas full-length and other constructs are dimeric. This indicates STI1-II drives the oligomerization necessary for baseline condensation.
• STI1-I is Dispensable: Deletion of STI1-I ($\Delta$STI1-I) has no significant effect on baseline puncta formation compared to full-length.
• UBA and N-terminus: Deletion of the UBA domain ($\Delta$UBA) nearly eliminates puncta. Conversely, removing the N-terminal UBL domain (construct 109-624) enhances puncta formation and increases nuclear localization, suggesting the N-terminus acts as an autoinhibitor.

C. Domain Requirements for Heat Stress Response

• Non-Redundant Roles: Both STI1 domains are required for the full heat stress response.
  • $\Delta$STI1-II: Shows almost no stress-induced puncta formation (remains diffuse).
  • $\Delta$STI1-I: Shows a substantial attenuation of the stress response (reaches only ~0.6% puncta area vs. ~1.2% for full-length).
• STI1-I Latency: The contribution of STI1-I to stress-induced condensation is "latent" in the full-length protein. When the N-terminal autoinhibitory region (UBL) is removed (e.g., in construct 178-624), STI1-I strongly promotes phase separation. This suggests STI1-I interacts with the N-terminus to suppress its activity under baseline conditions, but this suppression is relieved or overcome during heat stress.

D. Correlation with Biophysics

• There is a strong negative correlation (Spearman $\rho$ = -0.84) between in vitro saturation concentration ($c_{sat}$) and cellular puncta area. Constructs with lower $c_{sat}$ (higher propensity to phase separate) form more puncta in cells.
• $\Delta$STI1-II fails to phase separate in vitro under tested conditions, mirroring its failure to form puncta in vivo.

5. Significance

• Mechanistic Insight into ALS/FTD: The findings provide a structural framework for understanding how ALS-linked mutations in STI1 domains or the PXX region might differentially affect basal protein homeostasis versus stress responses. Mutations disrupting STI1-II would likely cause a total loss of condensation function, while those affecting STI1-I might specifically impair stress adaptation.
• Regulation of Phase Separation: The study reveals a sophisticated regulatory mechanism where the N-terminal UBL domain autoinhibits the STI1-I domain. This coupling suggests that cellular factors binding the UBL (e.g., proteasomal subunits) could relieve this inhibition, dynamically tuning UBQLN2 condensation in response to proteostatic load.
• Condensate Specificity: The data highlights that different condensates (stress granules vs. p62 bodies) recruit UBQLN2 via distinct mechanisms and domain requirements, challenging the view of a single "UBQLN2 condensate" state.
• Therapeutic Implications: Understanding that STI1-II is the primary driver of oligomerization while STI1-I modulates stress responsiveness offers specific targets for modulating UBQLN2 aggregation in neurodegenerative diseases.