Cooling fast and slow: Characterising the effects of vitrification in cryo-EM and the subsequent recovery of equilibrium populations

Through extensive molecular dynamics simulations of the Trp-cage miniprotein, this study demonstrates that while rapid vitrification in cryo-EM can perturb biomolecular ensembles, a developed thermodynamic inference framework can accurately recover equilibrium populations, validating cryo-EM as a reliable technique for studying biomolecular conformational landscapes.

The Gist

Imagine you have a bustling dance floor filled with people (the proteins) moving, spinning, and changing partners in a liquid room (the water). Scientists want to take a super-clear photo of everyone to see exactly what they are doing. But there's a problem: the camera needs to be in a vacuum, which means no liquid is allowed. If you just take a photo of the liquid, it would look like a blurry mess.

So, the scientists use a trick called cryo-EM. They grab the dance floor and plunge it instantly into a bucket of freezing liquid nitrogen. This turns the water into "glass" (vitrification) so fast that the dancers get frozen in mid-motion, preserving a snapshot of their dance.

The Big Question:
The worry was: Does the act of freezing them so fast change the dance? Maybe the dancers get stuck in weird poses they wouldn't normally hold, or maybe the fastest dancers get left behind while the slow ones get frozen in place. If the freeze changes the picture, the photo isn't a true representation of the dance floor.

What the Scientists Did:
To solve this mystery, the researchers didn't just take photos; they built a massive, ultra-detailed computer simulation. Think of it like a video game where they could control time.

1. The Simulation: They created a tiny digital world with a "Trp-cage" protein (a very small, simple dancer) and water.
2. The Speed Test: They ran the simulation at seven different speeds, from a slow, gentle freeze (like a winter night) to a super-fast plunge (like a lightning strike). They ran this for a total of 50 milliseconds—which is a blink of an eye for us, but an eternity for a tiny protein.
3. The Map: They created a "map" (called a Markov State Model) of all the different poses the protein likes to take when it's warm and happy.

What They Found:

• The Water Doesn't Care: The protein didn't change how the water froze. The water turned to glass just as it would on its own.
• The Fast vs. Slow Freeze: They discovered that if a protein pose takes a long time to change (a slow dance move), the freezing process is too fast to mess it up. It gets frozen exactly as it was. However, if a pose is very "jittery" and changes super fast, the freezing process might trap it in a slightly different spot than it would be at room temperature.
• The "Missing" Dancers: In a normal slow freeze, some of the most energetic, fast-moving dancers might disappear from the photo because they get trapped in a weird spot. But in their rapid freeze, nobody disappeared. Everyone was still there, just maybe slightly shifted.

The Solution:
The scientists realized that while the freeze might nudge the "jittery" dancers, it doesn't destroy the data. They invented a mathematical "decoder ring" (a thermodynamic inference framework). This tool allows them to look at the frozen photo, realize where the jittery dancers were nudged, and mathematically "undo" the nudge to figure out exactly where they were standing before the freeze.

The Bottom Line:
This paper is like a stamp of approval for cryo-EM. It tells us: "Don't worry about the freezing process messing up the picture." Even though we are freezing things incredibly fast, we can still trust the photos to show us the true, natural behavior of these tiny biological machines. We can now use these frozen snapshots to understand how proteins move and dance in real life, not just how they look when they are frozen.

Technical Summary

Problem Statement

Single-particle cryogenic electron microscopy (cryo-EM) is a premier technique for determining near-atomic resolution structures of biomolecules. However, the technique requires samples to be suspended in a vitrified (glass-like) water matrix to survive the high vacuum of the microscope. This process involves plunging samples into cryogens, resulting in rapid cooling rates that are inherently non-equilibrium.

The central challenge addressed in this paper is the uncertainty regarding how this vitrification process affects the conformational ensemble of biomolecules. Specifically, it is unknown whether the rapid cooling:

1. Traps the biomolecules in non-representative conformations.
2. Alters the relative populations of different conformational states compared to their equilibrium distribution at physiological temperatures.
3. Obscures the ability to derive accurate conformational ensembles, which is critical for understanding biomolecular dynamics and function.

Methodology

To investigate these effects, the authors employed a multi-faceted computational approach using the Trp-cage miniprotein as a model system:

• Molecular Dynamics (MD) Simulations: They conducted extensive MD simulations totaling over 50 milliseconds. These included:
  • Equilibrium Sampling: Simulations at 277 K to establish a baseline for the protein's natural conformational landscape.
  • Non-Equilibrium Cooling: Simulations of the protein undergoing rapid cooling at seven different rates spanning three orders of magnitude. The slowest cooling rates were calibrated to match experimental cryo-EM preparation conditions.
• Water Dynamics Analysis: The authors analyzed molecular mobility and density-temperature equations of state for water both with and without the protein to determine if the protein influences the vitrification of the solvent matrix.
• Markov State Modeling (MSM): To link conformational changes to kinetics, they constructed an MSM based on 5.4 milliseconds of equilibrium sampling at 277 K. This model allowed them to categorize states based on their characteristic timescales (kinetic stability).
• Thermodynamic Inference Framework: For states identified as "labile" (susceptible to cooling artifacts), the authors developed a novel mathematical framework to mathematically infer and recover the original equilibrium populations from the perturbed, vitrified ensembles.

Key Contributions

1. Decoupling Solvent and Solute Effects: The study demonstrates that the presence of the protein does not alter the fundamental vitrification process of water, confirming that the solvent matrix forms glass similarly regardless of the solute.
2. Kinetic Stability as a Predictor: The authors established a critical correlation between the characteristic time of a conformational state (from the MSM) and its robustness to cooling artifacts. States with characteristic times longer than the duration of the cooling process remain stable, while faster transitions are more susceptible to perturbation.
3. Discovery of State Preservation: Contrary to the fear that rapid cooling might eliminate certain states, the study found that while some states vanish in a static equilibrium ensemble at low temperatures (230 K), no states are lost in the non-equilibrium cooled ensembles. The cooling process effectively "freezes" the high-temperature ensemble rather than allowing it to relax to a low-temperature minimum.
4. Correction Framework: The development of a thermodynamic inference framework provides a practical tool to correct for cooling-induced perturbations, allowing researchers to reconstruct the true equilibrium populations from cryo-EM data.

Results

• Vitrification Integrity: Water vitrification proceeds unaltered by the Trp-cage protein, ensuring the solvent environment in simulations is physically realistic.
• State Robustness: MSM states with slow kinetics (long characteristic times) are robust against the non-equilibrium cooling artifacts.
• Ensemble Fidelity: The non-equilibrium cooling process preserves the diversity of the conformational ensemble observed at higher temperatures. Unlike a slow cooling to equilibrium at 230 K (which would lose high-energy states), the rapid cooling traps the ensemble in a state that reflects the higher-temperature distribution.
• Population Recovery: The proposed inference framework successfully accounts for perturbations in labile states, enabling the accurate recovery of equilibrium populations from the vitrified data.

Significance

This work provides a crucial validation for the use of cryo-EM in structural biology beyond static structure determination. By demonstrating that vitrification does not inherently destroy conformational diversity and by offering a method to correct for kinetic artifacts, the paper establishes that cryo-EM can be a reliable and accurate biophysical technique for studying biomolecular ensembles. This opens the door to using cryo-EM to map energy landscapes, study transient intermediate states, and understand the dynamic mechanisms of biomolecules in near-native conditions.