Quantitative Spatial Analysis of Chromatin Biomolecular Condensates Using Cryoelectron Tomography

Academic Background

Biomolecular condensates are membrane-less organelles formed through liquid-liquid phase separation (LLPS) within cells, playing critical roles in key biological processes such as gene expression and signal transduction. However, due to limitations in traditional imaging techniques, high-resolution structural information about the interior of condensates has long been lacking, hindering a deeper understanding of their functional mechanisms. Chromatin, the primary organizational form of genetic material in eukaryotic nuclei, undergoes dynamic condensation and decondensation that directly influences gene regulation. Yet, the fine structure and molecular arrangement principles of chromatin condensates remain unclear.

This study, led by Michael K. Rosen’s team, aimed to address the following key questions:
1. How to overcome the structural disruption of liquid-like condensates caused by traditional cryo-EM sample preparation techniques
2. How to achieve precise localization and orientation analysis of individual nucleosomes within high-density condensates
3. Comparing the molecular organization of in vitro reconstituted chromatin condensates with native chromatin

Source of the Paper

• Authors: Huabin Zhou (first author), Joshua Hutchings, and 13 collaborators from the University of Texas Southwestern Medical Center, HHMI Janelia Research Campus, and other institutions
  
• Corresponding Authors: Elizabeth Villa and Michael K. Rosen
  
• Journal: Proceedings of the National Academy of Sciences (PNAS)
  
• Publication Date: May 6, 2025
  
• DOI: 10.1073/pnas.2426449122
  

---

Research Workflow and Methodological Innovations

1. Breakthroughs in Sample Preparation

Limitations of Traditional Methods

• Key Findings: Conventional blotting and self-wicking methods led to:
  
  • Distorted condensate morphology (from spherical to flattened structures)
    
  • Orientation bias of nucleosomes at the air-water interface (AWI) (70% parallel to the interface)
    
  • Chromatin disassembly, resulting in exposed DNA (Figure 1)
    

High-Pressure Freezing–Focused Ion Beam (HPF-FIB) Integration

• Innovative Approach:
  
  • High-Pressure Freezing (HPF): Preserved spherical condensates intact within 25 μm ice layers
    
  • Correlative Fluorescence Localization: Identified milling regions using Alexa Fluor 594 labeling
    
  • Cryo-FIB Milling: Prepared 80–150 nm lamellae using the “waffle method” (Figure 2)
    
• Advantages:
  
  • Maintained physiological nucleosome density (intermediate between sparse blotting and over-packed self-wicking)
    
  • Random nucleosome orientation (consistent with isotropic liquid properties)
    

2. Development of Image Analysis Algorithms

Pitfalls of Traditional Template Matching

• Achieved only an F1 score of 0.76 (position + orientation) in simulated data
  
• Missing wedge effects caused resolution loss along the z-axis
  

Context-Aware Template Matching (CATM) Algorithm

• Two-Stage Workflow (Figure 3):
  
  1. Deep Learning Localization: Initial centroid positioning using DeepFinder
     
  2. Local Template Optimization:
     
     • Retained multiple candidate templates (CCC > 0.3)
       
     • Incorporated steric exclusion to resolve particle overlaps
       
     • Optimized neighboring particle pairs via Monte Carlo sampling
       
• Performance Improvements:
  
  • Achieved F1 scores of 0.99 (position) and 0.96 (orientation) in simulated data
    
  • Nucleosome orientation distribution perfectly matched theoretical randomness
    

3. Multi-Scale Structural Analysis

In Vitro Reconstituted System (12-Nucleosome Arrays)

• Resolution Milestone:
  
  • Obtained 6.1 Å resolution structure from 126,126 particle averages (Figure 4c)
    
  • First observation of nucleosome-nucleosome interaction interfaces
    
• Network Analysis:
  
  • Valency distribution entropy of 0.74 ± 0.02, indicating heterogeneous organization
    
  • DBSCAN clustering revealed small clusters of 4–20 nucleosomes (Figure 5g)
    
  • Surface nucleosomes were 37% less likely to form clusters than interior ones (Figure 5i)
    

Native Chromatin (Hela Nuclei & NIH3T3 Cells)

• Structural Features:
  
  • Achieved 12 Å resolution nucleosome structure from Hela nuclei (35,503 particles)
    
  • Identified two nucleosome classes in NIH3T3 cells (12 Å and 22 Å), the latter potentially containing linker histone H1 (Figure 6g)
    
• Conserved Findings:
  
  • Valency distribution entropy of 0.77 ± 0.01, highly similar to in vitro systems
    
  • Chromatin fiber length differences did not affect local packing patterns
    

---

Research Conclusions and Impact

Scientific Significance

1. Methodological Contributions:

   • Established an HPF-FIB-CATM pipeline, setting a paradigm for structural studies of liquid biomolecular condensates
     
   • CATM algorithm is generalizable to other condensates with large components (e.g., centrosomes, transcription factories)
     

2. Chromatin Biology Insights:

   • Revealed inherent heterogeneity in nucleosome networks independent of fiber length
     
   • Surface tension mechanism: Unsaturated valency at interfaces drives inward cohesion (Figure 5i)
     

3. Disease Relevance:

   • Provides structural foundations for diseases involving aberrant chromatin condensation (e.g., cancer, neurodegeneration)
     

Technical Highlights

• Sample Prep Innovation: First native-state cryo-preservation of liquid condensates
  
• Algorithm Breakthrough: CATM solves particle assignment challenges in high-density conditions
  
• Resolution Record: Achieved 6.1 Å structure of chromatin condensates in situ
  

Applications

• Drug Development: Screening small molecules targeting nucleosome interaction interfaces
  
• Synthetic Biology: Rational design of artificial chromatin condensates
  
• Cryo-EM Technology: Open-sourced DeepFinder-CATM pipeline (GitLab)