Identifying the Point of No Return: Integrating Single-Cell Transcriptomics and Proteomics to Map Irreversible Fibroblast Commitment in Idiopathic Pulmonary Fibrosis

Arnab Nath

Department of Developmental Biology and Genetics, Indian Institute of Science (Deepak K Saini Lab)

Department of Computer Science and Automation, Indian Institute of Science (Chiranjib Bhattacharyya Lab)

1. Overview

Idiopathic pulmonary fibrosis (IPF) is a progressive, deadly lung disease caused by abnormal fibroblast activity and extracellular matrix deposition. Despite two FDA-approved medications (pirfenidone and nintedanib), IPF is incurable, with a median life expectancy of 3-5 years ¹. A significant knowledge gap is understanding when fibroblasts change from a reversible activated state to an irreversible myofibroblast fate—the 'point of no return.’ This change dictates the therapeutic window: early intervention can reverse fibrosis, but late-stage illness is resistant. This study will reanalyse publicly available single-cell RNA sequencing (scRNA-seq) datasets from IPF patients to reconstruct fibroblast developmental trajectories using pseudotime analysis and identify the molecular commitment point at which cells become locked into pathogenic states. To confirm gene signatures and explore causative pathways, we will perform single-cell proteomics on TGF-β-induced lung fibroblasts over time. This regulated experimental system replicates the quiescent → active → myofibroblast transition in vitro, enabling mechanistic validation of commitment-point genes identified from patient data.

Funding and facilities: The first section of this proposal is a pure computational reanalysis study that uses only publicly available datasets and conventional bioinformatics tools. Computing infrastructure is provided free of charge via lab resources. All required software (Seurat, Monocle3, CellRank, and scVelo) is open source. The LC-MS/MS analysis will be performed at the Indian Institute of Science Mass Spectrometry Facility.

Timeline: 6 months. Months 1-2 involve acquiring, QCing, and integrating datasets for IPF scRNA-seq research, as well as setting up TGF-β time-course experiments and preparing single-cell proteomics samples. Month 3: Using patient data to infer trajectory and create commitment signatures. Month 4: Generate and analyse single-cell proteomics data. Months 5-6: integrate transcriptomics and proteomics, validate commitment genes at the protein level, and prepare the article.

Data sharing commitment: All LC-MS procedures, raw chromatograms, all analysis code, integrated single-cell objects, trajectory coordinates, and gene signatures will be shared on GitHub and Zenodo under a CC-BY license. ShinyApps will provide an interactive browser to explore fibroblast trajectories. A preregistration will be placed on ResearchHub before the analysis begins, and the results will be shared as a preprint regardless of the outcome.

2. Introduction

Background –

IPF has a significant global impact, with adjusted prevalence estimates ranging from 0.33 to 4.51 per 10,000 people and incidence from 0.09 to 1.30 per 10,000 people worldwide ². The condition is characterised by progressive lung scarring caused by abnormal fibroblast activation. Current medicines delay but do not reverse fibrosis, in part because there are no biomarkers distinguishing reversible from irreversible disease stages ^1,3. Animal models suggest that early-stage fibrosis can be regressed after damage removal, while established myofibroblast foci persist ⁴. This implies a biochemical 'commitment point' beyond which cells cannot return to quiescence.

Single-cell transcriptomics revealed remarkable fibroblast heterogeneity in IPF lungs. Recent investigations have identified several fibroblast subpopulations, including quiescent, active (high ECM gene expression), and myofibroblasts (ACTA2+ contractile phenotype) ^5,6. However, this research examined cells as distinct clusters without recreating the developmental pathways that linked them. Pseudotime analysis, which arranges cells along a continuous differentiation trajectory, can uncover transition states and identify genes indicative of irreversible commitment.

The epithelial-mesenchymal transition (EMT), an emerging research topic in IPF biology, involves several intermediate stages, with some cells becoming irreversibly mesenchymal while others retain plasticity ⁷. Likewise, we posit that the transition from fibroblast to myofibroblast entails intermediate 'pre-committed' phases in which cells can still undergo dedifferentiation if they receive treatment, succeeded by a molecular checkpoint that locks them into pathogenic programming. Proteomics can also show this trajectory, and combining proteomics and transcriptomics can provide a deeper understanding of the cell's changes.

