The study, published in Frontiers in Acoustics by Sam Bleker and colleagues at HAWK University of Applied Sciences and Arts, the University Medical Center Göttingen, and the Medical Laser Center Lübeck, is notable less for a single dramatic number than for the unusual rigor of its temporal accounting. Rather than simply measuring how much tracer crossed a barrier after ultrasound exposure, the team measured transport during the sixty-second sonication window and again in the nine minutes that followed, treating these intervals as fundamentally different biological events. That distinction, they argue, has been missing from much of the in vitro literature.

The experimental system was deliberately reductionist. Human cerebral microvascular endothelial cells, the hCMEC/D3 line widely used as a workhorse model of the blood-brain barrier, were grown on ThinCert permeable inserts for six days until they formed continuous monolayers. Lucifer Yellow, a small hydrophilic fluorescent dye that normally crosses such barriers only slowly through paracellular routes, was placed in the donor compartment above the cells. The researchers then applied continuous-wave focused ultrasound at 1.2 megahertz and 80 volts peak-to-peak for exactly sixty seconds, with no microbubbles, no pulse modulation, and no amplitude ramping.

Characterizing the acoustic field proved to be its own challenge, and the team approached it with unusual candor. Using a calibrated PVDF hydrophone, they scanned frequencies from 0.1 to 2.0 megahertz and found that 1.2 megahertz produced the highest mean peak rarefactional pressure at the focal distance of 5.0 millimeters from the lens apex. Because direct measurement at the biological exposure voltage was not performed, they scaled pressures from the 100-volt calibration condition by a factor of 0.8, arriving at an estimated peak rarefactional focal pressure of 0.838 plus or minus 0.054 megapascals. The authors are explicit that this number describes the calibration geometry, not the actual pressure at the cell layer inside the well-and-insert system, where reflections, standing waves, and lateral spread can substantially reshape the field.

The central result was unambiguous in direction. Across eight independent paired experiments, each comparing an ultrasound-exposed monolayer with an unexposed control from the same culture preparation, the mean control-adjusted increase in Lucifer Yellow accumulation during the exposure interval was 0.283 nanomoles, with a 95 percent confidence interval of 0.185 to 0.380 nanomoles. All eight experiments showed the effect in the same direction, and the standardized effect size, Cohen’s dz of 2.43, was exceptionally large. After Holm correction for multiple comparisons, the p-value was 0.000475. Leave-one-experiment-out analyses confirmed that no single pair drove the result.

What happened next was equally informative. During the nine minutes after the ultrasound was switched off, the additional control-adjusted accumulation was only 0.035 nanomoles, with a confidence interval spanning zero and a Holm-adjusted p-value of 0.555. In other words, the transport enhancement was essentially confined to the period of active insonification. The cumulative difference at ten minutes remained positive, but the authors caution that this largely reflects tracer that crossed during exposure, not continued leakage afterward. They also stress that the absence of detectable post-exposure accumulation does not prove the barrier had fully recovered, since no formal equivalence test or later functional measurement was performed.

The viability data add a layer of caution. In five separate paired experiments, the team stained monolayers immediately after exposure with Calcein-AM and ethidium homodimer-1, standard live-dead indicators, and quantified four image-derived metrics using condition-independent segmentation thresholds. The results were mixed: some experiments showed apparent increases in certain metrics, others showed decreases, and no reproducible ultrasound-associated direction emerged. Standardized effect sizes were small, and the authors explicitly warn that these findings cannot establish biological safety or the absence of delayed or spatially heterogeneous injury, particularly given the small sample size and the lack of a deliberately damaged positive control.

Perhaps the most intriguing and least settled observation came from the cell-free experiments. In a preliminary session conducted before the cellular series, three paired comparisons of ultrasound-exposed and unexposed cell-free inserts, inserts coated with Matrigel but lacking cells, showed no comparable positive response during the exposure interval. All three paired differences were negative. This hints that the transport enhancement in the cellular experiments may depend on the presence of living endothelial cells rather than on ultrasound-driven convection through the membrane alone. But the authors are careful: because all three cell-free pairs came from a single session, they cannot reliably quantify or exclude non-cellular contributions, and independent replication would be needed to separate endothelial, membrane, and fluid-system effects.

