Code from: Critical role for the TGF-β1/mTORC1 signalling axis in defining the transcriptional identity of CTHRC1+ pathologic fibroblasts
Data files
Jul 24, 2026 version files 110.36 KB
-
COL1A1_qPCR_code.R
17.40 KB
-
CTHRC1_qPCR_code.R
18.01 KB
-
DEseq2_AZD_Rapa_TGF.R
13.56 KB
-
DEseq2_Interaction_Factors_v2.R
11.63 KB
-
mTOR_CTHRC1_Paper_Figures.R
3.76 KB
-
Plotting_Pathways_using_clusterProfiler.R
4.46 KB
-
README.md
1.93 KB
-
Scissor_R_script_from_GitHub.R
39.61 KB
Abstract
Excessive extracellular matrix (ECM) deposition leading to organ dysfunction, represents the concluding pathological outcome of many chronic fibrotic conditions. Single-cell studies have identified disease-associated fibroblast populations, including a CTHRC1+ subpopulation with high collagen gene expression. While TGF-β1-driven fibrogenesis involves cooperation between Smad and mTORC1/4E-BP1 signaling, the role of mTORC1 in regulating ECM transcriptional programs and fibroblast heterogeneity remains poorly understood. To address this, we analyzed RNA-seq datasets from primary human lung fibroblasts (pHLFs) to examine the effects of mTOR inhibition on the TGF-β1-induced transcriptome. Differential expression analysis compared dual mTOR inhibition (AZD8055) with allosteric mTORC1 inhibition (rapamycin) (GSE102674), followed by CRISPR-Cas9-mediated disruption of either RPTOR or RICTOR to disrupt mTORC1 and mTORC2 (RICTOR) signaling, respectively and subsequent RNA-seq (GSE272972). These analyses identified a broad set of TGF-β1-responsive genes dependent on mTORC1 activity. To investigate how mTOR signaling influences fibroblast subpopulation identity in the setting of the setting of idiopathic pulmonary fibrosis (IPF), we integrated bulk RNA-seq data with publicly available human IPF single-cell RNA-seq datasets (GSE136831, GSE135893) using the Scissor framework. Scissor integrates bulk phenotype information with single-cell transcriptomic data to identify phenotype-associated cell subpopulations. TGF-β1-stimulated pHLFs displayed transcriptional similarity to a CTHRC1+ fibroblast subpopulation enriched in IPF, whereas disruption of mTORC1 signaling abolished this overlap, indicating a critical role for the TGF-β1/mTORC1 axis in establishing this pathogenic fibroblast state. Additional analyses characterized baseline fibroblast populations and their transcriptional signatures. Functional validation using a selective mTORC1 inhibitor, RMC-5552, demonstrated suppression of TGF-β1-induced COL1A1 and CTHRC1 expression across multiple fibroblast lines. This repository provides the code used for differential expression and pathway analysis, bulk–single-cell integration (Scissor), and statistical modeling, including a linear mixed-effects model to account for donor and passage variability.
Dataset DOI: 10.5061/dryad.q2bvq8419
Description of the data and file structure
4 scripts (DEseq2_AZD_Rapa_TGF.R, DEseq2_Interaction_Factors_v2.R, mTOR_CTHRC1_Paper_Figures.R, Plotting_Pathways_using_clusterProfiler.R):
Bulk RNA-seq datasets to assess the impact of CRISPR/cas9 mTORC1 and mTORC2 knockouts and mTOR pharmacological inhibitors on the TGFbeta1 response in primary human lung fibroblasts
Dataset 1: CRISPR/Cas9 gene edited pHLFs targeting either mTORC1 or mTORC2
Dataset 2: Treatment with the dual mTOR inhibitor AZD8055 or the partial mTORC1 inhibitor rapamycin
1. and 2. Two sets of DEseq2 analysis using complex design formula (interaction analysis)
3. Figure generation for CTHRC1+ pathogenic fibroblasts publication, including dotplots, cnetplots, barplots
4. Pathways analysis using ClusterProfiler, including visualisation using dotplots, cnetplots etc
1 script (Scissor; "Scissor_R_script_from_GitHub.R"):
Analysing publicly available scRNA-seq datasets and matching to bulk RNA-seq in pHLFs using Scissor tool to further understand fibroblast heterogeneity.
Scissor tool (https://github.com/sunduanchen/Scissor) was applied to map bulk RNA-seq datasets where mTORC1 and mTORC2 were depleted using CRISPR/Cas9 gene editing to pre-existing IPF scRNA-seq datasets. Fibroblasts from Habermann et al (GSE135893) and Adams et al. (GSE136831) were extracted and used as reference scRNA-seq datasets.
1. Scissor analysis and visualisation of results
Linear mixed model (LMM) fitting for both COL1A1 and CTHRC1 can be produced using R code block in files “COL1A1_qPCR_code.R “ and “CTHRC1_qPCR_code.R”.
