A transcriptomic resilience gap between the substantia nigra and cerebellum is associated with regional vulnerability...

The mechanisms underlying the selective vulnerability of the substantia nigra (SN) in Parkinson's disease (PD) are incompletely understood. We computed the tissue-specificity index τ for 19,355 protein-coding genes across 68 GTEx v11 tissues, defining a gene as tissue-specific when τ ≥ 0.8, median TPM > 1, and the tissue of interest was the absolute maximum. Under this definition, the SN harboured only 12 SN-specific genes whereas the cerebellum (CB) contained 367 CB-specific genes. The direct pairwise contrast was not statistically significant (exact p = 0.090). The primary evidence is a negative correlation between signature density and PD vulnerability tier across all 13 brain subregions (Spearman ρ = -0.709, exact permutation p = 0.0098). This gradient, replicated across GTEx releases, was robust to relaxed tissue-specificity criteria and alternative τ thresholds and persisted after controlling for cell-type composition. Resilience-associated gene sets were more highly expressed in the CB (14 of 17 sets). The classical antioxidant system was the only set higher in the SN. Single-nucleus RNA-seq of 434,340 human SN nuclei showed that the SN-specific signature is dominated by, but not restricted to, dopaminergic neurons and is largely maintained in PD donors. Integration with the latest PD genome-wide association study showed that 12 of 367 CB-specific genes (3.27%) overlapped genome-wide significant PD variants versus 1.30% of a matched background (p = 0.0036). The SN-specific set overlapped no significant variant, a result that is uninformative given its size. CB-specific genes formed a significantly interaction-sparse STRING network, whereas the SN-specific network comprised a single dopamine-synthesis module (TH and SLC6A3). The SN-specific set showed a non-significant tendency toward fewer human paralogues. These findings support a transcriptomic resilience gap, defined as a narrow SN-specific repertoire alongside a broader and genetically proximal CB-specific repertoire. We propose this asymmetry as a hypothesis for regional selective vulnerability and a framework for prioritising region-specific neuroprotective targets.