What mechanism could explain the response? The authors decline to commit. Acoustic radiation force, oscillatory deformation of the cells, fluid motion, or local shear stress could activate mechanosensitive signaling pathways, alter cortical actin tension, or transiently reorganize cell-cell junctions. Membrane permeabilization or vesicular transport cannot be ruled out either. A thermal contribution is also possible, since the sixty-second continuous-wave exposure was delivered without temperature monitoring, and the liquids were not degassed. The absence of visible bubbles during exposure does not exclude microscopic cavitation, and no passive cavitation detection was performed. The study therefore establishes that something happened during insonification, but not precisely what.

The broader context makes the work more than a laboratory curiosity. Focused ultrasound combined with intravenously administered microbubbles has already progressed from preclinical studies to early clinical trials for temporarily opening the blood-brain barrier in conditions such as Alzheimer’s disease and amyotrophic lateral sclerosis. Rodent studies have also demonstrated targeted and reversible opening of the blood-retinal barrier. But those approaches depend on circulating microbubbles, which introduce their own variability and safety considerations. Demonstrating that continuous-wave focused ultrasound can modulate tracer transport across a human endothelial monolayer without any added bubbles suggests that a simpler, bubble-free paradigm might deserve closer investigation, at least under controlled in vitro conditions.

The authors are equally clear about the limits of generalization. The hCMEC/D3 line, while human and reproducible, exhibits lower electrical resistance and higher paracellular permeability than primary or stem-cell-derived barrier models, and it lacks the pericytes, glial interactions, basement membrane, and hemodynamic environment of a real neurovascular barrier. The cells are of cerebral, not retinal, origin, so the findings cannot be assumed to translate directly to the blood-retinal barrier, even though the two share fundamental endothelial features. The study also did not measure transendothelial electrical resistance or junctional protein organization, leaving baseline barrier tightness and post-exposure recovery uncharacterized by orthogonal measures.

What the study does deliver is a methodological template. By treating the paired experiment as the biological unit, correcting cumulative transport calculations for both decreasing compartment volume and tracer removed during repeated sampling, and analyzing exposure and post-exposure intervals separately, the researchers have built a framework that other groups can adopt to disentangle when, not just whether, ultrasound affects endothelial transport. Future work, they suggest, should include randomized well positions, applicator-coupled sham controls, direct acoustic characterization near the monolayer, temperature monitoring, cavitation detection, time-resolved electrical resistance measurements, and replication in human retinal microvascular endothelial cells and more complex multicellular barrier models.

For now, the finding stands as a carefully bounded but genuinely novel observation: a short burst of continuous-wave focused ultrasound, delivered without microbubbles, measurably and reproducibly increased the passage of a small hydrophilic tracer across a human brain endothelial monolayer, and it did so predominantly while the sound was on. Whether that transient window of enhanced transport can be harnessed for drug delivery to the brain or retina, and whether it can be achieved without collateral cellular stress, remains an open and compelling question for the field.

Subject of Research: Focused ultrasound modulation of tracer transport across human cerebral endothelial monolayers in vitro

Article Title: Focused ultrasound increases Lucifer Yellow transport across hCMEC/D3 endothelial monolayers

Article References: Bleker, S., Redeker, K.-E. M., Meußler, G., Brinkmann, R., Brockmöller, J., & Russmann, C. (2026). Focused ultrasound increases Lucifer Yellow transport across hCMEC/D3 endothelial monolayers. Frontiers in Acoustics, 4, Article 1968185. https://doi.org/10.3389/facou.2026.1968185

Image Credits: AI Generated

DOI: 10.3389/facou.2026.1968185

Keywords: focused ultrasound, blood-brain barrier, hCMEC/D3, endothelial monolayer, Lucifer Yellow, drug delivery, microbubbles, in vitro model, acoustic bioeffects, blood-retinal barrier, paracellular transport, continuous-wave ultrasound

The peptide, known as MOTS-c, is transcribed from the mitochondrial open-reading-frame of the 12S rRNA-c region. Since its discovery, MOTS-c has attracted intense interest as one of a handful of mitochondrial-derived peptides that can act as signaling molecules, influencing metabolism, stress responses and aging far beyond the organelle where they are made. What the Nanjing-led team has done is to trace its role in the eye with unusual precision. Using retinal tissue, they showed that MOTS-c is predominantly localized in retinal endothelial cells, the thin, tightly connected cells that form the inner lining of blood vessels and constitute the blood-retinal barrier, the critical interface that keeps the light-sensitive tissue protected and properly nourished.

