Organizing Principles of Astrocytic Nanoarchitecture in the Mouse Cerebral Cortex

Christopher Salmon, Tabish A. Syed, J. Benjamin Kacerovsky, Nensi Alivodej, Alexandra L. Schober, Michael T. Pratte, Michael P. Rosen, Miranda Green, Adario DasGupta, Hojatollah Vali, Craig A. Mandato, Kaleem Siddiqi, Keith K. MuraibioRxiv 2021 · DOI

TL;DR

This paper leverages high-resolution electron microscopy and advanced computational analysis to reveal the intricate nanoarchitecture of astrocytes in the mouse cerebral cortex. It demonstrates that astrocytes are not haphazardly branched but rather organized into specific structural motifs and hierarchical assemblies, offering new insights into their functional adaptations.

What Problem Does This Paper Address?

Astrocytes, crucial for brain health and function, possess a complex and finely branched structure that has historically been difficult to fully characterize. While their broad roles are known, the fundamental building blocks and organizing principles of their nanoarchitecture remained undefined. This knowledge gap stems from the lack of high-resolution datasets capable of resolving nanoscale features and the absence of computational tools to effectively analyze this intricate morphology. Understanding this detailed structure is essential for attributing specific astrocytic functions to their anatomy and for comprehending how they integrate into neural circuits.

Methods

The researchers developed an experimental-computational pipeline to tackle astrocytic nanoarchitecture. This involved:

  • High-Resolution Electron Microscopy: Utilizing focused ion beam scanning electron microscopy (FIB-SEM) to generate detailed 3D ultrastructural datasets of layer 2/3 mouse somatosensory cortex at an 8nm sectioning interval and ~4.13nm axial resolution.
  • Computer Vision and Shape Analysis: Developing novel computational approaches to analyze astrocyte geometry, complexity, and interactions. This included:
  • Defining a "protrusiveness" measure for astrocytic surfaces.
  • Computing local thickness using maximal inscribed spheres.
  • Combining these measures to create an "M-score" for classifying astrocytic parts into "cores," "expansions," and "constrictions."
    • Constructing directed acyclic graphs to analyze the hierarchical organization of these parts.
  • Synaptic Analysis: Segmenting postsynaptic densities (PSDs) and analyzing their spatial relationships with astrocytic structures and mitochondria.
  • Clustering Algorithms: Employing Dominant Sets (DSet) clustering to identify groups of synapses ("synaptic clusters") around astrocytes.
  • Distance Measurements: Using fast-marching methods to compute geodesic distances between various cellular components (PSDs, mitochondria, astrocyte surfaces).

Datasets

The study utilized high-resolution FIB-SEM datasets from the layer 2/3 somatosensory cortex of adult mice. Specifically, three reference volumes were analyzed, with voxel sizes of 4.13 x 4.13 x 8nm. These datasets were made publicly available for further research.

Key Insights and Ideas

  • Beyond the "Sponge": Contrary to the long-held "sponge-like" or "spongiform" view, astrocytes are not characterized by frequent loops and holes. Instead, their structure is more akin to a "thorny bush" or "shrub" with shallow but wide networks of branches.
  • Hierarchical Organization: Astrocytes are organized into distinct subcompartments (cores, expansions, constrictions) that form a hierarchical network, revealing a more structured arrangement than previously appreciated.
  • Synaptic Clustering: Synapses are not randomly distributed around astrocytes but rather organize into discrete clusters, suggesting that astrocytes actively partition synaptic space.
  • Mitochondrial Positioning: Astrocytic mitochondria are not uniformly distributed but are strategically positioned closer to larger synaptic clusters, implying a functional relationship between energy supply and synaptic activity.
  • Specialized Organelles: Endfeet exhibit specialized mitochondrial and endoplasmic reticulum (ER) structures, suggesting adaptations for their unique role in gliovascular coupling and metabolic support.

