Integrating single-cell transcriptomics and network pharmacology identifies ADORA1 as a candidate receptor for gastrodin derivatives in lung adenocarcinoma.
TL;DR
These analyses nominate ADORA1 as a candidate receptor and Gastrodin_T9 as a lead among gastrodin derivatives in LUAD; both candidates warrant mechanistic validation in LUAD-specific models.
Abstract
Background
Gastrodin, the main bioactive constituent of Gastrodia elata, shows antitumor activity, but its cell-type-specific mechanism in LUAD remains elusive.
Methods
Single-cell RNA-seq data (GSE131907; 208,506 cells, 58 samples, 44 patients) were analyzed with five frameworks: AUCell drug-sensitivity scoring, scTenifoldNet/scTenifoldKnk network/knockout analysis, decoupleR pathway inference, CellChat communication mapping, and MEBOCOST metabolic communication. Ten gastrodin derivatives were ranked by TOPSIS, docked against human ADORA1 (PDB: 5UEN), and validated by surface plasmon resonance (SPR) and an LPS-induced acute inflammation model in Kunming mice (CD68 immunohistochemistry, PARP Western blot, cytokine ELISA).
Results
ADORA1 was the only core target with both receptor and metabolic-sensor roles. TOPSIS ranked Gastrodin_T9 first (0.592), despite T8's highest network betweenness (0.1389). Docking gave modest affinities (-4.18 to -4.93 kcal/mol) with a shared 12-residue core; T9 covered most of the pocket (37/41). SPR supported the same order for three derivatives (T9 > T7 > T1; apparent K_D 22.79-68.18 μM, n = 3 runs), whereas native-gastrodin signal was too weak for reliable estimation. Lung conditioned medium showed Th1-skewed cytokines: IL-12 rose from 55.5 to 304.1 pg/mL under T9 (p < 0.001), IL-10 was unchanged, and the IL-12/IL-10 ratio shifted four-fold (p < 0.001). Derivatives raised the cleaved-PARP ratio modestly (1.05-1.15-fold), a preliminary change; CD68 staining was heaviest in the native-gastrodin group.
Conclusions
These analyses nominate ADORA1 as a candidate receptor and Gastrodin_T9 as a lead among gastrodin derivatives in LUAD. The evidence is hypothesis-generating and biophysical; both candidates warrant mechanistic validation in LUAD-specific models.