Pericyte Snai1 Drives Pathological Retinal Angiogenesis Through Extracellular Matrix Remodelling

Pathological retinal neovascularization, the aberrant growth of new blood vessels that complicates diabetic retinopathy, retinal vein occlusion and retinopathy of prematurity, remains a leading cause of blindness worldwide, and a new study now points to an unexpected culprit hiding in the walls of the retinal microvasculature. Writing in Experimental & Molecular Medicine, researchers report that the transcription factor Snai1, when upregulated in pericytes, reprograms these mural cells into pro-angiogenic builders that remodel the perivascular extracellular matrix and thereby fuel disorganized vessel growth. Using the oxygen-induced retinopathy model in mice, the team showed that genetic deletion of Snai1 specifically in pericytes markedly reduced pathological neovascular tufts and vascular leakage while improving retinal function, all without disturbing physiological vascular development. The work also identified a pharmacologically tractable downstream node, the collagen crosslinking enzyme Lox, whose inhibition with an intravitreal injection phenocopied the genetic ablation and promoted what the authors describe as vascular normalization.

Pericytes are essential mural cells of the retinal microvasculature that maintain vascular homeostasis by regulating lumen diameter, preserving the integrity of the inner blood-retinal barrier, and communicating with endothelial cells through intercellular junctions and paracrine signalling. They also play a key role in vascular remodelling by restoring endothelial coverage and modifying the extracellular matrix, the protein meshwork that provides both a mechanical scaffold and biochemical signals for surrounding cells. Pericyte dysfunction is considered pivotal in retinal vascular disease: under ischaemic conditions pericytes detach from capillaries and alter their secretory profile, and with disease progression they may trans-differentiate into myofibroblasts, contributing to retinal fibrosis. Despite this recognition, the molecular characteristics of pericytes during pathological neovascularization and the mechanisms by which they regulate aberrant vascularization have remained unclear, and effective therapeutic targets have been lacking. Current clinical treatments based on anti-VEGF therapy face challenges including poor response, recurrence and unfavourable long-term outcomes, motivating the search for mechanisms beyond the VEGF pathway.

To dissect pericyte behaviour during pathological angiogenesis, the researchers turned to single-cell RNA sequencing data from the oxygen-induced retinopathy model, in which neonatal mice are exposed to 75 percent oxygen for five days and then returned to room air, triggering retinal ischaemia and subsequent neovascularization. From 363 pericytes identified in normal and OIR retinas, pseudotime trajectory analysis revealed three pericyte states along a connected differentiation route. State1 pericytes, designated quiescent, predominated in normal retinas and highly expressed genes related to aerobic glycolysis, gene-expression regulation and retina-specific functions. State3 pericytes, designated dysfunctional, expanded dramatically in OIR retinas and showed elevated expression of genes involved in protein translation, extracellular matrix-receptor interaction and cytoskeletal rearrangement. Kyoto Encyclopedia of Genes and Genomes and Gene Ontology enrichment analyses supported a pro-angiogenic phenotype, with blood vessel development among the enriched terms. Morphologically, pericytes in OIR retinas shifted from elongated and vessel-aligned to contracted and clustered at branch points, with enlarged cell bodies and region-dependent loss of coverage that was most severe within neovascular tufts.

Searching for the transcription factors driving this pathological transition, the team compared quiescent and dysfunctional pericytes and identified Runx1, Lef1, Foxc2, Snai1 and Dr1 as the top five upregulated transcription factors in the dysfunctional state. Foxc2 and Snai1 exhibited specific high expression within the pericyte population, and because the group had previously shown that Snai1 in endothelial cells promotes the early phase of ocular neovascularization, they focused on Snai1. Immunofluorescence confirmed minimal Snai1 expression in normal retinas but marked upregulation in OIR, where the protein localized predominantly to neovascular tufts and co-localized with pericyte nuclei. Notably, during physiological retinal vascular development Snai1 expression remained low through the stages examined. The finding extended to human tissue: in specimens from patients with diabetic retinopathy, Snai1 was detected at significantly higher levels in proliferative membranes than in largely avascular fibrous subretinal membranes, supporting its participation in disease-related angiogenesis.

The decisive functional test came from genetically engineered mice carrying floxed Snai1 alleles combined with tamoxifen-inducible Pdgfrb-CreERT, allowing pericyte-specific knockout. Analysis at postnatal day 7 revealed that conditional Snai1 deletion had minimal effects on developmental vascular outgrowth, morphology or pericyte coverage, indicating that the factor is dispensable for physiological angiogenesis. In the OIR model, however, Snai1 knockout mice exhibited a significant reduction in neovascular area at postnatal day 17, while the vaso-obliterated area remained unchanged, showing that pericyte Snai1 promotes pathological rather than compensatory vessel growth. Functional assessments strengthened the picture: fluorescein angiography and dextran injection revealed reduced vascular leakage, suggesting improved vessel integrity, and full-field electroretinography showed that deletion significantly increased the amplitudes of both a-waves, reflecting photoreceptor function, and b-waves, reflecting inner retinal function, to near-normal levels. The attenuated neovascularization was associated with decreased endothelial cell proliferation rather than altered endothelial apoptosis, and pericyte coverage improved significantly, contributing to recovery of vascular structure and function.