Research Questions –

Can single-cell trajectory analysis identify a discrete molecular commitment point in the fibroblast-to-myofibroblast transition in IPF, and which genes mark this point of no return?

Do patient scRNA-seq-identified commitment genes exhibit protein-level dynamics during TGF-β-induced fibroblast activation in vitro, indicating a commitment of transition?

Hypotheses –

H1: Pseudotime ordering of IPF lung fibroblasts shows a consistent trajectory from quiescent to active to myofibroblast states, with clear inflexion points in gene expression dynamics.

H2: The transition to irreversibility is marked by a gene 'commitment signature', which includes sustained upregulation in all myofibroblasts, low expression in reversibly activated fibroblasts, and, enrichment for epigenetic regulators and transcription factors that lock in profibrotic programs.

H3: When cross-referenced with clinical outcome data, the commitment signature will be expressed more in rapid IPF progressors than in slow progressors, and it will correspond with recognised fibrosis markers (FVC decline, fibroblastic foci burden on histology).

H4: RNA velocity studies will reveal that cells before the commitment point can move bidirectionally along the pathway, whereas post-commitment cells progress unidirectionally towards myofibroblast destiny.

H5: Commitment genes will show concordant RNA-protein dynamics in TGF-β-treated fibroblasts, with protein-level upregulation occurring 12-24h after transcript changes, and the commitment point (pseudotime 0.6-0.7 in patient data) corresponding to 48-72h TGF-β treatment in vitro.

Impact Statement –

This effort will establish the first molecular definition of irreversible fibroblast commitment in human IPF. The commitment gene signature meets the following clinical needs:

Early intervention biomarkers: Patients who have a high commitment signature in lung biopsies or bronchoalveolar lavage (BAL) cells will be prioritised for aggressive therapy or clinical trials.

Drug development: The commitment genes are medicinal targets. The signature also includes goals for preclinical studies: successful medicines should diminish commitment gene expression in animal models.

Understanding treatment failures: Current IPF medications produce variable results. Trajectory analysis may demonstrate that non-responders have reached the commitment point, whereas responders have reversibly activated fibroblasts. This explains heterogeneity and informs patient selection in future trials.

Even negative results are beneficial. If there is no clear commitment point (a smooth, continuous trajectory), it suggests that fibroblast plasticity is maintained throughout the disease, favouring a long-term suppression strategy rather than early intervention.

3. Methods

Participants/Sample

Publicly available data will be used. Biological triplicates will be employed for time-course proteomics analysis.

Materials and Procedures

Data Sources and Sample Selection

We will analyze publicly available scRNA-seq datasets from IPF patients deposited in Gene Expression Omnibus (GEO), Single Cell Portal, and European Nucleotide Archive (ENA).

Computational Pipeline

Each dataset will go through regular QC before being integrated with Harmony or ScVI. The subpopulation of fibroblasts will be identified using conventional marker gene expression. For trajectory analysis, complementary approaches such as Monocle3, CellRank, and scVelo will be used. The consensus trajectory will be defined as regions in which all three approaches agree on cell ordering. Transcriptional entropy, RNA velocity reversal, and gene expression dynamics will be used to determine the inflection points. The commitment gene signatures will be graded according to their consistency in expression changes and biological roles. The gene signatures will be confirmed against independent datasets and then applied to bulk RNA-seq data.

Single-Cell Proteomics Validation

TGF-β1 will be applied in human immortalized lung fibroblasts to induce myofibroblast differentiation in time-course mannar. Each timepoint will have biological triplicates. These samples will be used for single cell proteomics analysis using an already established lab protocol.

Integrating with trajectory data

Using transcriptional profiles, we will link in vitro timepoints to patient trajectory pseudotime. Protein expression dynamics will be compared to RNA dynamics to identify:

concordant genes (RNA-protein correlation >0.7),

post-transcriptional regulation (protein survives after RNA declines),

protein-only markers that transcriptomics cannot capture.

