Architecture of the Serotonergic Projectome in the Mouse Brain

Key results

1. 01

   ShownSerotonin projections from dorsal and median raphe (DR, MR) are well described by five basis patterns. Hierarchical clustering of brains in this space gives five projectomic groups: hippocampal, basal ganglia, cortical, medial interbrain, and brainstem/lateral thalamic.

   Matrix of 110 brains x 280 regions. Held-out reconstruction error from speckled cross-validation plateaued at k = 5, at the replicate-variability baseline and well below the shuffle baseline. The factorization explained 73.5% of variance. 101 of 110 brains kept their group under 280-D clustering; 79.2% of 250 subsampling trials gave five groups. · Determining whole-brain serotonin axon…; DR and MR serotonin neurons comprise five…; Axon data analysis using unsupervised NMF; Figure legends

2. 02

   SuggestedGroup membership appears to follow functional relatedness rather than physical proximity. Central amygdala injections fell in the basal ganglia group but basolateral amygdala injections in the cortical group. VTA and nucleus accumbens core injections fell in the medial interbrain group, not the basal ganglia group.

   Central amygdala injections clustered with basal ganglia, basolateral amygdala with cortical; VTA and accumbens core with medial interbrain, not basal ganglia. The authors call this consistent with functional divergence. Same-group different-site cosine similarity matched same-site similarity (except cortical) and exceeded between-group similarity. Marker regions per group: 16, 8, 18, 28, 81. · DR and MR serotonin neurons comprise five…; Organization of the mouse brain from the…; Marker region analysis

3. 03

   ShownWithin groups, structure runs from discrete subgroups to continuous variation. Basal ganglia innervation splits into subpatterns that pair topographically across caudoputamen and globus pallidus externus, while cortical group brains vary continuously.

   NMF gave k = 4 subpatterns in caudoputamen (17 brains) and k = 3 in globus pallidus externus, with dorsal, ventral and caudal subpatterns correlated across the two regions. The 27 cortical-group brains (k = 3) did not form distinct clusters. No significant subpatterns for the medial interbrain or brainstem/lateral thalamic groups. · Within-group organization: from discrete…; Axon data analysis using unsupervised NMF; Figure legends

4. 04

   SuggestedTranscriptomic cluster and somatic position jointly predict group identity. Each group maps to a combination of transcriptomic clusters rather than a single one. Neurons collateralize preferentially within a group.

   Spatial transcriptomics with retrograde barcodes: 8,715 Tph2+ neurons, 635 barcode-positive. Multi-barcode rate 15.2% (62/408) in within-group mice vs 4.8% (11/227) in cross-group mice. Random forest on single barcoded cells: 74.0% accuracy combined vs 41.4% transcriptome-only and 58.4% position-only. Supervised NMF on raphe cell-body maps: 83.6% (chance 20%). · Projection-specific cell body distributions in…; Transcriptomic and spatial identity jointly…; STARmap data processing

5. 05

   SuggestedProjection-specific Tph2 depletion gives dissociable behavioral profiles across eight paradigms. The basal ganglia group diverged most. A multivariate discriminant model separated groups and generalized to a second batch using different marker regions.

   54 features. Basal ganglia group: fewer marbles buried, more food intake after fasting, less immobility in forced swim. Cross-validated LDA accuracy 79% (females) and 60% (males) vs permutation chance 25.9 ± 7.2% and 26.1 ± 7.0%. Batch 2 within-group similarity exceeded between-group similarity in all groups in both sexes, except controls. · Projection-specific Tph2 depletion produces…; Serotonergic projectomic groups organize the…; Multivariate behavioral classification using…