Key Results

  • Astrocytic processes were successfully decomposed into three main part types: cores, expansions, and constrictions, with constrictions being prevalent at branch tips and expansions being predominantly embedded.
  • A directed acyclic graph analysis revealed a shallow but wide network of astrocytic parts, with a few "hub" parts connecting to up to 45 other components.
  • Looping structures were found to be scarce, contradicting the "sponge-like" morphology.
  • Analysis of tripartite synapses showed that PSD landing points were significantly more likely to be found on embedded parts and constrictions, though expansions also received landing points.
  • Mitochondria in endfeet displayed distinct morphologies (flattened, disk-like) compared to non-endfoot regions (tubular), with differences in surface area-to-volume ratio, curvature, and thickness.
  • Endfoot ER showed a higher proportion of sheet-like cisterns compared to the tubular networks found in non-endfoot regions.
  • Mitochondria-ER contact sites (MERCs) were significantly larger and more abundant in endfeet compared to non-endfoot regions.
  • Synaptic clusters were found to be inversely correlated with the mean distance to mitochondria, indicating that larger clusters are positioned closer to astrocytic mitochondria.

Limitations and Future Work

  • Subregion Analysis: The study focused on subregions of astrocytes, limiting the ability to apply computational analysis to entire cells. Future work could explore methods for reconstructing larger volumes of whole astrocytes.
  • Technical Challenges: Automated segmentation of thin astrocytic processes remains a significant challenge. While progress has been made with FIB-SEM, limitations in field of view and section thickness for larger volumes persist.
  • Developmental and In Vivo Studies: The authors suggest future research should include developmental studies, chronic in vivo imaging, and new genetic lines to investigate how astrocytes organize synaptic clusters and how mitochondria adjust their positioning in response to activity.
  • Functional Integration: Integrating these detailed structural datasets with biophysical modeling environments is expected to improve our understanding of astrocyte function and their role in neural circuits.

Summary updated

Abstract

SUMMARY Astrocytes have complex roles in central nervous system (CNS) health and disease. Underlying these roles is an elaborate architecture based on frequent, extremely fine, but seemingly haphazard branches, as well as prominent features including tripartite synaptic complexes and perivascular endfeet. While broad categories of structures in astrocytes are known, the fundamental building blocks that compose them and their organizing principles have yet to be adequately defined. This is largely due to the absence of high-resolution datasets that can reveal nanoscopic features of astrocytes (i.e. 10-20nm diameter in x, y, and z) and a lack of computational approaches that can effectively interrogate astrocyte shape, organization, and nanoarchitecture. Here, we produced and analyzed multiple, high-resolution datasets of layer 2/3 mouse somatosensory cortex using focused ion beam scanning electron microscopy (8nm intervals) and computer vision approaches to provide a principled, quantitative analysis of astrocytic nanoarchitecture. A decomposition of astrocytes into fundamental ‘parts’ led to the discovery of unique structural components, recurring structural motifs, and assembly of parts into an organized hierarchy. New relationships were also discerned between astrocytic processes and other CNS microanatomy including mitochondria, tripartite synapses, and cerebrovasculature. By deploying computational resources to quantitatively understand the organizing principles and nanoarchitecture of astrocytes, this study reveals the specialized anatomical adaptations of these complex cells within the CNS. One Sentence Summary Using high-resolution serial electron microscopy datasets and computer vision, this study provides a systematic analysis of astrocytic nanoarchitecture from multiple samples of layer 2/3 of adult mouse neocortex, and presents quantitative evidence that astrocytes organize their morphology into purposeful, classifiable assemblies with unique structural and subcellular organelle adaptations related to their physiological functions.
BibTeX
@misc{salmon2021organizing,
  author = {Christopher Salmon and Tabish A. Syed and J. Benjamin Kacerovsky and Nensi Alivodej and Alexandra L. Schober and Michael T. Pratte and Michael P. Rosen and Miranda Green and Adario DasGupta and Hojatollah Vali and Craig A. Mandato and Kaleem Siddiqi and Keith K. Murai},
  title = {{Organizing Principles of Astrocytic Nanoarchitecture in the Mouse Cerebral Cortex}},
  year = {2021},
  doi = {10.1101/2021.11.05.467391},
  howpublished = {bioRxiv},
}
Figure