That localization matters because the earliest events in diabetic retinopathy are vascular. Chronically high blood sugar injures endothelial cells, weakens the junctions between them, and allows fluid and blood components to leak into the retina. It also damages mitochondria within the endothelial cells themselves, promoting the accumulation of reactive lipid peroxides that attack cellular membranes. To test whether MOTS-c is a casualty of this process, the researchers exposed human retinal endothelial cells to high glucose concentrations and studied a streptozotocin-induced diabetic mouse model, a standard experimental system in which the drug destroys insulin-producing cells and produces sustained hyperglycemia. In both settings, high glucose reduced the expression of MOTS-c in the retinal vasculature, suggesting that the loss of this mitochondrial signal is part of the injury program that diabetes unleashes on the eye.

The therapeutic logic followed directly. When the team supplied MOTS-c to human retinal endothelial cells suffering high-glucose-induced injury, the peptide protected the endothelial barrier and ameliorated mitochondrial dysfunction. The same benefit appeared in the diabetic mice: MOTS-c treatment rescued vascular endothelial damage in the retina. These genetic and pharmacological approaches, applied both in cell culture and in living animals, established that the peptide’s dysregulation is not merely a byproduct of disease but a mechanistically meaningful contributor to the vascular breakdown that drives diabetic retinopathy.

The deeper question was how a 16-amino-acid peptide could accomplish this. To find out, the researchers turned to unbiased molecular profiling. Transcriptomic analyses, which survey changes in gene expression across the whole genome, and lipidomic analyses, which map the constellation of fats and phospholipids inside the cells, converged on a surprising pathway: MOTS-c increased the activity of phospholipase A2, a family of enzymes that cleave phospholipids, the molecules that form the backbone of every cellular membrane, and thereby regulated phospholipid metabolism. In other words, the mitochondrial peptide appears to reprogram how endothelial cells maintain and remodel their own membranes under diabetic stress.

Among the differentially expressed genes associated with this upregulated phospholipase A2 activity, one stood out. Phospholipase A2 group VI, known as PLA2G6, showed the most significant upregulation following MOTS-c treatment. PLA2G6, an intracellular calcium-independent phospholipase, is best known in other contexts for its role in lipid homeostasis and, when mutated, in neurodegenerative disease. In the diabetic retina, the new work assigns it a protective job: PLA2G6-mediated degradation of lipid peroxides served as a primary mediator of the endothelial barrier protection seen during MOTS-c treatment. Lipid peroxides are the corrosive products of unsaturated fatty acid oxidation, and their accumulation is a hallmark of ferroptosis and related forms of oxidative cell death. By accelerating their breakdown, PLA2G6 effectively detoxifies the membrane environment of stressed endothelial cells.

The causal chain was then tested in the most demanding way, by removing the link. When the researchers knocked down PLA2G6, the protective effect of MOTS-c against vascular endothelial damage collapsed, both in vitro and in vivo. The peptide could no longer defend the endothelial barrier without its enzymatic partner. This loss-of-function result elevates the finding from correlation to mechanism: MOTS-c acts upstream, PLA2G6 acts downstream, and the degradation of lipid peroxides is the functional bridge between them. It also suggests that patients or experimental systems with insufficient PLA2G6 activity might respond poorly to MOTS-c-based interventions, a consideration that could shape future strategies for patient selection.

