The exercise-mimetic question — SLU-PP-332, AICAR, and what we've learned

The recurring pattern

Exercise mimetics have come along roughly once a decade since the 2000s, each with a similar narrative arc: a preclinical paper showing that activating a specific molecular pathway produces exercise-like metabolic adaptations in mice — increased running capacity, improved insulin sensitivity, fat oxidation. The papers generate substantial press and biohacker interest. The compound enters clinical development. Translation proves harder than the mouse data suggested.

AICAR — AMPK activation

AICAR (5-aminoimidazole-4-carboxamide ribonucleotide) was the first major exercise-mimetic candidate. Narkar et al. (2008, Cell) showed AICAR alone produced ~44% increased running endurance in untrained mice. The mechanism — AMPK activation — connected directly to one of the major metabolic pathways exercise itself activates.

The translation has been complicated. AICAR has had limited clinical development outside specific rare-disease indications. The pharmacokinetic challenges (poor oral bioavailability), the breadth of AMPK's downstream effects, and the difficulty of achieving the dosing required for the rodent endurance effects in humans together kept the compound from becoming a clinical exercise mimetic.

GW501516 — PPARδ agonism

GW501516 (cardarine) was the more cautionary case. The same Narkar 2008 paper showed GW501516 produced exercise-like metabolic adaptations in trained mice. Recreational use in athletic communities followed. Subsequent rodent toxicology revealed dose-dependent carcinogenicity across multiple organ systems. Development was halted; the compound is now banned by WADA at all times for athletes.

The GW501516 story is instructive specifically because the preclinical efficacy was real. The drug really did mimic exercise in mice. The cancer signal that emerged in longer rodent studies wasn't a translation failure of the efficacy claim — it was the discovery of a separate problem that the initial efficacy work hadn't characterized.

SLU-PP-332 — pan-ERR agonism

SLU-PP-332 is the current candidate generating significant attention. The 2024 Mol Metab paper (Billon et al.) showed daily SLU-PP-332 in mice produced increased running capacity, weight loss without reduced food intake, and skeletal-muscle gene-expression patterns matching exercise. The molecular target — pan-agonism of estrogen-related receptors — is mechanistically novel and connects to the PGC-1α coactivator pathway exercise itself uses.

The pattern so far is the familiar one: striking mouse data, generating substantial interest, with the translational work still ahead. We'd want to see human pharmacokinetic and safety characterization, longer-duration animal toxicology to characterize whether ERR pan-agonism produces unexpected long-term effects (the field has not done this for the broad-receptor activation profile), and dose-response work in human metabolic indications.

What the history actually teaches

Three patterns recur:

1. Mouse-to-human translation is harder than the mouse data suggests. Compounds that produce 30–50% endurance increases in mice rarely produce comparable effects in humans, partly because mouse skeletal muscle responds to interventions differently than human muscle, and partly because the daily dosing and pharmacokinetic profiles required for the rodent effects are difficult to reproduce in humans.
2. Long-term toxicology can surface late. The GW501516 cancer signal didn't appear in early studies. Drug-development paths for exercise mimetics specifically need extended toxicology because the use case (chronic supplementation in healthy adults) is high-bar.
3. Exercise itself is doing more than activating one pathway. The cardiovascular, neurological, psychological, and metabolic benefits of exercise are produced by a network of effects that no single-pathway pharmacology has come close to reproducing.