Immune Cells in Knee Joints Show Lipid-Driven Inflammation, Study Finds

Knee osteoarthritis has long been dismissed as simple wear and tear, a mechanical consequence of decades of load-bearing on aging joints. But a growing body of research has reframed the condition as a disease of the whole joint, in which inflammation, immune dysregulation, and metabolic imbalance all play active roles. Now, a study published in PLOS One has added a striking new piece to that picture, showing that the inflammatory troublemakers inside the arthritic knee may be driven, in part, by the way they handle fat. By combining single-cell transcriptomics with an arsenal of machine-learning tools, a research team led by Xin Chen and Jifeng Zhao has identified an enzyme called LTA4H as a central player linking lipid metabolism to inflammatory immune cells in the joint’s soft tissues.

The team focused on a compartment of the knee that has attracted increasing attention in osteoarthritis research: the infrapatellar fat pad and the synovium, the thin membrane lining the joint capsule. The fat pad, once considered mere structural padding beneath the kneecap, is now recognized as a metabolically active tissue that can secrete inflammatory molecules. Together with the synovium, it forms a microenvironment where immune cells, fat-laden cells, and connective tissue cells interact. It is also a site of visible change in osteoarthritis: the tissue becomes thickened, infiltrated with immune cells, and scarred by fibrosis. Understanding which cells in this compartment behave abnormally, and which genes drive that behavior, could open new avenues for treating a disease that affects hundreds of millions of people worldwide and for which there is currently no disease-modifying therapy.

To dissect this complexity, the researchers turned to single-cell transcriptomics, a technique that measures the complete set of messenger RNA molecules in thousands of individual cells, allowing scientists to identify distinct cell types and states within a tissue rather than averaging signals across a mixed population. The team integrated publicly available single-cell data from cells derived from the infrapatellar fat pad and synovium of knee osteoarthritis patients with bulk transcriptomic datasets from synovial tissue of patients and healthy controls. This dual approach gave them both the fine-grained resolution of single-cell analysis and the statistical power of larger bulk cohorts for validation.

A central innovation of the study was its use of multiple scoring algorithms to quantify lipid metabolism-related transcriptional activity in each individual cell. Rather than relying on a single gene or a single computational method, the researchers calculated composite scores across the single-cell dataset, then searched for genes whose expression tracked with those scores. They layered on additional filters: differential expression between osteoarthritis and control samples, protein–protein interaction network analysis, and five separate machine-learning approaches to prioritize candidate genes. Only genes that survived this cross-screening, and that could be validated in external bulk datasets and characterized at the single-cell level, advanced to the final stage of analysis.

The results pointed unambiguously toward the immune system. Among all the cell types in the fat pad and synovium compartment, myeloid cells, the family that includes monocytes, macrophages, and other innate immune cells, showed the most consistent increases in lipid metabolism-related transcriptional scores in osteoarthritis samples. This finding fits with a broader shift in the field’s understanding of osteoarthritis, in which macrophages infiltrating the synovium are increasingly seen as key drivers of joint inflammation and cartilage degradation. What the new study adds is the suggestion that these cells’ abnormal lipid handling may be part of what makes them inflammatory.

From fifty genes associated with the composite lipid metabolism scores, which were enriched in pathways involving lipid metabolic processes, lysosomal function, and PPAR signaling, a master regulator of cellular metabolism, three candidates emerged from the cross-screening: PLIN2, LTA4H, and PPT1. PLIN2 encodes a protein that coats lipid droplets inside cells, PPT1 is a lysosomal enzyme involved in lipid degradation, and LTA4H encodes leukotriene A4 hydrolase, an enzyme that sits at a critical junction in the production of leukotrienes, potent inflammatory lipid messengers. Of the three, LTA4H showed the strongest association with the lipid metabolism-related transcriptional scores, making it the study’s lead candidate.

Single-cell characterization revealed that LTA4H was predominantly enriched in myeloid cells, particularly monocytes and inflammatory macrophages, and that its expression was elevated in osteoarthritis myeloid cells and in cells with high lipid metabolism scores. The researchers then traced the relationships between cell populations using cell–cell communication analysis, myeloid subclustering, and pseudotime analysis, a computational technique that orders cells along a trajectory resembling a developmental or activation pathway. Together, these analyses linked LTA4H-positive myeloid cells to inflammatory signaling and to the transition of monocytes, which circulate in the blood and enter inflamed tissue, into inflammatory macrophages, the effector cells that amplify tissue inflammation. In other words, LTA4H appeared to mark, and possibly help shape, the very cells that fuel the inflammatory fire in the osteoarthritic joint.

To probe what LTA4H might be doing functionally, the team performed an in silico knockout analysis, a computational simulation of what happens to gene regulatory networks when a gene is removed. The analysis implicated LTA4H in regulatory programs related to leukocyte migration, chemotaxis, phagocytosis, and cytokine receptor binding, all processes central to how immune cells move into damaged tissue, engulf debris, and communicate inflammatory signals. The findings are biologically coherent: LTA4H catalyzes the final step in the synthesis of leukotriene B4, a lipid mediator that recruits neutrophils and monocytes to sites of inflammation. An enzyme that both handles inflammatory lipids and sits within lipid-metabolism-associated transcriptional programs is a plausible link between the two faces of the disease.

Crucially, the team did not stop at human datasets. In a rat model of knee osteoarthritis, they examined synovial tissue and found that LTA4H protein expression was increased, and that the protein spatially colocalized with CD68, a well-established marker of macrophages. This colocalization means that the LTA4H signal in the arthritic joint tissue was found precisely where macrophages reside, providing protein-level confirmation of the single-cell transcriptomic findings and suggesting that the monocyte-to-macrophage axis identified computationally has a physical correlate in diseased tissue.

The study’s significance lies less in any single gene than in the framework it establishes. By treating lipid metabolism as a measurable, cell-by-cell property and then using machine learning to trace its genetic underpinnings, the researchers have produced a myeloid-cell-centered model of the osteoarthritic fat pad and synovium, one in which LTA4H-linked lipid handling and inflammatory activation are intertwined. If future work confirms that LTA4H or its downstream leukotriene products actively drive disease progression rather than merely marking it, the enzyme could become a drug target in a field desperately short of them. Existing leukotriene-pathway drugs, developed for asthma and allergy, could in principle be repurposed for testing in joint disease, though any such leap would require extensive validation. For now, the study offers something equally valuable: a detailed molecular map of where inflammation and metabolism meet inside the osteoarthritic knee, and a demonstration that single-cell technology combined with machine learning can navigate that map with unprecedented precision.

Subject of Research: Single-cell transcriptomic and machine-learning analysis of lipid metabolism and inflammatory myeloid cell signatures in knee osteoarthritis infrapatellar fat pad and synovium

Article Title: Single-cell transcriptomics and machine learning characterize LTA4H-linked lipid metabolism and inflammatory signatures in the knee osteoarthritis IPFP/synovium compartment

Article References: Chen, X., Li, S., Wang, X., Chen, W., Qi, S., & Zhao, J. (2026). Single-cell transcriptomics and machine learning characterize LTA4H-linked lipid metabolism and inflammatory signatures in the knee osteoarthritis IPFP/synovium compartment. PLOS One, 21(10), e0360615. https://doi.org/10.1371/journal.pone.0360615

Image Credits: AI Generated

DOI: 10.1371/journal.pone.0360615

Keywords: knee osteoarthritis, single-cell transcriptomics, machine learning, LTA4H, lipid metabolism, myeloid cells, macrophages, infrapatellar fat pad, synovium, inflammation, leukotrienes, PLIN2