Multisensory integration in pyramidal cell dendrites in parietal cortex of awake behaving mice

RWTH Publications (RWTH Aachen)

Abstract

The brain's remarkable ability to integrate information from multiple sensory modalities is known as multisensory integration (MSI). MSI is fundamental for creating a coherent perception of the world and guiding adaptive behavior. While the behavioral advantages of MSI, such as enhanced detection and discrimination, are well established, the underlying neural computations, particularly at the cellular and subcellular levels within the neocortex, remain largely unresolved. This thesis investigated the anatomical substrates and functional mechanisms of visuotactile integration in the mouse brain, focusing on the rostrolateral visual area (RL), a region implicated as a parietal association hub. We specifically examined the role of Layer 2/3 (L2/3) pyramidal neurons and their dendrites, hypothesizing that dendritic computations are critical for implementing MSI and are subject to modification through learning. Functionally, we investigated neural correlates of MSI using behavioral paradigms combined with chronic widefield calcium imaging. Mice trained on a visuotactile evidence accumulation task exhibited significant behavioral benefits during multisensory trials, demonstrating enhanced perceptual sensitivity compared to unisensory conditions. Initial population-level imaging using widefield calcium recordings confirmed RL as a key integration hub, displaying enhanced, often supra-linear, population activity during multisensory trials that correlated with improved behavioral performance. To delineate the circuitry supporting visuotactile integration in RL, we employed viral tracing techniques. Retrograde tracing confirmed that RL receives convergent input from primary visual (V1) and somatosensory (S1) cortices, as well as thalamic nuclei. Quantitative analysis revealed the highest density of projecting somata originated fromV1, followed by S1, with significantly lower densities from thalamic sources, highlighting the importance of direct cortico-cortical pathways. Dual anterograde tracing provided crucial insights into the laminar organization of these inputs within RL. We discovered a striking segregation: S1 projections predominantly targeted superficial layers (L1/L2/3), the principal receiving zone for apical dendrites of L2/3neurons, while V1 projections primarily innervated mid-to-deep layers. This anatomical arrangement provides a compelling structural basis for modality-specific inputs interacting with distinct dendritic compartments, potentially enabling non-linear integration. Subsequent 2-Photon calcium imaging at single-cell resolution during a similar discrimination task further substantiated RL's role, revealing distinct populations of neurons responsive to visual, tactile, or multisensory stimuli. However, simple averaging of somatic responses from functionally classified cells did not always reveal clear supra-linear summation, highlighting the need for subcellular investigation and analysis of learning-induced changes. A central aim was to probe MSI at the dendritic level. We successfully established methodologies combining sparse neuronal labeling with chronic, simultaneous 2-Photon calcium imaging of somatic and apical dendritic activity in L2/3 pyramidal neurons of V1 and RL in awake, behaving mice. Using this longitudinal imaging approach, we compared neuronal responses during passive sensory exposure before (naive) and after (trained) associative learning in a detection task. This revealed profound learning-induced plasticity, particularly within RL. Both dendritic and somatic responses to tactile and visuotactile stimuli were dramatically amplified in RL following training, while responses in V1 and responses to visual stimuli showed less pronounced changes. These functional plasticity findings, particularly the strong enhancement of tactile responses in RL, align remarkably well with our anatomical data showing dense S1input specifically targeting the superficial layers of RL where the apical dendrites ofL2/3 neurons reside. This convergence strongly suggests that associative learning potentiates synaptic inputs onto these dendritic compartments. While the complete analysis of dendritic dynamics during active task performance is ongoing, the current data provide substantial evidence for experience-dependent modulation of sensory processing at both somatic and dendritic levels. In conclusion, this thesis provides a multiscale investigation into the anatomical basis and functional implementation of visuotactile integration and learning in the mouse parietal association cortex. We demonstrated the convergence of visual and somatosensory pathways onto RL with distinct laminar targeting, providing a structuralsubstrate for integration. We confirmed RL's role as an integration hub wherepopulation activity correlates with better performance in MSI-dependent behavioraltasks. Most critically, we revealed profound learning-induced plasticity within RL,characterized by a dramatic amplification of tactile and multisensory responses in L2/3neurons, observable at both the somatic and dendritic levels. By establishing methodsfor subcellular imaging in behaving animals and linking anatomical connectivity withfunctional plasticity, this work strongly supports the hypothesis that experience-dependent modulation of dendritic integration within associative cortical areas like RLis a key mechanism underlying the brain's ability to effectively combine informationacross senses to guide perception and behavior. Future research completing the analysisof active dendritic processing and employing targeted manipulations will furtherelucidate the precise computational roles of dendrites in multisensory perception andlearning.

Authors 1

1. Abou Rachid Sacha corresponding Aachen

   RWTH Aachen University

   Affiliation as printed

   RWTH Aachen