Data mining historical Chinese medical recipe collections and nuclear receptor profiling identify plant fractions...

Glucocorticoids (GCs) play a prominent role in the management of chronic inflammatory diseases like rheumatoid arthritis and psoriatic arthritis, but their use is associated with various adverse effects. The therapeutic and adverse effects of GCs can be partly explained by modes of engagement with their target, the glucocorticoid receptor (GR). Identifying compounds that modulate GR through alternative mechanisms of action may provide a strategy to decouple therapeutic efficacy from side effects. Historical manuscripts on Chinese pharmacotherapy, which document the empirical use of natural products in treating inflammatory diseases, represent an underexplored source in drug discovery. These texts offer a historical library of plant materials and their metabolites, enabling strategic pre-selection of plant candidates for GR-targeted screening.

This study utilizes the Chinese Historical Healthcare Manuscripts Database, a newly compiled corpus comprising over 41,000 medical recipes from 227 historical Chinese manuscripts, to identify plant-based treatments for rheumatoid arthritis and psoriatic arthritis. Pareto front analysis, a multi-objective optimization method, was applied to 1,897 relevant recipes to identify plants that consistently ranked high across multiple metrics, suggesting their effectiveness in historical practice. The results were evaluated by comparison with modern Chinese materia medica dictionaries. Extracts from ten of these resulting plants underwent fractionation and were screened for GR modulation using Nuclear Receptor Activity Profiling (NAPing).

Pareto front analysis identified 32 botanical drugs statistically associated with historical disease indications resembling rheumatoid arthritis, psoriatic arthritis, and psoriasis. Nineteen of these are explicitly described in Chinese materia medica dictionaries for the treatment of such diseases. None of the plant fractions tested by NAPing replicated classical GC-induced GR-coregulator binding, but three induced unique binding interactions, suggesting alternative GR modulation mechanisms.

This study illustrates how combining data mining of historical pharmaceutical recipes with molecular screening can accelerate the discovery of new and possibly safer GR modulators. Such approaches may inform future translational strategies for treating chronic inflammatory diseases.