High Resolution Solvated Models Reveal Mechanisms of Allosteric Activation of mTORC1 by RHEB

By integrating AlphaFold-3 modeling with molecular dynamics simulations, this study reveals how RHEB binding induces global structural and energetic remodeling of the mTORC1 complex to stabilize ATP coordination and preorganize the kinase for catalysis.

The Gist

The Big Picture: The Cell's "Master Switch"

Imagine your body is a massive, bustling city. In this city, there is a Master Switch called mTORC1. This switch decides whether the city should grow, build new buildings, and stock up on food (cell growth), or slow down and conserve energy when supplies are low.

For a long time, scientists knew that a tiny helper protein called RHEB acts like a "Green Light" to flip this switch on. But there was a problem: when scientists tried to take a picture of this switch with the best cameras available (Cryo-EM), the image was blurry. It was like looking at a complex machine through a foggy window. They could see the general shape, but they couldn't see the tiny gears turning or the specific bolts tightening that actually made the machine work.

The Problem: The Foggy Window

The paper explains that while we had these blurry pictures, we couldn't fully understand how RHEB turns the switch on.

• The Blur: The pictures were missing about 16% of the parts (like missing puzzle pieces).
• The Limitation: Because of the blur, scientists couldn't run computer simulations to see how the machine moves, vibrates, or changes shape in real-time.

The Solution: A High-Tech "3D Printer" and a "Virtual Wind Tunnel"

To fix this, the researchers used a clever three-step strategy to build a crystal-clear, 3D movie of the machine:

1. The AI Architect (AlphaFold-3): They used a super-smart AI (like a master architect who has seen every building in the world) to predict what the missing puzzle pieces looked like based on the protein's blueprint.
2. The Virtual Wind Tunnel (MDFF): They took this AI prediction and forced it to fit perfectly into the blurry "foggy" pictures from the real experiments. Think of this like molding clay into a shape that matches a faint shadow.
3. The Relaxation Phase (Molecular Dynamics): Finally, they let the computer "shake" the model gently. This allowed the atoms to settle into their most natural, comfortable positions, filling in the gaps and smoothing out the wrinkles.

The result? A high-definition, fully solvated (drenched in virtual water) model that shows exactly how the machine moves.

The Discovery: How the "Green Light" Works

Once they had this crystal-clear model, they watched what happened when RHEB (the Green Light) arrived. Here is what they found, using some analogies:

1. The "Squeeze and Stretch" Effect

Imagine the mTORC1 machine is shaped like a rugby ball.

• Without RHEB: The ball is loose and floppy.
• With RHEB: RHEB grabs the ball and squeezes the top and bottom closer together while stretching the sides.
• The Result: This squeezing action pulls the two main halves of the machine (RAPTOR and mTOR) tighter together, while pushing the side supports (mLST8) slightly apart. It's like tightening a screw to lock the machine into place.

2. The "Perfectly Aligned" Engine

Inside the machine is an engine (the Kinase Domain) that needs fuel (ATP) to run.

• Before RHEB: The engine parts are slightly misaligned. The fuel tank is a bit wobbly, and the spark plugs aren't quite touching the right spot.
• After RHEB: The squeezing action from step 1 forces the engine parts to snap into perfect alignment.
  • The Spark: A critical part of the engine (a specific amino acid called D2338) moves closer to the fuel.
  • The Magnesium: The machine swaps out some "water" helpers for "fuel" helpers, creating a tighter, more efficient grip on the magnesium ions needed for the reaction.
  • The Energy: This realignment makes it energetically "cheaper" (more favorable) for the machine to grab and hold the fuel. It's like the difference between trying to catch a slippery fish with wet hands versus catching it with a net.

3. The "Jiggle" Factor

The researchers also looked at how much the machine wiggles.

• The Paradox: Usually, you want a machine to be stiff and stable. But here, RHEB makes one specific part of the engine (the N-lobe) wobble more.
• Why? This extra "jiggle" is actually good! It makes the engine more flexible and ready to accept a new passenger (the substrate) to be worked on. It's like a door that is slightly loose on its hinges; it's easier to swing open quickly when someone needs to walk through.

