Coordinated Head Direction Representations in Mouse Anterodorsal Thalamic Nucleus and Retrosplenial Cortex

Marie-Sophie H. van der Goes, Jakob Voigts, Jonathan P. Newman, Enrique H.s. Toloza, Norma J. Brown, Pranav Murugan, Mark T. HarnettbioRxiv 2022 · DOI

TL;DR

This study reveals that the mouse's sense of direction (head direction, HD) is remarkably coordinated between the thalamus (ADn) and the retrosplenial cortex (RSC), even when visual cues are removed. The findings suggest a strong feedforward flow of HD information from ADn to RSC, challenging previous ideas about how visual cues update our internal compass.

What Problem Does This Paper Address?

Navigating our environment relies on a robust internal sense of direction, often called the "head direction" (HD) system. This system integrates self-motion cues with external sensory information, like visual landmarks, to maintain our orientation. While we know the brain regions involved in HD coding (like the anterodorsal thalamus, ADn, which relays this information to the cortex), how visual cues precisely update and anchor this internal compass remains a puzzle. Specifically, researchers are keen to understand if visual information is first processed in the cortex (like the retrosplenial cortex, RSC) and then relayed to subcortical areas, or if there's a more coordinated update across these regions. This paper aims to untangle this by investigating the temporal dynamics and connectivity of HD representations in ADn and RSC during visual cue manipulations.

Methods

The researchers employed simultaneous tetrode recordings from the ADn and RSC in freely moving mice. They used a dark circular arena with a single prominent LED cue that could be rotated. This allowed them to observe how HD representations in both regions changed in response to visual cue rotations and in darkness. They also used advanced techniques like:

  • Simultaneous Two-Area Tetrode Recording: This is the core method, allowing for direct comparison of neural activity in ADn and RSC at high temporal resolution.
  • Decoding HD: They developed a decoding strategy using a linear-Gaussian generalized model (GLM) to infer HD from population neural activity, providing a more continuous measure of orientation than traditional cell-based analyses.
  • Spike Cross-Correlation: This technique was used to identify putative monosynaptic connections between neurons in ADn and RSC, revealing the directionality and strength of their functional connectivity.
  • Retrograde Monosynaptic Rabies Tracing: This anatomical technique was used to map the direct connections between ADn and RSC.

Datasets

Not explicitly discussed as training/evaluation datasets in the traditional machine learning sense. The study generated its own electrophysiological and behavioral data from mice during controlled experiments.

Key Insights and Ideas

  • Synchronized HD Updates: ADn and RSC update their HD representations in near-perfect synchrony when a visual cue is rotated, suggesting a coordinated rather than sequential processing.
  • Darkness Doesn't Break Coordination: This synchronization persists even when the visual cue is turned off, indicating that internal mechanisms, not just external cues, maintain HD coordination.
  • ADn to RSC Feedforward Dominance: Anatomical and functional connectivity data strongly suggest a dominant feedforward pathway from ADn to RSC, with minimal reciprocal connections from RSC back to ADn. This challenges the idea that RSC is the primary hub for integrating visual information to update HD.
  • RSC's Role: While not the primary driver of visual updates, RSC's HD code is influenced by ADn and likely integrates this information with other sensory inputs to contribute to spatial representations.

Key Results

  • HD Cell Proportion: 48% of recorded cells in ADn and 20% in RSC met the criteria for HD cells.
  • Cue Rotation Response: ADn and RSC HD ensembles showed highly correlated responses to cue rotations (correlation coefficient=0.52), indicating a synchronized shift in their HD representations.
  • Darkness Coordination: Even in darkness, ADn and RSC decoded HD errors were correlated (r=0.333 for cue-on, r=0.311 for cue-off), demonstrating sustained coordination.
  • Connectivity Asymmetry: Monosynaptic connections were found to be overwhelmingly from ADn to RSC (6.4% of pairs) compared to RSC to ADn (0.08% of pairs).
  • Decoding Accuracy: The GLM decoder achieved an average error of ~60° in ADn and ~72° in RSC, with performance dropping at higher angular velocities.

Limitations and Future Work

  • Angular Velocity Influence: While the study controlled for angular velocity, its precise role in the observed coordination and potential differences between regions warrants further investigation.
  • RSC's Complex Receptive Fields: The paper notes that RSC neurons have more complex, multimodal receptive fields, suggesting they integrate more than just HD information. Future work could explore how these complex fields interact with HD coding.
  • Cerebellar Input: The authors mention that cerebellar inputs might mediate HD coordination in darkness, a hypothesis that could be directly tested in future studies.
  • Specific Circuit Mechanisms: While the feedforward ADn-to-RSC pathway is identified, the precise circuit mechanisms by which RSC integrates this information with other sensory modalities for spatial cognition remain to be fully elucidated.
  • Landmark Coding Cells: The authors suggest future experiments could investigate dedicated "landmark-coding cells" in RSC, which might play a role in visual anchoring.

Summary updated

Abstract

Abstract The sense of direction is critical for survival in changing environments and relies on flexibly integrating self-motion signals with external sensory cues. While the anatomical substrates involved in head direction (HD) coding are well known, the mechanisms by which visual information updates HD representations remain poorly understood. Retrosplenial cortex (RSC) plays a key role in forming coherent representations of space in mammals and it encodes a variety of navigational variables, including HD. Here, we use simultaneous two-area tetrode recording to show that RSC HD representation is nearly synchronous with that of the anterodorsal nucleus of thalamus (ADn), the obligatory thalamic relay of HD to cortex, during rotation of a prominent visual cue. Moreover, coordination of HD representations in the two regions is maintained during darkness. We further show that anatomical and functional connectivity are consistent with a strong feedforward drive of HD information from ADn to RSC, with surprisingly little reciprocal drive in the corticothalamic direction. Together, our results provide direct evidence for a concerted global HD reference update across cortex and thalamus, and establish the underlying functional connectivity that supports this coordination.
BibTeX
@misc{goes2022coordinated,
  author = {Marie-Sophie H. van der Goes and Jakob Voigts and Jonathan P. Newman and Enrique H.S. Toloza and Norma J. Brown and Pranav Murugan and Mark T. Harnett},
  title = {{Coordinated Head Direction Representations in Mouse Anterodorsal Thalamic Nucleus and Retrosplenial Cortex}},
  year = {2022},
  doi = {10.1101/2022.08.20.504604},
  howpublished = {bioRxiv},
}
Figure