A Recurrent Slc17a6-High Glutamatergic Microdomain in the Parabrachial Region Following Paclitaxel Exposure: A Whole-Brain Spatial Transcriptomic Reanalysis

Abstract

BackgroundPaclitaxel is a widely used antineoplastic agent whose therapeutic utility is limited by persistentneurological adverse effects, including chemotherapy-induced peripheral neuropathy (CIPN). Mostmechanistic studies of paclitaxel neurotoxicity have focused on peripheral sensory neurons, dorsal rootganglia, axonal transport, mitochondrial dysfunction, neuroimmune signaling, and spinal sensitization.Increasing evidence, however, indicates that chemotherapy also reorganizes supraspinal neural systemsinvolved in nociceptive processing. Whether paclitaxel exposure is associated with reproducible, spatiallylocalized molecular states within specific brain regions remains poorly understood. MethodsWe performed an exploratory secondary analysis of publicly available whole-brain spatial transcriptomicdata from paclitaxel-exposed and control mouse brain sections (GSE325554). Spatial gene-expressionmatrices were integrated with tissue coordinates and mapped computationally to the Allen Mouse BrainCommon Coordinate Framework. Candidate molecular states were initially examined across genesassociated with glutamatergic transmission, somatostatin signaling, MAPK signaling, serotonergicsystems, neuronal activity, and pain-related pathways.A spatial Gate score incorporating Sst, Slc17a6, and Mapk14 initially identified localized candidatedomains. Component decomposition subsequently demonstrated that Slc17a6, encoding vesicularglutamate transporter 2 (VGLUT2), was the most reproducible component within the parabrachial region(PB), whereas PB-specific Sst enrichment was absent and Mapk14 enrichment was inconsistent acrosspaclitaxel sections. We therefore refined the analysis around a Slc17a6-high PB spatial state.Spatial reproducibility was quantified between independent paclitaxel sections using Jaccard and Diceoverlap statistics. To determine whether the observed overlap exceeded that expected from intrinsic PBgeometry, we generated 10,000 PB-constrained spatial-null configurations using rotation and translationof the microdomain while approximately preserving its spatial extent. ResultsWhole-brain spatial transcriptomic reanalysis identified a recurrent, spatially organized Slc17a6-highglutamatergic microdomain within the parabrachial region after paclitaxel exposure.The two independently analyzed paclitaxel sections exhibited substantial spatial concordance of theSlc17a6-high PB microdomain, with a Jaccard index of 0.506 and a Dice coefficient of 0.672. Under a PBconstrained spatial-null model, the mean null Jaccard index was 0.158, with 95th and 99th percentiles of1 DOI: 10.5281/zenodo.222656450.400 and 0.481, respectively. The observed Jaccard overlap exceeded the 99th percentile of the spatialnull distribution (P = 0.0044). Dice-based testing produced the same spatial-null probability (P = 0.0044).Comparison with the evaluable control section suggested greater PTX–PTX than PTX–control spatialconcordance. PTX1–Ctrl2 and PTX2–Ctrl2 Jaccard indices were 0.235 and 0.182, respectively, comparedwith 0.506 between PTX1 and PTX2. The resulting ΔJaccard was +0.298. Barcode-level bootstrap analysissupported the direction of this effect in 94.7% of bootstrap iterations, although its 95% bootstrapconfidence interval crossed zero (−0.037 to 0.402), indicating that treatment-level specificity remainsincompletely established.Atlas-based refinement consistently assigned the recurrent state to the parabrachial region in bothpaclitaxel sections. The available two-dimensional atlas coordinates did not retain sufficient informationfor definitive assignment to individual PB subnuclei. ConclusionsThese findings identify a reproducible spatial organization of a Slc17a6-high glutamatergic state withinthe parabrachial region following paclitaxel exposure. The result extends previous evidence implicatingglutamatergic PB circuitry in neuropathic and nociplastic pain by demonstrating a spatially recurrentmolecular architecture in paclitaxel-exposed brain tissue.Because the source dataset was generated to investigate long-term paclitaxel-associated brain changesrather than painful neuropathy, and because no pain behavior was linked directly to the analyzedsections, the identified PB microdomain should not yet be interpreted as a causal substrate of CIPN.Instead, it represents a spatially defined candidate central node for experimental investigation ofpersistent chemotherapy-associated neural states.