Transient and layer-specific reduction in neocortical PV inhibition during sensory association learning

Dika Kuljis, Eunsol Park, Stephanie E. Myal, Claudia Clopath, Alison L. BarthbioRxiv 2020 · DOI

TL;DR

Learning a new association, like a whisker-based reward task, rapidly reduces inhibition from specific neurons (PV neurons) in the superficial layers of the somatosensory cortex. This "disinhibition" is crucial for enabling prolonged neural activity that underlies learning, and it's specifically triggered by reward, not just sensory input.

What Problem Does This Paper Address?

When we learn something new, especially through association (like linking a specific touch to a reward), our brains change. We know that the connections between neurons, particularly the excitatory ones, get stronger. However, the role of inhibitory neurons – the "brakes" of the brain – in this learning process has been less clear. Specifically, how do these inhibitory neurons, particularly the parvalbumin (PV)-expressing ones, change their activity during learning, and what effect does this have on the overall circuit? This paper aims to pinpoint the specific changes in PV inhibition during a sensory association learning task and understand their functional consequences.

Methods

The researchers used a multi-pronged approach:

  • Behavioral Training: They developed an automated system to train mice in a whisker-based reward-association task (SAT). This allowed for precise control and measurement of learning progress.
  • Electrophysiology: Using optogenetics (light-activated channels in neurons) and whole-cell patch-clamp recordings in brain slices, they measured the strength of inhibitory signals (PV-IPSCs) from PV neurons onto pyramidal (Pyr) neurons in different layers of the somatosensory cortex.
  • Anatomical Analysis: They used advanced microscopy and fluorescent labeling techniques to quantify the physical connections (synapses) between PV neurons and Pyr neurons.
  • Computational Modeling: They built a simplified model of the cortical circuit to simulate how changes in PV inhibition might affect neural activity and recurrent network dynamics.

Datasets

Not explicitly discussed as datasets for training/evaluation in the machine learning sense. The study relies on experimental data collected from mice.

Key Insights and Ideas

  • Learning is Layer-Specific: The reduction in PV inhibition isn't uniform across the cortex. It specifically targets superficial layers (L2/3) of the somatosensory cortex, leaving deeper layers (L5) largely unaffected. This suggests a targeted rewiring process.
  • Reward is the Key Driver: Simply experiencing the sensory stimulus (airpuff) without the reward (water) doesn't lead to this disinhibition. This highlights that the brain is distinguishing between mere sensory input and learning-associated reward.
  • Disinhibition Enables Recurrent Activity: The computational model revealed that reducing PV inhibition in L2/3 is critical for enabling prolonged, recurrent neural activity. This sustained activity is thought to be essential for consolidating learned associations.
  • Structural Changes Underlie Functional Changes: The reduction in PV inhibition isn't just a temporary change in neuron firing. It's associated with a physical loss of synapses on the Pyr neurons, indicating structural plasticity.

Key Results

  • Rapid Suppression: After just 24 hours of sensory association training, PV-IPSC amplitude in L2/3 Pyr neurons was suppressed by approximately 35%.
  • Layer-Specific Effect: This suppression was observed in L2/3 Pyr neurons but not in L5 Pyr neurons.
  • Transient Nature: The reduced PV inhibition returned to baseline levels after 5 days of training, coinciding with plateaued behavioral performance.
  • Structural Basis: Anatomical analysis showed a corresponding ~39% reduction in PV-associated synapses on L2/3 Pyr neurons.
  • Reward-Dependent: Pseudotraining (sensory stimulus without reward) did not induce any significant changes in PV inhibition.
  • Modeling Confirmation: The computational model demonstrated that a 35% reduction in L2/3 PV inhibition was sufficient to initiate prolonged recurrent activity between L2/3 and L5 neurons.

Limitations and Future Work

  • Presynaptic vs. Postsynaptic: While the study found a reduction in PV synapses on Pyr neurons, it suggests this is due to the dismantling of postsynaptic structures rather than large-scale retraction of PV axons. However, further investigation into the precise mechanisms of synapse removal is warranted.
  • PV Neuron Location: The study couldn't pinpoint the exact laminar location of the PV neurons that reduced their output to L2/3 neurons. Future experiments could explore this.
  • Other Inhibitory Neurons: The role of other inhibitory neuron types (like somatostatin-expressing neurons) in SAT-induced plasticity remains to be explored.
  • Projection Target Specificity: It's unclear if PV disinhibition is restricted to all L2/3 Pyr neurons or a specific subset defined by their projection targets.

The authors suggest that future work should focus on understanding how reward signals are integrated to trigger these specific PV disinhibition mechanisms and whether this plasticity is restricted to certain layers or neuron subtypes.

Summary updated

Abstract

Abstract Sensory and motor learning reorganizes neocortical circuitry, particularly manifested in the strength of excitatory synapses. Prior studies suggest reduced inhibition can facilitate glutamatergic synapse plasticity during learning, but the role of specific inhibitory neurons in this process has not been well-documented. Here we investigate whether inhibition from parvalbumin (PV)-expressing neurons is altered in primary somatosensory cortex in mice trained in a whisker-based reward-association task. Anatomical and electrophysiological analyses show PV input to L2/3, but not L5, pyramidal (Pyr) neurons is rapidly suppressed during early stages of sensory training, effects that are reversed after longer training periods. Importantly, sensory stimulation without reward does not alter PV-mediated inhibition. Computational modeling indicates that reduced PV inhibition in L2/3 selectively enables an increase in translaminar recurrent activity, also observed during SAT. PV disinhibition in superficial layers of the neocortex may be one of the earliest changes in learning-dependent rewiring of the cortical column. Impact statement Tactile learning is associated with reduced PV inhibition in superficial layers of somatosensory cortex. Modeling studies suggest that PV disinhibition can support prolonged recurrent activity initiated by thalamic input.
BibTeX
@misc{kuljis2020transient,
  author = {Dika Kuljis and Eunsol Park and Stephanie E. Myal and Claudia Clopath and Alison L. Barth},
  title = {{Transient and layer-specific reduction in neocortical PV inhibition during sensory association learning}},
  year = {2020},
  doi = {10.1101/2020.04.24.059865},
  howpublished = {bioRxiv},
}
Figure