# Organization of mouse prefrontal cortex subnetwork revealed by spatial single-cell multi-omic analysis of SPIDER-Seq

## TL;DR

This study uses a new technique called SPIDER-Seq to map the mouse prefrontal cortex's complex organization and neural connections.

## Contribution

The novel SPIDER-Seq method integrates viral barcoding, single-cell sequencing, and spatial-omics to create a detailed multi-modal atlas of the mouse prefrontal cortex.

## Key findings

- Projection neurons in the PFC occupy distinct territories aligned with their wiring patterns.
- Neurons with reciprocal circuit connections show higher co-projection probabilities to downstream nuclei.
- Machine learning accurately predicts neuron projections using gene profiles and spatial data.

## Abstract

Deciphering the connectome, anatomy, transcriptome and spatial-omics integrated multi-modal brain atlas and its underlying organization principles remains a great challenge. We developed a Single-cell Projectome-transcriptome In situ Deciphering Sequencing (SPIDER-Seq) technique by combining viral barcoding tracing with single-cell sequencing and spatial-omics. This empowers us to delineate an integrated single-cell spatial molecular, cellular, anatomic and projectomic atlas of the mouse prefrontal cortex (PFC). The projectomic and transcriptomic cell clusters display distinct modular organization principles, but are coordinately configured in the PFC. The projection neurons gradiently occupied different territories in the PFC aligning with their wiring patterns. Importantly, they show higher co-projection probability to the downstream nuclei with reciprocal circuit connections. Moreover, we integrated the projectomic atlas with its distinct spectrum of neurotransmitters/neuropeptides with their receptor-related gene profiles in order to demonstrate the PFC neural signal transmission network, by which means we uncovered potential mechanisms underlying the complexity and specificity of neural transmission. Finally, leveraging machine learning, we predicted neuron projections with high accuracy by combining gene profiles and spatial information. As a proof of concept, we used this model to predict projections of fear recall engram neurons. This study facilitates our understanding of the brain multi-modal network and neural computation.