The team, led by Shilong Liu and senior author Zhenhua Zhu of the First Affiliated Hospital of Hunan University of Chinese Medicine, began by mining publicly available single-cell RNA sequencing datasets from the Gene Expression Omnibus. After quality control and integration, they analyzed 41 CRSwNP samples and 7 healthy controls, resolving the nasal tissue microenvironment into 13 major cell populations. They then zoomed in on the two compartments most directly responsible for tissue architecture: the epithelium, which forms the mucosal barrier, and the fibroblasts of the underlying lamina propria, which manufacture the extracellular matrix. Epithelial cells resolved into 13 subpopulations and fibroblasts into 8, each defined by distinct functional states.

The compositional shifts they observed were striking and asymmetric. In the epithelial compartment, six of the thirteen subpopulations changed significantly in disease. Inflammatory CXCL8-positive cells and MUC5AC-positive goblet cells expanded, while homeostatic FOXJ1-positive ciliated cells, the workhorses of mucociliary clearance, shrank. Basal-like populations marked by KRT15 and COL17A1 grew, signaling a tissue locked into a perpetual injury-repair mode. Fibroblasts shifted more modestly, but the direction was telling: inflammatory CCL2-positive fibroblasts increased while APOD-positive homeostatic fibroblasts declined. The epithelium, in other words, tilted toward inflammation and aberrant secretion, while the stroma edged toward activation.

Differential expression and gene set enrichment analysis deepened the story. Diseased epithelial cells upregulated genes tied to injury responses and remodeling, including SERPINB4, KRT13, MMP7, and VIM, while losing genes that maintain barrier function and ciliary differentiation such as SCGB1A1, FOXJ1, and MUC5B. Enriched pathways included epithelial-mesenchymal transition, hypoxia, inflammatory response, and TGF-beta and WNT signaling. Fibroblasts showed a complementary profile: POSTN, COL1A2, SERPINE1, and CCL2 rose, while homeostatic genes like PI16 and MFAP4 fell, with collagen metabolism and TGF-beta production prominently enriched. Shared hub genes such as IL6 and SERPINE1 appeared in both cell types, hinting at a common inflammatory and matrix-remodeling program.

Pseudotime analysis using Monocle3 and CytoTRACE2 then reconstructed how these states emerge. Epithelial cells appeared to branch from a secretory starting point toward ciliated, mucous, and injury-repair fates, with homeostatic genes like EPCAM and CDH1 declining along the trajectory while KRT15, COL17A1, and the EMT-associated transcription factor SNAI2 increased. Fibroblasts progressed from quiescent CFD- and APOD-expressing states toward activated, contractile phenotypes marked by TAGLN, POSTN, and MYL9, with collagen genes COL1A1 and COL1A2 peaking in mid-to-late pseudotime. The picture is one of gradual, coordinated drift away from tissue homeostasis rather than an abrupt switch.

The most consequential findings came from CellChat, a computational framework that infers ligand-receptor communication between cell populations. In CRSwNP, the overall communication network grew denser and stronger, with two modules standing out. The COLLAGEN module, built on pairs such as COL1A1 and COL1A2 binding CD44, SDC4, and integrins, flowed predominantly from activated fibroblast subpopulations toward specific epithelial subsets. The TGF-beta module, involving TGFB1 binding TGFBR1/TGFBR2 and ACVR1, was more bidirectional, with epithelial and fibroblast compartments signaling reciprocally. Extending the analysis to myeloid cells, the team found that TGFB1 transcripts concentrated in a GPNMB-expressing macrophage state, marked by SPP1, TREM2, and MMP9, that was significantly expanded in polyps, echoing scar-associated macrophages described in fibrotic liver and other tissues.

