SARM1 activation promotes axonal degeneration via a two-step phase transition.

SARM1 is a key executioner of axonal degeneration, acting through NAD⁺ depletion by NADase activity of its TIR domain. Although normally autoinhibited, SARM1 becomes activated in response to axonal damage; however, the underlying mechanism remains unclear. Here, using a class of pyridine-containing compounds that trigger SARM1-dependent axon degeneration, we uncover a two-step activation process. First, NMN primes the base exchange activity of SARM1, generating covalent adducts between ADP-ribose (an NAD⁺ hydrolysis product) and the compounds. These ADP-ribose conjugates then serve as molecular glues to promote the assembly of superhelical SARM1 filaments, in which TIR domains adopt an active configuration. After reaching solubility limits, these filaments condense into stable, phase-separated assemblies with full enzymatic activity. Unexpectedly, several clinical-stage SARM1 inhibitors targeting its TIR domain also form such adducts, paradoxically promoting its activation. These findings reveal a molecular mechanism that spatially restricts SARM1 activation to damaged axons and offer new guidance for therapeutic strategies targeting SARM1.