Molecular changes in the habenula mark the onset of depression

Summary

This study mapped the topographically organized, time-locked molecular remodeling in the lateral habenula following chronic corticosterone exposure in male mice, using spatial transcriptomics and two-photon calcium imaging. It demonstrates that increased excitatory responses to aversive stimuli develop within a specific time window prior to the emergence of depression-like behaviors, and that this condition is reversible via ketamin or chemogenetic silencing.

Why it matters

Depression treatments often fail because the underlying neurobiological cascade has already become entrenched. Identifying a molecularly defined 'initiation window' in a critical stress-response circuit could, in principle, guide earlier and more targeted interventions. Additionally, the study provides one of the first spatially resolved atlases of the habenula, a region long associated with mood but previously insufficiently mapped at single-cell resolution.

Highlights

• Spatial transcriptomics revealed distinct and temporally ordered DEG signatures in three LHb subregions (medial, central, and lateral) following 1, 7, and 21 days of CORT administration.
• Two-photon imaging showed a gradual and measurable increase in excitatory responses to aversive stimuli, progressing in parallel with the transcriptional timeline.
• Reduction of LHb activity via ketamin or chemogenetic silencing applied during the earliest molecular change window produced antidepressant effects, indicating the existence of a therapeutic window.

Critical take

The model relies on a single inbred, entirely male mouse strain exposed to pharmacological CORT added to drinking water. Consequently, the 'initiation window' is a product of a controlled dosing regimen rather than the heterogeneous, self-paced stress trajectories observed in humans. Furthermore, the spatial transcriptomics data consist of static snapshots at three time points; therefore, the claimed 'spatiotemporal' dynamics were not directly observed in real time but were derived through interpolation.