The Bottom Line

This paper is a breakthrough because it moves us from guessing how the machine works to seeing it in high definition.

• Old View: RHEB turns the switch on. (We knew this, but didn't know how).
• New View: RHEB acts like a master mechanic. It physically reshapes the machine, tightening the frame, aligning the gears, swapping out the lubricants, and adding just the right amount of "wiggle" to the engine. This prepares the machine to catch fuel and work at maximum speed before the actual work even begins.

This new understanding is a huge deal for medicine. Since this "Master Switch" is often stuck in the "ON" position in many cancers, understanding exactly how it gets turned on helps scientists design better drugs to jam the gears or block the Green Light, potentially leading to new cancer treatments.

Technical Summary

1. Problem Statement

The mechanistic target of rapamycin complex 1 (mTORC1) is a massive (~1.2 MDa) dimeric complex essential for regulating cell growth. While Cryo-EM structures (e.g., PDB IDs 6BCU and 6BCX) have provided a foundational view of how the small GTPase RHEB allosterically activates mTORC1, they suffer from significant limitations:

• Resolution Heterogeneity: The maps have non-uniform resolution, particularly in flexible loops and inter-subunit interfaces.
• Missing Data: Approximately 16–17% of residues (~1,400 atoms) are unresolved in the experimental models.
• Dynamic Blindness: The static, low-resolution nature of Cryo-EM prevents the study of atomic-level dynamics, energetic changes, and the precise mechanism of ATP binding and catalysis.
• Modeling Limitations: Standard homology modeling cannot fill large gaps, and direct application of AlphaFold to such large, conformationally flexible complexes often results in atomic clashes or failure to capture large-scale rearrangements.

Consequently, a high-resolution, atomistic, and fully solvated model was needed to understand the allosteric mechanism by which RHEB primes mTORC1 for catalysis.

2. Methodology

The authors developed a hybrid computational pipeline to bridge the gap between Cryo-EM data and high-resolution molecular dynamics (MD):

• Initial Modeling (AlphaFold-3):

  • Used AlphaFold-3 to generate initial monomeric models of mTOR, mLST8, RAPTOR, and RHEB.
  • Employed a "divide-and-reassemble" strategy due to sequence length limits, modeling monomeric halves first and then assembling the full dimeric complex.
  • Aligned these models to the Cryo-EM density maps (6BCU for RHEB-bound, 6BCX for RHEB-free) to minimize RMSD.

• Refinement (Molecular Dynamics Flexible Fitting - MDFF):

  • Performed MDFF using NAMD to fit the AlphaFold models into the experimental Cryo-EM density maps.
  • Applied biasing potentials derived from the density maps to drive heavy atoms into high-density regions while maintaining secondary structure and chirality restraints.
  • This step resolved missing residues and corrected large-scale conformational shifts.

• Equilibration and Production MD:

  • Ligand Docking: Added ATP, GTP, and Mg²⁺ ions to the active sites, aligning them with Cryo-EM coordinates.
  • Progressive Relaxation: Conducted a multi-stage equilibration (energy minimization, heating from 0K to 300K, and NPT equilibration) with gradually decreasing harmonic restraints. This allowed modeled (previously missing) regions to relax while preserving experimentally resolved regions.
  • Production Simulations: Ran multiple 15–25 ns NPT production simulations for four systems: mTORC1 ± RHEB ± ATP.

• Validation and Analysis:

  • Structural Metrics: Validated models against Cryo-EM using LDDT, GDT-TS/HA, TM-score, and Interface Similarity (IS/ITM) scores.
  • Energetics: Calculated ATP binding enthalpy ($\Delta H$) using MM/GBSA (Molecular Mechanics/Generalized Born Surface Area) to dissect electrostatic and van der Waals contributions.
  • Dynamics: Analyzed cumulative variance of coordinate fluctuations ($\sigma^2_{CVCF}$) to map the energy landscape and domain flexibility.
  • Active Site Analysis: Examined ATP solvent accessibility (pSASA), Mg²⁺ coordination spheres, and distances between catalytic residues (e.g., D2338) and ATP.

