Comparison of extracellular vesicles and mechanically induced vesicles for structure determination of membrane proteins

This study compares natively shed extracellular vesicles (EVs) and mechanically derived vesicles (MVs) from HER2-overexpressing SKBR3 cells, concluding that MVs offer a more suitable platform for determining the structure of membrane proteins in their native membrane environments due to their superior structural uniformity compared to the highly diverse EVs.

The Gist

The Big Picture: Taking a Snapshot of a Busy City

Imagine a cell is like a bustling, crowded city. The membrane proteins (like HER2) are the important buildings, traffic lights, and signs on the city's walls. Scientists want to take a high-resolution photo of these buildings to understand how they work and how to fix them when they break (which happens in diseases like cancer).

However, taking a photo of a building inside a crowded city is incredibly hard. The streets are too narrow, there's too much traffic, and the buildings are often hidden behind other structures.

To solve this, scientists tried two different ways to "pack up" a piece of the city wall and bring it into the lab for a closer look. They compared two types of "moving trucks":

1. The Natural Delivery Truck (EVs): These are tiny bubbles that cells naturally spit out into the bloodstream. Think of them as mailboxes that the city drops off to communicate with neighbors.
2. The Mechanical Delivery Truck (MVs): These are bubbles created by scientists physically squeezing the cells through a tiny syringe, like squeezing toothpaste out of a tube to get a piece of the wall.

The goal of this study was to see which "truck" gives us the cleanest, most accurate photo of the HER2 building.

---

The Experiment: Two Types of Bubbles

The researchers used a specific type of breast cancer cell (SKBR3) that is famous for having a massive number of HER2 "buildings" on its surface. They collected both the natural bubbles (EVs) and the squeezed bubbles (MVs).

1. The "Natural" Bubbles (EVs)

• What they found: These bubbles were like a chaotic moving van. When they looked inside, they found a mix of everything: old furniture, random trash, and even parts of the city's skeleton (cytoskeleton).
• The Problem: Because they were so full of "junk" and had weird shapes (some were long, some were cross-linked), it was very hard to get a clear photo. The inside was too dense, making the image blurry.
• Analogy: Trying to take a photo of a specific house through a window that is covered in fog, dirt, and other people standing in the way.

2. The "Mechanical" Bubbles (MVs)

• What they found: These bubbles were much more uniform. They were like neat, round shipping containers. They were mostly empty inside, with just the wall and the buildings attached to it.
• The Advantage: Because they were cleaner and rounder, the "fog" was gone. The scientists could see the buildings on the wall much more clearly.
• Analogy: Taking a photo of the same house, but this time the window is clean, and the street is empty.

The Special Tool: The "Velcro" Trap

One big worry was that when you squeeze the cell (to make the MVs), you might accidentally flip the wall inside out. If the HER2 building gets flipped, the "front door" faces the wrong way, and the photo is useless.

To fix this, the scientists used a clever trick involving DARPins (which are like tiny, custom-made Velcro hooks).

• They made a hook that only sticks to the front door of the HER2 building.
• They put these hooks on magnetic beads.
• When they mixed the bubbles with the beads, only the bubbles with the HER2 building facing the right way got stuck to the magnet.
• They then gently washed the bubbles off the magnet, ensuring they only studied the correctly oriented buildings.

The Result: A Blurry but Promising Photo

The scientists then used a super-powerful microscope (Cryo-EM) to take pictures of the HER2 buildings on the Mechanical Vesicles (MVs).

• The Good News: They successfully identified the HER2 building! They could see its shape and size. It looked like the scientists had finally found the right house in the city.
• The Bad News: The photo wasn't "4K Ultra HD" yet. It was more like a "low-resolution sketch."
  • Why? The HER2 building is very flexible. Its top part (the part sticking out) wiggles around a lot, like a jellyfish tentacle. Because it moves so much, the photo comes out a bit blurry.
  • Also, the bubbles were still a bit thick, which made the image slightly hazy.

The Conclusion: Which Truck Wins?

The paper concludes that while the Natural Bubbles (EVs) are interesting for studying how cells talk to each other, they are too messy for taking high-quality structural photos.

The Mechanical Bubbles (MVs) are the winners for structural biology. They are cleaner, more uniform, and provide a better platform for seeing how these important proteins look in their natural environment.

In short: If you want to study the architecture of a cell's wall, don't wait for the cell to spit out a bubble. Instead, gently squeeze the wall out, use a special magnet to grab the right pieces, and you'll get a much clearer picture of the machinery at work.

Why This Matters

Understanding the exact shape of HER2 helps scientists design better drugs to fight cancer. If we know exactly what the "lock" (HER2) looks like, we can build a better "key" (medicine) to open it or jam it, stopping the cancer from growing. This study shows a new, cleaner way to get that blueprint.

Technical Summary

1. Problem Statement

Membrane proteins are critical for cellular function, and their structure and conformation are heavily influenced by their native lipid environment. Traditional structural biology methods (e.g., X-ray crystallography, standard cryo-EM) typically require isolating proteins from their native membranes using detergents or reconstituting them into synthetic lipid bilayers, thereby losing the native context.

