ISG15-Driven Müller Glial Reprogramming Emerges Early in Diabetic Retinopathy
Diabetic retinopathy is often framed as a vascular complication, but the earliest injury may begin in the retinal support network that keeps neurons and capillaries stable. In this study, an ISG15+ Müller glial subtype expanded in diabetic mouse retina and carried a distinctly inflammatory, interferon-stimulated signature. The finding matters because Müller glia sit at the center of retinal metabolism, antioxidant defense, and vascular support; when their internal stress programs shift, the surrounding microvasculature becomes easier to injure.
The work links microglial interferon signaling to a glial state that weakens the retina from within. Rather than treating Müller glia as a uniform reactive population, the data point to a spatially clustered subtype that appears in response to diabetic stress and contributes directly to early blood-retinal barrier breakdown.
Microglial IFN-β Appears to Set the Program in Motion
At six months after streptozotocin induction, ISG15 signal was detected in multiple retinal compartments, including Müller glia, vascular structures, and microglia. In the retinal glial compartment, the ISG15-positive cells were not spread evenly across the tissue. They appeared in clusters, suggesting a localized state change rather than a diffuse, pan-Müller response.
Transcriptomic reanalysis of a published single-cell RNA-sequencing dataset reinforced that pattern. Two parent clusters were annotated as Müller glia, and focused reclustering resolved six Müller subclusters. One of those subclusters carried a classic interferon-stimulated gene program, including Isg15, Ifit1, Ifit3, Irf7, and related markers. The cellular picture is straightforward: diabetic stress did not merely activate Müller glia in general; it selected for a specific interferon-responsive subtype.
One sentence captures the biological pivot. Microglial IFN-β drove the Müller cell state that seemed to start the barrier injury.
Primary Müller glia exposed to high glucose alone did not show an increase in basal free ISG15 or detectable ISGylation. The response emerged when the cells were treated with conditioned medium from high glucose-exposed microglia, and the same effect was reproduced by IFN-β. Inhibition of type I interferon receptor 1 blunted that response, placing interferon signaling upstream of ISG15 induction. In vivo, Ifnb1 RNA signals were associated with microglia, retinal Ifnb1 expression rose in diabetic retina, and microglial depletion reduced Ifnb1 expression while limiting the expansion of ISG15+ Müller glia.
Barrier Leakage Tracks With Loss of Müller Cell Antioxidant Defense
The downstream injury was not abstract. Müller glia–specific Isg15 deletion preserved blood-retinal barrier integrity, reduced acellular capillaries, and lowered Evans blue leakage. Those are concrete vascular readouts, and they point to a retina in which capillary support and endothelial containment are both failing less often when ISG15 is removed from Müller glia.
The same intervention also reduced oxidative stress, lipid peroxidation, and glutathione redox imbalance. That cluster of changes fits the biology of Müller glia, which normally buffer oxidative load and help maintain a stable metabolic environment for adjacent vascular cells. When that defense falters, endothelial cells are exposed to a more injurious milieu, and barrier fragility follows.
Conditioned medium from Müller glia altered endothelial behavior as well. Loss of ISG15 or disruption of ISGylation blunted interferon-driven oxidative stress in Müller glia and reduced Müller glia–conditioned medium–dependent endothelial migration and tube formation. In practical terms, the glial secretome shifted from supportive to disruptive, and endothelial cells responded by behaving in ways that are more consistent with unstable, abnormal remodeling.
PRDX6 ISGylation Links Interferon Signaling to Oxidative Collapse
Mechanistically, IFN-β promoted ISGylation of peroxiredoxin 6 (PRDX6) and drove its proteasome-dependent degradation without changing Prdx6 transcript levels. That detail is important because it places the injury at the post-translational level: the antioxidant protein was not being transcriptionally silenced, it was being marked for destruction. PRDX6 helps protect cells from oxidative damage, so its loss would be expected to leave Müller glia less able to neutralize reactive species and lipid peroxides.
This creates a clean mechanistic chain. Microglial IFN-β activates an ISG15 program in Müller glia, ISG15-dependent modification destabilizes PRDX6, and antioxidant defenses collapse. Once that happens, oxidative stress accumulates, lipid injury rises, and the surrounding microvasculature becomes more leak-prone.
Why this cellular sequence matters
Diabetic retinopathy has long been treated as a disease of capillaries, but capillary failure does not arise in isolation. The retina depends on glial cells to buffer redox stress, regulate metabolic exchange, and preserve endothelial quiescence. A Müller glial subtype that loses PRDX6 is not a passive bystander; it becomes a source of oxidative strain and a contributor to vascular instability.
That biology also helps explain why early diabetic retinal injury can emerge before advanced neovascular disease is obvious. A localized interferon response can reprogram a subset of Müller cells, and a small number of dysfunctional support cells can destabilize the larger neurovascular unit.
The authors provide a strong case for the IFN/ISG15 axis as an early pathogenic pathway in diabetic retinopathy. The study stays within preclinical territory, so the next question is whether the same ISG15-positive Müller glial state appears in human disease and whether targeting interferon-driven ISGylation can preserve retinal function without impairing necessary innate immune defense. Those are the clinical tests that now matter.