Taken together, the study identifies a novel role for MOTS-c dysregulation in mediating vascular endothelial damage in diabetic retinopathy and proposes MOTS-c supplementation as a potential therapeutic approach, one that works by regulating phospholipid homeostasis and protecting the endothelial barrier through PLA2G6. The conceptual appeal is considerable. Rather than blocking a single growth factor after the vasculature has already begun to fail, a MOTS-c-based strategy would aim to reinforce the fundamental resilience of endothelial membranes early in the disease, when intervention could still prevent the cascade of leakage, inflammation and pathological vessel growth that ultimately destroys vision. It also adds to a growing appreciation that mitochondrial-derived peptides are not metabolic footnotes but active hormones with tissue-specific protective functions.

Substantial work remains before this biology reaches the clinic. The findings, published as an open-access article in Cell Death Discovery on 10 October 2026, were obtained in cell culture and in a mouse model of diabetes, and translating a mitochondrial peptide into a safe, durable therapy for human eyes will require solving questions of delivery, dosing, stability and long-term safety that the current study did not address. Human diabetic retinas, exposed to years of metabolic fluctuation, may behave differently from acute experimental hyperglycemia. Yet the study offers something the field has lacked: a defined molecular axis, running from a mitochondrial genome-encoded peptide through phospholipase A2 activity and lipid peroxide clearance to the integrity of the blood-retinal barrier. If that axis holds up in further preclinical and clinical testing, the smallest genome in our cells may turn out to hold one of the larger answers to diabetic blindness.

Subject of Research: The role of the mitochondrial-derived peptide MOTS-c and phospholipase A2 group VI in protecting retinal endothelial cells in diabetic retinopathy

Article Title: Mitochondrial-derived peptide MOTS-c rescues vascular endothelial damage via PLA2G6 in diabetic retinopathy

Article References: Liao, Y., Xiang, J., Xia, Y., Liu, C., He, C., Tong, J., Chen, Y., Yao, G., & Xie, Z. (2026). Mitochondrial-derived peptide MOTS-c rescues vascular endothelial damage via PLA2G6 in diabetic retinopathy. Cell Death Discovery. https://doi.org/10.1038/s41420-026-03384-5

Image Credits: AI Generated

DOI: 10.1038/s41420-026-03384-5

Keywords: MOTS-c, mitochondrial-derived peptide, diabetic retinopathy, retinal endothelial cells, PLA2G6, phospholipase A2, lipid peroxides, endothelial barrier, mitochondrial dysfunction, phospholipid metabolism, blood-retinal barrier, diabetes complications

Diabetic retinopathy is a devastating complication of diabetes, characterized by progressive damage to the microvascular network that nourishes the light-sensitive tissue at the back of the eye. For decades, clinicians have focused their attention on the blood vessels themselves, treating the leaky capillaries and abnormal new vessel growth with laser therapy, injections of anti-VEGF drugs, or surgery. But a growing body of evidence points to a quieter culprit: chronic neuroinflammation, driven in large part by microglia, the resident immune cells of the retina. When these cells become chronically activated in a high-glucose environment, they release a cocktail of pro-inflammatory cytokines that erode the structural and functional integrity of the retinal vasculature from the outside in.

The research team approached the problem with an unusually systematic strategy. Rather than testing piceatannol blindly, they combined network pharmacology with transcriptomic analysis to map the compound’s potential molecular targets against the gene signatures of diabetic retinopathy. The computational screen identified 159 overlapping targets shared between the polyphenol and the disease. Among these, one receptor rose to the top: C-C chemokine receptor type 5, or CCR5, a G protein-coupled receptor best known to immunologists for its role in directing immune cell traffic and, famously, as an entry co-receptor exploited by HIV.

Molecular docking studies predicted that piceatannol could bind directly to CCR5, and the researchers went on to validate that binding affinity experimentally. This mattered because the receptor’s ligand, the chemokine CCL5, was significantly upregulated in the diabetic retina. The picture that emerged was of a signaling cascade, CCL5 binding to CCR5 and triggering the pro-inflammatory transcription factor NF-κB, that could potentially be intercepted at its receptor by a dietary polyphenol. Single-cell and expression analyses added a crucial layer of specificity: CCR5 was found to be expressed predominantly in retinal microglia, placing the immune cells squarely at the center of the pathway.