• Current Limitations: While Cryo-Electron Tomography (Cryo-ET) allows for imaging in situ, the dense molecular crowding of native cell membranes and the small size of many membrane proteins result in low signal-to-noise ratios, making high-resolution structure determination difficult.
• The Gap: There is a need for a platform that preserves the native membrane environment but offers sufficient homogeneity and clarity for high-resolution structural analysis. The authors investigate two types of cell-derived vesicles—naturally shed Extracellular Vesicles (EVs) and mechanically induced Membrane Vesicles (MVs)—to determine which is superior for structural studies of the receptor HER2.

2. Methodology

The study utilized the human breast cancer cell line SKBR3, which naturally overexpresses HER2.

• Vesicle Production:
  • EVs: Harvested from serum-free spent media after 14 hours of culture (naturally shed).
  • MVs: Generated by mechanically forcing cells through a syringe (extrusion), creating plasma membrane vesicles.
• Affinity Purification Strategy:
  • To ensure the correct orientation ("outside-out") and enrichment of HER2, the authors developed a purification protocol using DARPins (Designed Ankyrin Repeat Proteins).
  • A specific DARPin (G3) targeting the extracellular domain (ECD) of HER2 was biotinylated and immobilized on streptavidin-coated magnetic beads.
  • Vesicles were captured, washed, and eluted using 3C protease cleavage to release the vesicles without denaturing them.
• Characterization Techniques:
  • Cryo-Electron Tomography (Cryo-ET): Used to analyze vesicle morphology, internal density, and protein distribution.
  • Mass Spectrometry (MS): Performed on low-sample loads to quantify proteome composition and compare protein abundance between EVs and MVs.
  • Single Particle Cryo-EM (SPA): Applied to purified MVs to attempt high-resolution reconstruction of HER2.
  • Computational Analysis: Used AlphaFold2 for structural prediction, DeepLoc for subcellular localization prediction, and correlation coefficient (CC) analysis to distinguish HER2 from other abundant membrane proteins.

3. Key Contributions

• Comparative Analysis of Vesicle Types: The study provides the first direct comparison of EVs and MVs specifically for the purpose of in situ membrane protein structural determination.
• Novel Purification Protocol: Development of a DARPin-based affinity purification method that enriches HER2-containing vesicles while maintaining native orientation and allowing for gentle elution.
• Workflow for Native Membrane SPA: A computational workflow combining particle picking, 2D/3D classification, and AlphaFold-guided filtering to isolate specific receptors from the "noise" of a complex native membrane proteome.

4. Key Results

A. Morphological and Compositional Differences

• Heterogeneity: EVs were highly heterogeneous in size and shape (classified into 9 groups, including elongated, dense, and multivesicular types). In contrast, MVs were significantly more homogeneous, with 77.5% appearing as round, low inner-density vesicles.
• Proteome Composition:
  • EVs: Enriched in cytoskeletal proteins, Rab proteins, and exosomal markers. They contained higher internal density, likely due to encapsulated macromolecules.
  • MVs: Enriched in ribosomal proteins, proteasome components, and glycolysis enzymes. They had lower internal density, making them more electron-transparent.
• HER2 Abundance: Both vesicle types contained high levels of HER2, but the morphological heterogeneity and high internal density of EVs made them less suitable for high-resolution imaging.

B. Structural Determination of HER2

• Cryo-ET Limitations: Sub-tomogram averaging failed to yield a high-resolution structure of HER2, likely due to the receptor's low molecular weight (<200 kDa), conformational flexibility, and the high contrast of the membrane.
• Single Particle Analysis (SPA) Success:
  • Using over 100,000 micrographs of MVs, the authors applied a rigorous classification workflow.
  • They successfully reconstructed a 3D map of the HER2 extracellular domain (ECD) at ~8.7 Å resolution.
  • The structure clearly matched the known shape of HER2 subdomains I-III.
  • Intracellular Domain (ICD): The ICD density was weak or absent, confirming the high flexibility of the linker between the transmembrane helix and the intracellular tail in the native membrane.
• Specificity Validation: By comparing the experimental density map against AlphaFold2 predictions of the top 34 most abundant membrane proteins, the authors confirmed that the reconstructed density correlated best with HER2 (CC = 0.724) compared to other candidates (e.g., NCLN, CC = 0.720 but present in very low copy numbers).

5. Significance and Conclusion

• MVs are Superior for Structural Biology: The study concludes that Mechanically Induced Vesicles (MVs) are a more suitable platform than naturally shed EVs for determining the structure of membrane proteins in their native environment. Their morphological homogeneity and lower internal density reduce background noise, facilitating better contrast for cryo-EM.
• Proof of Concept: The research demonstrates that it is possible to resolve the structure of a flexible, single-pass transmembrane receptor (HER2) directly from a native plasma membrane environment, albeit at modest resolution.
• Future Directions: The authors suggest that while current resolution is limited by the flexibility of the receptor and the thickness of the ice (due to large vesicle size), future advancements in cryo-EM labeling technologies and image processing will enable higher-resolution reconstruction of individual receptors within cellular contexts.

In summary, this paper establishes a robust pipeline for isolating and structurally analyzing membrane proteins in native membranes, identifying mechanically induced vesicles as the preferred substrate for such studies.