scRNA-seq reveals persistent aberrant differentiation of nasal epithelium driven by TNFα and TGFβ in post-COVID syndrome

Abstract

Post-COVID syndrome (PCS) affects approximately 3-17% of individuals following acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection and poses a potential global health burden. While improved assessment strategies are emerging, mechanistic insights and treatment options remain limited. This study investigates molecular mechanisms underlying PCS using single-cell RNA (scRNA) transcriptomics combined with in vitro validation. scRNA analysis is performed on nasal biopsies from 25 patients with moderate or severe PCS to investigate differential cell types, signalling pathways, and cell-cell communication. Air-liquid interface cultures are used to validate findings, focusing on the TNFα-TGFβ axis. Severe PCS shows reduced numbers of ciliated cells, increased immune cell infiltration, and heightened inflammatory signaling that drives TGFβ and TNFα upregulation, in the absence of a detectable viral load. These changes trigger epithelial-mesenchymal transition, basal cell expansion and a mis-stratified nasal epithelium. In vitro experiments confirm TGFβ and TNFα as causal cytokines promoting ciliated cell loss and increased basal cell abundance. These findings indicate a sustained severe PCS is not driven by ongoing viral load but by immune cell activity and chronic cytokine production. Targeting the TNFα-TGFβ axis may mitigate immune-mediated nasal tissue damage and support epithelium restoration, offering a potential therapeutic strategy for PCS.

Fig. 1. Schematic representation of sample collection and processing.

Nasal samples were collected using a currette, capturing the epithelium and local immune population. Samples were processed and dissociated to create a single-cell suspension for downstream single-cell RNA sequencing analysis. Created in BioRender  https://BioRender.com/tvlyurc.

Fig. 2. scRNA-seq analysis of the cellular composition of nasal samples from moderate and severe PCS patients.

A Cell-type annotated UMAP plot of all integrated samples. 17 distinct clusters were detected by cluster gene signatures. B Violin plots of marker gene expression (ln-transformed counts per million) of the 17 distinct clusters collapsed to 12 common cell types found in the conducting airways. C Stacked bar plots of cell cluster frequency in moderate and severe PCS patients. D Relative differences in cell cluster proportions between moderate and severe PCS groups. Each point represents the observed log₂ fold-difference in cell-type proportion (severe vs. moderate). Statistical significance was assessed using permutation testing with FDR correction. Error bars denote the 95% confidence intervals (2.5th and 97.5th percentiles), calculated from 1000 permutations. Significance is color-coded: red (FDR < 0.05 & |log₂FD | > 0.5), blue (FDR < 0.05), and gray (not significant). Vertical black dashed lines indicate the ±0.5 log₂FD thresholds. Group sizes were moderate (11 samples, 15,014 cells) and severe (14 samples, 29,680 cells). E Hazard and odds ratios of nasal and respiratory diseases significantly increased in post-COVID patients. Each ratio is plotted in pairs, where light blue and dark blue indicate the hazard and odds ratios. Error bars represent the 95% confidence intervals. The dotted horizontal line at 1.0 indicates no effect; values above this line signify an increased risk or association with the outcome. Specifically, an odds ratio >1 indicates greater odds of the outcome, while a hazard ratio >1 indicates a higher rate of the outcome over time. A total of 52,833 post-COVID cases and 51,310 controls were included in the analysis.

Fig. 3. Enriched pathological pathways and cell-cell interactions with PCS severity.