3. Key Contributions

• Generation of Complete Atomistic Models: Created fully solvated, high-resolution models of the ~1.2 MDa mTORC1 dimer with and without RHEB, filling ~16% of the missing residues from Cryo-EM.
• Hybrid Modeling Pipeline: Established a robust workflow combining AlphaFold-3, MDFF, and progressive equilibration to model large, dynamic multi-protein complexes that are otherwise intractable for pure AI prediction or pure MD.
• Mechanistic Insight into Allosteric Activation: Provided a detailed molecular explanation of how RHEB binding transitions mTORC1 from an inactive to a catalytically competent state.

4. Key Results

A. Global Conformational Rearrangement

• Complex Geometry: RHEB binding causes the complex to contract along the major axis (bringing RAPTOR copies closer) and expand along the minor axis (moving mLST8 copies apart).
• Domain Shifts: RHEB induces a significant compaction of the N-HEAT and M-HEAT domains of mTOR (decreasing separation by ~7 Å in the refined model vs. ~16 Å in raw Cryo-EM).
• Kinase Domain: This compaction drives the N-lobe and C-lobe of the kinase domain closer together, narrowing the catalytic cleft.

B. Energetic Stabilization of ATP Binding

• Enthalpic Gain: RHEB binding makes ATP binding enthalpically more favorable by approximately 30 kcal/mol.
• Pre-organization: Surprisingly, direct non-bonded interactions between ATP and the active site residues are weaker in the RHEB-bound state. Instead, RHEB induces a pre-organization of the active site residues that creates a highly favorable electrostatic environment, stabilizing the ATP-Mg²⁺ complex enthalpically.
• State Dependence: RHEB stabilizes the +ATP state locally but destabilizes the -ATP state at longer ranges, effectively lowering the energy barrier for ATP binding.

C. Active Site Remodeling

• ATP Conformations: In the absence of RHEB, ATP exists in a single, deeply buried state. RHEB binding creates two distinct ATP populations: one buried and one more solvent-exposed.
• Catalytic Competence: The RHEB-bound states show shorter distances between the catalytic residue Asp2338 and the ATP $\gamma$-phosphate (~8.4–10 Å vs. ~11.5 Å in RHEB-free), positioning the complex for phosphoryl transfer.
• Mg²⁺ Coordination: RHEB alters the coordination sphere of Mg²⁺ ions, replacing water ligands with ATP phosphate oxygens and swapping a Glutamate ligand for an Aspartate (D2357), optimizing the geometry for catalysis.

D. Allosteric Dynamics

• Differential Stabilization: RHEB binding stabilizes the FAT and M-HEAT domains (reducing their flexibility) but destabilizes (increases flexibility of) the Kinase N-lobe.
• Catalytic Cleft Dynamics: The increased flexibility of the N-lobe and the catalytic cleft in the RHEB-bound state facilitates substrate accommodation and catalysis, contrasting with the rigid, inactive state of the RHEB-free complex.

5. Significance

• Mechanistic Clarity: This study moves beyond static structural snapshots to reveal the dynamic and energetic basis of mTORC1 activation. It demonstrates that allosteric activation is driven by a global remodeling that pre-organizes the active site for catalysis, rather than just simple domain closure.
• Therapeutic Implications: Understanding the specific allosteric pathway (RHEB $\to$ N/M-HEAT $\to$ Kinase Domain) offers new targets for third-generation inhibitors (Rapalinks) and allosteric drugs that could bypass resistance mutations (e.g., A2034V, F2108L) found in current ATP-competitive inhibitors.
• Methodological Advance: The pipeline serves as a template for studying other massive, flexible biological assemblies where Cryo-EM resolution is insufficient for dynamic analysis, enabling the integration of AI prediction with experimental density maps.