Localization, not selectivity: a causal node–cell–outcome framework for designing senescent-cell-directed senotherapeutics, with an application to inflammaging

Senolytic and senomorphic agents are dose-limited by toxicity, and the field's default response is greater target selectivity. This preprint argues that is only half right: a substantial share of these toxicities are on-target effects occurring in the wrong cell (for example BCL-xL inhibition causing thrombocytopenia in platelets, or NF-κB/JAK inhibition immunosuppressing host-defense immune cells), which target selectivity cannot remove; a third class is intrinsic to senolysis itself. The paper formalizes senotherapeutic pharmacology as a causal map from node to cell to outcome and jointly optimizes a node-engagement profile with a cell-selective delivery localization. Exhaustive optimization (98,301 designs) shows that only senescent-cell-targeted delivery reaches zero predicted adverse effects, that neither systemic nor platelet-sparing PROTAC delivery does, and that the clean optimum necessarily excludes BCL-xL and FOXO4–p53. The optimized profiles are mapped onto twelve clinically relevant agents (structures PubChem-sourced, formula and weight cross-checked in RDKit), and senescence-associated β-galactosidase (galacto-prodrug activation) and platelet-sparing protein degradation are identified as concrete delivery instruments. The per-node efficacy weights are illustrative heuristics; the contribution is the design principle and optimization framework, offered as a hypothesis-generating tool, not a validated therapeutic. Companion to US Provisional Patent Application No. 64/121,903. All code and data are released for reproduction.