Planned Analyses

Primary analysis (H1-H2): reconstruct fibroblast trajectories, locate commitment points using entropy/velocity/GAM approaches, and obtain 10-15 gene signatures. Report signature performance (ROC-AUC) in discriminating myofibroblasts from activated fibroblasts.

Secondary studies (H3) will use linear mixed models and Cox regression to correlate the signature with clinical outcomes (FVC decrease and survival). Determine whether the signature improves predictive value beyond known biomarkers (Gender-Age-Physiology Index).

Tertiary analyses (H4) involve bidirectional RNA velocity analysis. Calculate the fraction of cells with backward-pointing velocity vectors in pre- and post-commitment regions (30-40% pre, <10%).

Proteomics analysis (H5): (i) Correlation of commitment gene RNA (patient scRNA-seq) and protein (TGF-β model) over pseudotime/timepoints. (ii) Identify proteins that exhibit non-linear dynamics, indicating post-translational control. (iii) Determine if a protein-level commitment signature (top 10) can better categorize cells than RNA alone (estimated AUC improvement: 0.75 -> 0.85).

Ethics and Data Management

This study relies solely on publicly available, de-identified datasets, and a cell line-based model. No permission from the ethical committee is necessary. All data sources are published, including data use agreements that allow for reanalysis. The analysis code will be version controlled on GitHub, with continuous integration testing. Processed objects (Seurat RDS files) and gene signatures will be stored on Zenodo with a DOI. An interactive Shiny app will allow the community to explore different trajectories. Prior to downloading data, a preregistration detailing the hypothesis and an analysis plan will be submitted on ResearchHub.

4. Pilot Data (Optional)

scRNA-seq datasets are publicly available. We developed the TGF-β-induced myofibroblast differentiation model in primary immortalised lung fibroblasts. The analytical pipeline and proteomics sample run procedure are already established in the lab.

5. Budget

Cloud computing for large-scale data integration: $1,200

Software licenses (Prism for figures, Biorender for schematics): $600

Data storage (Zenodo Pro for large file hosting, 1TB): $500

Cell consumables: $1,500

Proteomics cost: $1,200

Total: $5,000

Note: Personnel costs are not included as this work will be conducted by the applicant as part of ongoing doctoral research. Institutional computing cluster access is provided at no cost. All bioinformatics software is open source (Seurat, Monocle3, scVelo, CellRank).

6. References

1. Maher, T. M. & Strek, M. E. Antifibrotic therapy for idiopathic pulmonary fibrosis: time to treat. Respir Res 20, 205 (2019).

2. Maher, T. M. et al. Global incidence and prevalence of idiopathic pulmonary fibrosis. Respir Res 22, 197 (2021).

3. Stainer, A. et al. Molecular Biomarkers in Idiopathic Pulmonary Fibrosis: State of the Art and Future Directions. International Journal of Molecular Sciences 22, 6255 (2021).

4. Tashiro, J. et al. Exploring Animal Models That Resemble Idiopathic Pulmonary Fibrosis. Front. Med. 4, (2017).

5. Kato, K. et al. Impaired Myofibroblast Dedifferentiation Contributes to Nonresolving Fibrosis in Aging. Am J Respir Cell Mol Biol 62, 633–644 (2020).

6. Jin, C. et al. Single-cell RNA sequencing reveals special basal cells and fibroblasts in idiopathic pulmonary fibrosis. Sci Rep 14, 15778 (2024).

7. Li, W., Xie, Y., Chen, Z., Cao, D. & Wang, Y. Epithelial–mesenchymal transition in pulmonary fibrosis: molecular mechanisms and emerging therapeutic strategies. Front. Med. 12, (2025).

Identifying the Point of No Return: Integrating Single-Cell Transcriptomics and Proteomics to Map Irreversible Fibroblast Commitment in Idiopathic Pulmonary Fibrosis

Identifying the Point of No Return: Integrating Single-Cell Transcriptomics and Proteomics to Map Irreversible Fibroblast Commitment in Idiopathic Pulmonary Fibrosis

Funding Progress

Funders

Peer Reviews

Related Funding

Request for Proposals - Genomics and Proteomics

Nonprofit Recipient

Topics

DOI