Omics & gene networks

Interspecies Single-Cell RNA-Seq-Based Deconvolution of Bulk Transcriptomic Data to Characterize Mammary Cell Composition Dynamics in LPS-Challenged Dairy Ewes

---

Abstract

In livestock, the amount of bulk RNA-Seq data generated to study complex and economically relevant traits has increased exponentially over the last two decades. However, bulk RNA-Seq experiments measure gene expression from heterogeneous mixtures of multiple cell types, which can hide cell type-specific signals. Traditional gene expression analyses typically assume homogeneity within each sample, an assumption that may not be true in biological contexts characterized by inflammatory responses. To overcome this limitation, single-cell RNA sequencing (scRNA-Seq) technologies have been introduced into livestock research, offering the potential to investigate gene expression at cellular resolution. Nevertheless, these approaches remain technically demanding and expensive, limiting their widespread use. In this context, computational deconvolution of bulk RNA-Seq data has emerged as a valuable strategy to estimate the relative abundance of distinct cell types within complex tissues, leveraging existing scRNA-Seq datasets as reference profiles. Such integrative approaches allow the utilization of bulk RNA-Seq resources while gaining cell-level insights into physiological and pathological processes. The aim of this study was to evaluate the feasibility of using interspecies scRNA-Seq reference data to deconvolute bulk RNA-Seq datasets derived from Assaf ewes subjected to an experimental E.coli lipopolysaccharide (LPS) challenge in the mammary gland, as a model for subclinical mastitis. Milk somatic cells were collected from healthy glands (control), 6 hours post-challenge, and 24 hours post-challenge. The scRNA-Seq reference dataset was obtained from publicly available bovine mammary gland single-cell transcriptomes (PRJNA1235849). The reference single-cell reads were processed and aligned to the bovine genome using Seurat, while bulk RNA-Seq reads were aligned to the ovine reference genome using STAR. The R package CIBERSORTx was employed to estimate the relative proportions of epithelial mammary gland and immune cell types across the three experimental timepoints. The deconvolution analysis revealed a marked increase in neutrophil proportion at 6 hours post-LPS (95% CI= 0.898-0.933; padj=1.44e-7), accompanied by a transient reduction in secretory epithelial cell fraction (95% CI= 0.006-0.04; padj=3.84e-7). By 24 hours post-challenge, the cellular composition remained distinct from baseline, with persistent neutrophil dominance and significantly increased macrophage proportions (95% CI= 0.0933-0.151) compared with both control (padj=0.0290) and 6-hour post-LPS samples (padj= 0.0009). Differential expression analysis, after correcting for cell composition, identified genes involved in inflammatory signaling, such as IL6 (padj = 8.19 à— 10⁻⁴), CCL20 (padj = 1.08 à— 10⁻²), and PTGS2 (padj = 1.68 à— 10⁻²), all of which are associated with immune regulation in mastitis. Although we acknowledge the limitations of interspecies reference use, and recognize that future studies should generate scRNA-Seq data from the same populations employed in bulk RNA-Seq analyses, the results obtained were biologically coherent, and highlight the utility of cross-species scRNA-Seq datasets as preliminary tools for assessing tissue composition and cellular dynamics in livestock transcriptomics.

Keywords: 2026

How to Cite:

Suárez-Vega, A., Alonso-García, M., Pelayo, R., Gutiérrez-Gil, B. & Arranz, J., (2026) “Interspecies Single-Cell RNA-Seq-Based Deconvolution of Bulk Transcriptomic Data to Characterize Mammary Cell Composition Dynamics in LPS-Challenged Dairy Ewes”, World Congress on Genetics Applied to Livestock Production Digital Archive 2026(1): 2285415. doi: https://doi.org/10.31274/wcgalp.23692

Rights: 1