Molecularly targetable cell types in mouse visual cortex have distinguishable prediction error responses.

Area of Science:

• Neuroscience
• Computational Neuroscience
• Molecular Biology

Background:

• Predictive processing theory suggests the brain uses prediction error neurons.
• Neurons with negative and positive prediction error responses exist in visual cortex layer 2/3.
• The transcriptional identity of these neurons remains unclear.

Purpose of the Study:

• To determine if transcriptionally defined cell types in mouse visual cortex layer 2/3 exhibit distinct prediction error responses.
• To link specific gene expression profiles to negative and positive prediction error signaling.

Main Methods:

• Used the activity-dependent, photoconvertible marker CaMPARI2 to label neurons in vivo during prediction error-inducing stimuli.
• Performed single-cell RNA-sequencing on photolabeled populations.
• Validated findings using artificial promoters and AAV vectors for calcium imaging.

Main Results:

• Specific layer 2/3 transcriptional cell types, Adamts2 and Rrad, were enriched in populations responding to negative and positive prediction errors, respectively.
• Demonstrated that these distinct transcriptional cell types exhibit distinguishable prediction error responses.
• Showed that these cell types can be targeted using adeno-associated virus (AAV) vectors.

Conclusions:

• Transcriptionally distinct cell types in the visual cortex layer 2/3 are functionally specialized for processing different types of prediction errors.
• These findings provide a molecular basis for understanding prediction error signaling in the brain.
• The ability to target these cell types with AAV vectors opens new avenues for studying neural computation.