A single spectrum of neuronal identities across thalamus

James Phillips, Anton Schulmann, Erina Hara, Chenghao Liu, Lihua Wang, Brenda C. Shields, Wyatt Korff, Andrew L. Lemire, Joshua T. Dudman, Sacha B. Nelson, Adam W. HantmanbioRxiv 2017 · DOI

TL;DR

This study reveals that the thalamus, a crucial relay station in the brain, isn't organized into discrete functional modules as previously thought. Instead, it operates as a continuous "filter bank" with neuronal identities varying along a single spectrum, organized topographically along the mediolateral axis. This finding offers a simpler, more unified model for how the thalamus processes diverse information for the forebrain.

What Problem Does This Paper Address?

The thalamus is a complex structure in the brain, acting as a central hub for information flow from sensory systems, motor pathways, and the cortex. It comprises over 30 distinct nuclei, each projecting to specific areas of the forebrain. A long-standing question in neuroscience is whether there's a conserved organizational principle underlying these diverse thalamic pathways, or if each pathway is unique. Previous models often categorized thalamic nuclei into "core" and "matrix" types, but this dichotomy didn't fully capture the complexity and diversity of thalamic organization across different modalities (sensory, motor, cognitive, etc.). This paper aims to uncover a unifying organizational logic for thalamic pathways that applies across all major cortical modalities.

Methods

The researchers employed a multi-pronged approach:

  • Transcriptomic Atlas: They generated a comprehensive transcriptomic atlas of the thalamus by retrogradely labeling neurons from various forebrain projection targets. This involved microdissecting and performing RNA sequencing on cells from 22 thalamic nuclei across 120 samples.
  • Hierarchical Clustering: They used hierarchical clustering on the transcriptomes of these nuclei to identify relationships and group them based on gene expression patterns.
  • Principal Component Analysis (PCA): PCA was used to analyze gene expression differences and identify major axes of variance within the thalamus.
  • Single-Cell RNA Sequencing (scRNA-seq): To investigate neuronal diversity within nuclei, they performed scRNA-seq on individual projection neurons from motor, somatosensory, visual, auditory, and prefrontal projection classes.
  • Electrophysiology: Whole-cell recordings were performed on neurons from key thalamic nuclei representing different profiles to assess their functional properties.
  • Multi-Color Fluorescent In Situ Hybridization (Multi-FISH): This technique was used to spatially map the expression of specific marker genes within thalamic nuclei and identify intermediate cell populations.

Datasets

The study generated and analyzed novel transcriptomic datasets from mouse thalamus.

  • Pooled-cell RNAseq data: From 22 thalamic nuclei across 120 samples.
  • Single-cell RNAseq data: From individual projection neurons across motor, somatosensory, visual, auditory, and prefrontal projection classes.

Key Insights and Ideas

  • A Spectrum, Not Discrete Classes: The most striking finding is that thalamic nuclei don't form discrete, functionally distinct classes. Instead, their identities lie on a continuous spectrum, organized along a single axis of genetic variance.
  • Topographic Organization: This spectrum of neuronal identities is strongly aligned with the mediolateral axis of the thalamus, suggesting a fundamental link between genetic programming and spatial organization.
  • "Filter Bank" Model: The thalamus functions more like a "filter bank" that repeatedly applies a set of molecularly defined filters across different modalities, rather than a simple switchboard routing distinct information types.
  • Conserved Logic Across Modalities: This organizational principle isn't limited to sensory pathways but extends to motor, limbic, and cognitive systems, revealing a unifying architecture for thalamic processing.
  • Intermediate Cells: Single-cell analysis revealed the presence of "intermediate" cells that bridge these profiles, further supporting the idea of a continuous spectrum rather than sharp boundaries.

Key Results

  • Hierarchical clustering of transcriptomic data identified three major profiles of thalamic nuclei, which were best distinguished by their anatomical position along the mediolateral axis.
  • PCA revealed a primary axis of variance (PC1) that explained 38% of the gene expression differences and strongly correlated with the mediolateral position. This axis was enriched in genes encoding receptors and ion channels.
  • Electrophysiological recordings confirmed a systematic gradient of neuronal firing properties across these profiles, with neurons in different profiles exhibiting distinct action potential characteristics.
  • Single-cell RNA-seq data showed that individual neurons within nuclei also exhibit this spectral organization, with intermediate cells found at nuclear boundaries.
  • Multi-FISH experiments demonstrated that genes marking different profiles are spatially organized within nuclei, with some cells co-expressing markers, indicating a graded rather than discrete organization.

Limitations and Future Work

  • "Near-Comprehensive" Atlas: While extensive, the atlas is described as "near-comprehensive," suggesting that some nuclei or cell types might not have been fully captured.
  • Single-cell vs. Pooled-cell Matching: The authors note that pooled-cell and single-cell dissections were not precisely matched, though they found close agreement between the datasets.
  • Future Work: The authors suggest that future work should focus on understanding how this pattern of molecular variability intersects with input modalities and behavioral relevance. Investigating the functional implications of intermediate cell types and their potential for hybrid input-output transforms is also a promising direction. The study also opens avenues for exploring how this "filter bank" architecture is dynamically regulated and how it contributes to complex cognitive functions.

Summary updated

Abstract

ABSTRACT Uncovering common principles by which diverse modalities of information are processed is a fundamental goal in neuroscience. In mammalian brain, thalamus is the central processing station for inputs from sensory systems, subcortical motor systems, and cortex; a function subserved by over 30 defined nuclei 1,2 . Multiple thalamic nuclei send convergent information to each region of the forebrain, but whether there is a conserved architecture across the set of thalamic pathways projecting to each forebrain area has remained unresolved 3–5 . To uncover organizational principles of thalamic pathways, we produced a near-comprehensive transcriptomic atlas of thalamus. This revealed a common logic for thalamic nuclei serving all major cortical modalities. We found that almost all nuclei belong to one of three major profiles, with a given cortical area getting input from each of these profiles. These profiles lie on a single axis of variance aligned with the mediolateral axis of thalamus, and this axis is strongly enriched in genes encoding receptors and ion channels. We further show that each projection profile exhibits different electrophysiological signatures. Single-cell profiling revealed that rather than forming discrete classes, thalamic neurons lie on a spectrum, with intermediate cells existing between profiles. Thus, in contrast to canonical models of thalamus that suggest it is a switchboard primarily concerned with routing distinct modalities of information to distinct cortical regions, we show that the thalamocortical system is more akin to a molecularly-defined ‘filter bank’ repeatedly applied across modality. Together, we reveal striking covariation in the organization of thalamic pathways serving all input modalities and output targets, establishing a simple and comprehensive thalamic functional architecture.
BibTeX
@misc{phillips2017single,
  author = {James Phillips and Anton Schulmann and Erina Hara and Chenghao Liu and Lihua Wang and Brenda C. Shields and Wyatt Korff and Andrew L. Lemire and Joshua T. Dudman and Sacha B. Nelson and Adam W. Hantman},
  title = {{A single spectrum of neuronal identities across thalamus}},
  year = {2017},
  doi = {10.1101/241315},
  howpublished = {bioRxiv},
}
Figure