Authors
Yiwei Qin, Fang Zhang, Siqi Li, Haoran Wang, Jiye Liu
Published in
Naunyn-Schmiedeberg's archives of pharmacology. Nov 06, 2026. Epub Nov 06, 2026.
Abstract
Fibroblast-derived extracellular-matrix programs are prominent in non-small cell lung cancer (NSCLC), but bulk transcriptomic signals do not identify the relevant cellular compartment or establish whether candidate stromal programs respond to defined compounds. We integrated cell-resolved multi-omics analyses with fibroblast perturbation experiments to prioritize an Astragalus-associated stromal program and test its response to calycosin. A systematic public-data workflow integrated curated compound-target relationships with transcriptomic, co-expression, proteomic, single-cell, microenvironmental, dependency, protein-interaction, perturbational, and ranking-robustness evidence. Experimental validation used four independent biological repeats in TGF-β1-activated MRC-5 human lung fibroblasts together with a minimal four-donor primary NSCLC CAF validation set, with calycosin, the TGF-β receptor inhibitor SB431542, and COL3A1 loss- and gain-of-function perturbations. A separate concentration-response experiment evaluated calycosin at 5, 10, and 25 µM, with molecular, signaling, viability, and conditioned-medium effects on A549 and NCI-H520 spheroid invasion assessed across the dose range. The formal cross-cohort intersection identified four candidates shared by LUAD and LUSC-MMP1, MMP9, COL3A1, and PPARG. COL3A1 had concordant RNA-protein support, fibroblast-enriched single-cell expression, and recurrent associations with CAF signatures. In the principal TGF-β1-activated MRC-5 experiment, 25 µM calycosin reduced COL3A1 expression from 4.57- to 2.50-fold, collagen III secretion from 3.82 to 2.26 ng/105 cells/16 h, the α-SMA-high fraction from 64.5% to 33.5%, and CAGA activity from 4.14- to 2.38-fold. In a separate four-repeat dose series (0, 5, 10, and 25 µM), the means of eight molecular and functional endpoints declined with dose; paired-repeat linear-dose slopes were negative (two-sided P = 9.83 × 10-6 to 2.71 × 10-4), although not every replicate decreased strictly at 5 µM. In GSE153935, COL3A1 was enriched in fibroblasts (7/7 evaluable matched donors; exact signed-rank P = 0.0156). All four primary CAF donors showed a reduction in COL3A1, ACTA2, and conditioned-medium-driven A549 invasion with calycosin, but none of the three contrasts met the familywise Holm threshold (adjusted P = 0.0679). At the highest dose, MRC-5 live-cell fractions and tumor-spheroid viability were modestly reduced, so a viability-related contribution to invasion outcomes cannot be excluded. COL3A1 knockdown reduced selected activation endpoints and overexpression partly restored A549 invasion under calycosin treatment. Public cohort data, an independent single-cell dataset and perturbation experiments in MRC-5 fibroblasts support a calycosin-responsive, TGF-β-associated fibroblast phenotype marked by COL3A1. Four primary NSCLC CAF cultures showed the same direction of effect, but none of the calycosin comparisons remained significant after multiplicity adjustment, so these data are preliminary. COL3A1 perturbation altered selected fibroblast-conditioned readouts. Direct calycosin-COL3A1 binding, exclusive mediation, and a toxicity-independent anti-invasive effect remain to be tested.
PMID:
42993221
Bibliographic data and abstract were imported from PubMed on 07 Nov 2026.
Read full publication at:
Please sign in
to see all details.
Advertisement
Stats
- Recommendations n/a n/a positive of 0 vote(s)
- Views 12
- Comments 0