Crucially, the computational predictions held up at the protein level in human tissue. Examining nasal polyps from three patients and control mucosa from three individuals undergoing septal surgery, the researchers used Masson’s trichrome staining and tyramide signal-amplified immunofluorescence. Polyps showed denser subepithelial collagen deposition, stronger COL1A1 and alpha-SMA signals in the lamina propria, nuclear accumulation of phosphorylated Smad3 in epithelial cells, weakened E-cadherin, and increased N-cadherin and periostin, all statistically significant. One particularly revealing observation concerned architecture rather than abundance: in healthy mucosa, COL1A1 and periostin formed a compact, continuous band beneath the epithelium, but in polyps this band became loose and discontinuous, with matrix proteins diffusing into the lamina propria. Remodeling in CRSwNP, the data suggest, involves not just more matrix but disorganized matrix.

To test causality in a living system, the team established a CRSwNP-like mouse model using intranasal ovalbumin challenge combined with Staphylococcus aureus enterotoxin B, a superantigen long implicated in polyp formation. Over roughly four months, model mice developed thickened respiratory epithelium, polypoid protrusions, inflammatory infiltration, and significantly increased subepithelial collagen by Sirius Red staining under polarized light. Immunofluorescence mirrored the human findings: alpha-SMA-positive cells accumulated with COL1A1 deposition, epithelial E-cadherin fell, nuclear p-Smad3 rose, and N-cadherin and periostin increased. Western blotting confirmed elevated COL1A1, POSTN, N-cadherin, alpha-SMA, and an increased p-Smad3 to total Smad3 ratio.

The therapeutic implications emerged when the researchers treated model mice with dexamethasone, the corticosteroid that remains first-line therapy for CRSwNP. The drug partially reversed the remodeling phenotype: alpha-SMA and E-cadherin returned to levels statistically indistinguishable from controls, while COL1A1, POSTN, and N-cadherin declined but stayed above baseline. The p-Smad3/Smad3 ratio showed only a non-significant downward trend. This partial response is mechanistically coherent, since the glucocorticoid receptor is known to directly inhibit the transcriptional activation function of Smad3, but it also carries a sobering message: broad anti-inflammatory treatment alone may not fully reverse established structural remodeling, and pathway-targeted anti-fibrotic strategies deserve exploration.

The authors are careful about the limits of their work. CellChat and pseudotime analyses are computational inferences that nominate candidate signaling axes rather than prove physical ligand-receptor engagement, and no knockdown, pathway blockade, or co-culture experiments were performed. The two public cohorts differ in ethnicity and inflammatory endotype, one skewed toward Type 2 inflammation and the other toward IL-1beta-driven myeloid inflammation, and eosinophils, classic sources of TGF-beta1 in polyps, were too sparsely captured for subcluster analysis. Yet the reproducible detection of the same epithelial-fibroblast TGF-beta/Smad axis across cohorts differing in platform, ancestry, and sampling site argues that it is a shared feature of polyp remodeling rather than a cohort-specific artifact. If future spatially resolved and functional studies confirm this circuit, the collagen and TGF-beta dialogue between epithelium and stroma could become a tractable target for therapies aimed not just at shrinking polyps, but at preventing the scarring that makes them return.

Subject of Research: TGF-beta/Smad-mediated epithelial-fibroblast remodeling and collagen deposition in chronic rhinosinusitis with nasal polyps

Article Title: TGF‐β/Smad‐Associated Epithelial–Fibroblast Remodeling Underlies Collagen Deposition in Chronic Rhinosinusitis With Nasal Polyps

Article References: Liu, S., Xie, P., Deng, H., & Zhu, Z. (2026). TGF‐β/Smad‐Associated Epithelial–Fibroblast Remodeling Underlies Collagen Deposition in Chronic Rhinosinusitis With Nasal Polyps. Immunity, Inflammation and Disease, 14(10), Article e70532. https://doi.org/10.1002/iid3.70532

Image Credits: AI Generated

DOI: 10.1002/iid3.70532

Keywords: chronic rhinosinusitis, nasal polyps, TGF-beta, Smad3, fibroblasts, epithelial-mesenchymal transition, collagen deposition, single-cell RNA sequencing, CellChat, macrophages, dexamethasone, tissue remodeling