A Relative PROGENy pathway enrichment in moderate and severe PCS. Red indicates positive, whilst blue indicates negative enrichment for severe PCS. B Heatmap of the signaling pathway enrichment contributing to outgoing or incoming communication. The color bar indicates the relative signaling strength in PCS severity; red = increased in severe PCS, and blue = decreased in severe PCS. The solid-colored bars on the x- and y-axes indicate the sum of the incoming (x-axis) or outgoing (y-axis) signaling strength for each cell type. Comparison of total incoming signaling strength vs. total outgoing signaling strength across cell populations in moderate (C) and severe (D) PCS. Dot size is proportional to the number of outgoing and incoming inferred links associated with each cell population group. Dot colors indicate different cell population groups. All significant pathways (accumulated p-value < 0.05) are presented as absolute (E) or relative (F) information flow, ranked based on differences in the overall information flow between moderate (red) and severe (cyan) patients. The overall information flow of a signaling network was calculated by summarizing all communication probabilities in that network. Pathways colored red are enriched in moderate, those colored cyan are enriched in severe, and black are equally enriched in both conditions. G Significant ligand-receptor pair interactions are represented from all other cell groups to myeloid-dendritic cells in moderate (red) and severe (cyan) PCS. The dot color and size represent the calculated communication probability and p-values, respectively; n = 25. Empty space means the communication probability is zero. p-values are computed from a one-sided random permutation test (100 permutations). GF insulin-like growth factor, APP amyloid beta precursor protein, PTN pleiotrophin, CADM cell adhesion molecule, ADGRE5 adhesion G protein-coupled receptor E5, MK midkine, FN1 fibronectin 1, JAM junction adhesion molecule 2, CDH cadherin-1, PROS protein S, MIF macrophage migration inhibitory factor, NCL nucleolin, AXL receptor tyrosine kinase, CXCR4 CXC motif chemokine receptor 4, LGALS9 galectin 9.

Fig. 4. Presentation of cellular expression of TNFα and TGFβ and altered basal cell differentiation trajectory.

A Heatmap of PROGENy pathway enrichment, stratified by cell type; red indicates positive, whilst blue indicates negative enrichment in severe PCS. B UMAP of TGFβ1 and TNF expression; purple = increased expression, and gray = no expression. Gene expression is quantified as log₂Counts per million (CPM). C–F Pseudo-time analysis of nasal epithelial cell differentiation in moderate (C and D) and severe (E and F) PCS patients. C and E t-SNE of cell populations. D and F t-SNE plot with overlay of pseudo-time progression of cell development; purple = earlier in trajectory, and yellow = later in trajectory. Black lines represent the most likely path of cell maturation over the pseudo-time trajectory; n = 25.

Fig. 5. scRNA-seq analysis of ALI cultures stimulated with TGFβ or TNFα alone or in combination.

A UMAP of integrated scRNA-seq of primary nasal epithelial cells differentiated over 28 days across all stimulation conditions. B UMAP of individual stimulation conditions PBS, TNFα (10 ng/ml), TGFβ (1 ng/ml), and TNFα with TGFβ combined. Individual cell populations are identified for each condition based on canonical expression markers (Fig. S17). C Relative differences in cell proportions for each cluster across all conditions. Red colored shapes indicate a false discovery rate (FDR) < 0.05 and log2 fold-difference >0.5 compared to PBS control conditions; Squares (■) = TGFβ vs PBS, triangle (▲) = TNFα with TGFβ vs PBS, and a cross (╳) = TNFα vs PBS. D GSEA pathway enrichment analysis stratified by treatment conditions compared to PBS control. Color of the dots indicates significance, with red representing a smaller log-p-value and blue indicating higher. Dot size is proportional to the number of genes identified within each pathway; larger dots indicate gene sets with more associated genes for that pathway; n = 2 donors for each condition.

Fig. 6. Graphical summary of aberrant molecular and cellular mechanisms in the nasal epithelium in moderate vs severe PCS.

Severe PCS compared to moderate PCS is characterized by an increased abundance of basal epithelial cells due to increased TGFβ signaling. There is also reduced ciliated cell abundance, likely caused by cell-cell interaction between myeloid and T cells with differentiating basal cells with increased TNFα signaling, promoting altered epithelial differentiation. Created in BioRender. https://BioRender.com/g6hfbvy.