Phenotypic variation within and across transcriptomic cell types in mouse motor cortex

Federico Scala, Dmitry Kobak, Matteo Bernabucci, Yves Bernaerts, Cathryn R. Cadwell, Jesus Ramon Castro, Leonard Hartmanis, Xiaolong Jiang, Sophie Laturnus, Elanine Miranda, Shalaka Mulherkar, Zheng Huan Tan, Zizhen Yao, Hongkui Zeng, Rickard Sandberg, Philipp Berens, Andreas S. ToliasbioRxiv 2020 · DOI

TL;DR

This study tackles the long-standing challenge of unifying neuronal classification by integrating gene expression, morphology, and electrical activity from over 1300 mouse motor cortex neurons. They reveal that while broad neuronal families are distinct, individual cell types within these families exhibit a continuous spectrum of properties, suggesting a more fluid view of neural diversity than previously thought.

What Problem Does This Paper Address?

For decades, neuroscientists have grappled with how to best classify the staggering diversity of neurons in the brain. Traditionally, this classification relied on either morphology (shape) and electrical properties (how they fire) or on gene expression patterns. However, it's been technically difficult to measure all these properties simultaneously from the same individual neuron. This has led to separate taxonomies that don't always align, creating a "missing link" problem. This paper aims to bridge this gap by providing a comprehensive, multi-modal characterization of neuronal types in the mouse motor cortex, offering a more unified understanding of neuronal diversity.

Methods

The core of this study is the Patch-seq technique, a powerful method that allows researchers to perform three key measurements on a single neuron:

  1. Electrophysiology: Patch-clamp recordings to measure how the neuron fires in response to electrical stimulation.
  2. Morphology: Biocytin staining and subsequent reconstruction to visualize the neuron's intricate 3D structure (dendrites and axons).
  3. Transcriptomics: Single-cell RNA sequencing (scRNA-seq) to capture the neuron's gene expression profile.

They applied this to over 1300 neurons in the adult mouse motor cortex. To analyze the vast amount of data, they employed:

  • t-SNE (t-distributed Stochastic Neighbor Embedding): A dimensionality reduction technique used to visualize high-dimensional data (like gene expression or electrophysiological features) in a 2D space, revealing clusters and relationships between cells.
  • Sparse Reduced-Rank Regression (RRR): A statistical method to predict electrophysiological features from gene expression, identifying key genes involved in neuronal function.
  • k-Nearest Neighbors (kNN) classifier: Used to assess how well cells could be classified into transcriptomic families based on their different modalities.

Datasets

The study primarily uses data generated from their own Patch-seq experiments on mouse motor cortex neurons. They also leverage existing reference datasets for transcriptomic cell types, specifically from the BICCN consortium (Yao et al., in preparation) and Tasic et al. (2018), to assign their Patch-seq cells to known transcriptomic types.

Key Insights and Ideas

  • Families are distinct, but types are continuous: While broad categories of neurons (like excitatory neurons, or specific interneuron families such as Vip, Pvalb, Sst) have clearly different electrical and morphological signatures, the individual transcriptomic subtypes within these families often blend into each other.
  • A "parts list" is more like a spectrum: The idea of discrete, perfectly defined neuronal "types" might be an oversimplification. Instead, neurons within a family can exist on a continuum, with neighboring transcriptomic types sharing similar morpho-electric features.
  • Transcriptomic neighborhood predicts phenotype: The study found a strong correlation between how similar neurons are at the transcriptomic level and how similar their electrical and morphological properties are, especially within families.
  • Patch-seq is a game-changer: This work highlights the power of Patch-seq to integrate multiple data modalities, providing a much richer understanding of neuronal diversity than single-modality approaches.

Key Results

  • The study successfully characterized the morpho-electric phenotypes of nearly all 90 transcriptomically defined neural types in the mouse motor cortex.
  • They found that transcriptomic families were well-separated in morpho-electric space, but individual transcriptomic types within families showed continuous variation.
  • The correlation between transcriptomic distance and electrophysiological distance within families was high (r = 0.70 when pooled across all families).
  • The study identified specific genes that are predictive of electrophysiological properties, offering insights into the genetic basis of neuronal function.
  • They observed that some t-types exhibited variability that could be partially explained by underlying transcriptomic variability, further supporting the idea of continuous phenotypic landscapes.

Limitations and Future Work

  • Sparse coverage for some t-types: The authors acknowledge that some transcriptomic types were sparsely sampled, and suggest that future experiments using more specific Cre driver lines could improve coverage.
  • Morphological recovery challenges: Recovering complete morphologies, especially for fine axonal branches of interneurons, was challenging due to the slicing procedure and RNA extraction process. This might have led to missed morphological variability.
  • Need for theoretical frameworks: The authors emphasize the need for theoretical work on how to conceptualize and model this hierarchical, discrete-continuous cell variability.
  • Future directions: They suggest future studies should incorporate additional modalities like long-range projections, local connectivity, and in vivo functional characterization to build an even more complete picture of neuronal diversity. They also note that their analysis code and data will be made publicly available.

Summary updated

Abstract

Cortical neurons exhibit astounding diversity in gene expression as well as in morphological and electrophysiological properties. Most existing neural taxonomies are based on either transcriptomic or morpho-electric criteria, as it has been technically challenging to study both aspects of neuronal diversity in the same set of cells. Here we used Patch-seq to combine patch-clamp recording, biocytin staining, and single-cell RNA sequencing of over 1300 neurons in adult mouse motor cortex, providing a comprehensive morpho-electric annotation of almost all transcriptomically defined neural cell types. We found that, although broad families of transcriptomic types ( Vip , Pvalb , Sst , etc.) had distinct and essentially non-overlapping morpho-electric phenotypes, individual transcriptomic types within the same family were not well-separated in the morpho-electric space. Instead, there was a continuum of variability in morphology and electrophysiology, with neighbouring transcriptomic cell types showing similar morpho-electric features, often without clear boundaries between them. Our results suggest that neural types in the neocortex do not always form discrete entities. Instead, neurons follow a hierarchy consisting of distinct non-overlapping branches at the level of families, but can form continuous and correlated transcriptomic and morpho-electrical landscapes within families.
BibTeX
@misc{scala2020phenotypic,
  author = {Federico Scala and Dmitry Kobak and Matteo Bernabucci and Yves Bernaerts and Cathryn R. Cadwell and Jesus Ramon Castro and Leonard Hartmanis and Xiaolong Jiang and Sophie Laturnus and Elanine Miranda and Shalaka Mulherkar and Zheng Huan Tan and Zizhen Yao and Hongkui Zeng and Rickard Sandberg and Philipp Berens and Andreas S. Tolias},
  title = {{Phenotypic variation within and across transcriptomic cell types in mouse motor cortex}},
  year = {2020},
  doi = {10.1101/2020.02.03.929158},
  howpublished = {bioRxiv},
}

Loading related content...

Figure