Skip to main content
Back to timeline
bioRxivSource publication:

RNA outside cell bodies stops being discarded: SCOUBI maps axon-dendrite interfaces across the mouse brain

Synopsis

Much of the RNA in spatial transcriptomics falls outside segmented cell bodies and is usually discarded, yet neuronal mRNAs can be carried far from the soma into axons and dendrites. SCOUBI bins this extrasomatic space, trains a shallow neural network on known axon and dendrite marker genes while requiring significant ligand-receptor co-expression between neighboring axonal and dendritic bins, and calls loci where the two converge with ligand-receptor colocalization interfaces. On mouse hemibrain Xenium data it produces axon, dendrite and interface maps that match known neuroanatomy, with interface transcriptomes enriched for synaptic genes and reproducible across independent samples and two imaging platforms.

Source-provided article image: Revealing the transcriptomic organization of extrasomatic interfaces using SCOUBI
Figure 1 · bioRxiv · Page 8

Interpretation

Transcript signal outside cell bodies becomes an object of analysis rather than segmentation residue. In imaging spatial transcriptomics a substantial share of transcripts lies outside segmented cell bodies and is generally discarded by analysis tools; the abstract estimates up to half of the transcripts in such data may lie there. In the mouse hemibrain Xenium Prime dataset, binning at 2×2 micrometers left about 70% of non-zero-expression bins in unsegmented extrasomatic space, carrying about 42% of all detected transcripts.

Axon versus dendrite identity is set jointly by marker-gene weights and signaling between neighboring bins. Marker expression alone does not say how much weight each marker deserves, while choosing ligand-receptor pairs depends on reliable axon and dendrite labels. The model was trained on 13 axonal and 13 dendritic markers with a dual objective combining marker specificity and the CellWHISPER communication z-score; afterward 27% and 25% of non-zero-expression extrasomatic bins were classified axon- and dendrite-enriched.

Interfaces are defined as 2×2 bin windows where axonal and dendritic bins converge with ligand-receptor colocalization. Earlier tools for extrasomatic RNA identify granules from local transcript clustering and do not require axonal and dendritic signals to co-occur. The mouse hemibrain run yielded about 32K interfaces comprising about 68K bins; against mcDETECT granules, interfaces gave 325 enriched genes versus 147, with stronger enrichment of synapse-related terms.

Interface transcriptomes carry region-specific molecular signatures rather than only mirroring nearby cell bodies. Contrasting interface enrichment with cell-body expression in the same region isolates transcripts that exceed what local cellular abundance predicts. Four independent Xenium study-level sets shared 45 interface-enriched genes (SuperExactTest p = 1.05×10⁻¹⁵⁰); two CosMx datasets yielded 163 and 162 interface-enriched genes, 136 of them shared.

Perspective

The demonstrated scope is imaging spatial transcriptomics of mouse brain: one Xenium hemibrain dataset for the main analysis, plus four Xenium studies and two CosMx datasets for reproducibility checks; interfaces are explicitly a micron-scale bioinformatic construct, not individual synapses.

Extrasomatic transcript distributions combine active transport, local translation, passive diffusion and platform-specific technical noise including off-target probe binding, and these contributions are not disentangled; interfaces differ from granules in requiring axonal and dendritic signals to co-occur alongside ligand-receptor enrichment, which needs a broader spatial window.