To probe how Snai1 acts within pericytes, the researchers isolated primary mouse brain microvascular pericytes and modulated Snai1 expression by adenoviral overexpression or small interfering RNA knockdown. Altering Snai1 levels did not markedly affect pericyte proliferation, migration or classical contraction- and synthetic-state marker expression. The critical experiment compared two co-culture systems with the bEnd.3 endothelial cell line. In an indirect system using pericyte-derived conditioned medium, endothelial cell migration, proliferation and tube formation were unaffected by Snai1 manipulation. In a direct system where pericytes and endothelial cells were mixed and cultured together, Snai1-overexpressing pericytes strongly enhanced endothelial cell proliferation and tube formation, whereas Snai1-knockdown pericytes attenuated endothelial activation. These results indicate that Snai1 does not substantially alter pericyte behaviour in a cell-autonomous fashion but instead promotes endothelial angiogenic activity specifically through a contact-dependent mechanism, implicating the shared extracellular interface between the two cell types.

That interface is the extracellular matrix, and the sequencing data pointed squarely at it. Dividing pericytes into Snai1-high and Snai1-low subpopulations from the single-cell data, the researchers found that Snai1-high pericytes were highly enriched for pathways related to matrix organization and remodelling and exhibited significantly more and stronger interactions with endothelial cells, predominantly mediated by collagen-integrin ligand-receptor pairs. RNA sequencing of pericytes after Snai1 knockdown showed markedly downregulated expression of multiple collagen genes, including Col4a1 and Col6a3, and of key collagen crosslinking enzymes such as Lox and Plod. Quantitative PCR and western blotting confirmed reduced mRNA and protein levels of collagen IV, collagen II, Lox and Plod. Because Snai1 is a transcription factor, chromatin immunoprecipitation assays confirmed its direct binding to the promoter regions of Col2a1, Col4a1 and Lox. Functionally, knocking down Col4a1 or pharmacologically inhibiting Lox activity with beta-aminopropionitrile reversed the pro-angiogenic effects of Snai1-overexpressing pericytes on endothelial cells, establishing collagen synthesis and crosslinking as the mechanistic bridge.

In vivo evidence aligned with the cellular mechanism. Masson’s trichrome and Picrosirius staining revealed a significant reduction in total collagen deposition in the retinas of pericyte-specific Snai1 knockout OIR mice, and immunofluorescence showed reduced perivascular enrichment of type IV collagen, the major basement membrane component, and decreased levels of Lox. In control OIR retinas, type IV collagen formed dilated, thickened sleeves around neovascular tufts, whereas knockout retinas exhibited a more normalized, continuous pattern, and transmission electron microscopy confirmed a thinner and more organized basement membrane structure. Human samples corroborated the axis: vitreous liquid from patients with diabetic retinopathy showed elevated levels of Col4a1 and Lox compared with controls, and proliferative membranes displayed higher expression of type IV collagen and Lox than non-neovascular subretinal membranes, supporting excessive collagen deposition and crosslinking as a hallmark of pathological retinal neovascularization.

The study also closed the loop on upstream regulation, uncovering a self-amplifying cycle. Conditioned medium from hypoxic endothelial cells increased Snai1 protein levels in pericytes without markedly affecting Snai1 mRNA, pointing to post-transcriptional control. Among candidate endothelial secreted factors, only ANGPT2 and TGFβ1, not PDGFB, raised Snai1 protein levels, and cycloheximide decay experiments showed that both factors slow the degradation of Snai1 protein rather than promoting its transcription. Blocking integrin α5β1 with ATN-161 abolished the ANGPT2-induced Snai1 upregulation, identifying the receptor mediating that signal. Therapeutically, a single intravitreal injection of beta-aminopropionitrile at postnatal day 12 significantly reduced neovascular and vaso-obliterated areas, diminished pathological tufts, reduced leakage and improved electroretinographic responses in OIR mice with no apparent retinal toxicity, and the drug showed consistent efficacy in a laser-induced choroidal neovascularization model. The authors note limitations, including the modest transcriptional similarity between the OIR model and human proliferative diabetic retinopathy membranes, the unresolved downstream pathways activating endothelial cells, and the translational challenge of targeted delivery to pericytes. Still, by framing the Snai1-extracellular matrix axis as a route to reconstructing the vasculature toward a more physiological state rather than merely pruning vessels, the work offers a fresh perspective on vascular normalization therapy for blinding retinal disease.

Subject of Research: The role of pericyte Snai1 and extracellular matrix remodelling in pathological retinal neovascularization

Article Title: Pericyte Snai1 ablation prevents pathological angiogenesis in retinopathy via extracellular matrix remodelling-mediated inhibition of endothelial cell activation

Article References: Zhang, Q., Tang, B., Su, Y., Liu, P., Liang, L., Yang, Z., Zhang, P., Tian, C., Wen, T., Duan, Y., Jia, D., Zhou, Y., Dou, G., Han, H., Wang, Y., Sun, J., & Yan, X. (2026). Pericyte Snai1 ablation prevents pathological angiogenesis in retinopathy via extracellular matrix remodelling-mediated inhibition of endothelial cell activation. Experimental & Molecular Medicine. https://doi.org/10.1038/s12276-026-01874-1

Image Credits: AI Generated

DOI: 10.1038/s12276-026-01874-1

Keywords: pericytes, Snai1, retinal neovascularization, extracellular matrix, collagen crosslinking, Lox, oxygen-induced retinopathy, endothelial cells, vascular normalization, diabetic retinopathy, ANGPT2, TGF beta 1