To test the compound’s effects, the team built disease models at two scales. In the laboratory, they exposed microglia to high glucose concentrations to mimic the diabetic metabolic environment. In living animals, they induced diabetes in mice using streptozotocin, a chemical that destroys insulin-producing cells and produces sustained hyperglycemia. The results in the animals were striking. Mice treated with piceatannol showed restored retinal thickness, reduced formation of acellular capillaries, the ghost vessels that mark dead capillary segments in diabetic retinas, and attenuated vascular leakage, the pathological seepage of fluid and proteins that ultimately destroys vision.

The cellular mechanism behind these improvements centered on microglial polarization. Microglia, like their peripheral cousins the macrophages, can adopt broadly two functional states. The M1 phenotype is pro-inflammatory, unleashing tumor necrosis factor-alpha, interleukin-1 beta and interleukin-6, along with inducible nitric oxide synthase, all of which batter neighboring tissue. The M2 phenotype is restorative, promoting tissue repair through markers such as arginase-1, interleukin-10 and transforming growth factor beta 1. In the diabetic retina, the balance tips decisively toward M1. Piceatannol treatment, the researchers found, inhibited microglial activation and M1 polarization, reduced the secretion of pro-inflammatory cytokines, and promoted the expression of M2-associated markers, all through modulation of the CCL5/CCR5/NF-κB signaling pathway.

The downstream consequences for blood vessels were equally telling. In vitro experiments using human retinal endothelial cells showed that when high glucose pushed co-cultured microglia into their inflammatory state, the endothelial cells suffered: the tight junction proteins zonula occludens-1 and Claudin-5, which zip adjacent endothelial cells together to form the inner blood-retinal barrier, were degraded. Piceatannol suppressed the high-glucose-induced M1 polarization of the microglia and, as a result, preserved the expression of these tight junction proteins. In other words, the compound’s vascular protection appears to be indirect, mediated through the immune cells rather than through a direct effect on the endothelium itself.

The team then put the CCR5 hypothesis to a rigorous causal test. Using small interfering RNA to knock down CCR5 expression, and separately employing pharmacological activation of the receptor, they confirmed that CCR5 sits at the center of the protective mechanism. When the receptor was silenced, piceatannol’s benefits were diminished or recapitulated by the knockdown itself; when the pathway was artificially activated, the compound’s protective effects were counteracted. This pattern of gain- and loss-of-function evidence strengthens the argument that the CCL5/CCR5/NF-κB axis is not merely correlated with the disease process but is functionally required for the drug’s action.

The significance of the work lies partly in its translational framing. Piceatannol is a naturally occurring stilbenoid, meaning it is already present in the human food supply and has been studied for anti-inflammatory and vascular-protective properties in other contexts. Demonstrating a concrete, mechanistically defined role in diabetic retinopathy raises the possibility of repurposing a well-characterized molecule for a condition that currently affects tens of millions of people worldwide. The network pharmacology approach also offers a template for how natural products with dozens of potential targets can be systematically triaged, using docking, transcriptomics and weighted gene co-expression network analysis to converge on the handful of targets that actually matter.

Substantial caveats remain before patients should expect a piceatannol-based therapy. The findings come from cell culture and streptozotocin-induced mouse models, which reproduce key features of diabetic retinopathy but do not capture the full complexity of human disease, which unfolds over decades of metabolic stress. Questions about bioavailability, dosing, retinal drug delivery and long-term safety in a diabetic population remain open, and the study was funded by Chinese national and municipal science programs without reported involvement of commercial drug development. Still, by pinpointing a druggable receptor on retinal microglia and showing that a food-derived polyphenol can restrain the inflammatory cascade that dismantles the blood-retinal barrier, the study adds a compelling new chapter to the growing recognition that diabetic blindness is as much an immunological disease as a vascular one.

Subject of Research: The role of piceatannol in modulating microglial polarization via the CCL5/CCR5/NF-κB axis to protect retinal endothelial cells in diabetic retinopathy

Article Title: Piceatannol attenuates retinal endothelial cell dysfunction by modulating microglial polarization via the CCL5/CCR5/NF-κB axis

Article References: Tu, Y., Guo, Y., Zhang, Y., Sun, H., Zhu, X., Shen, G., Gu, Y., Xie, L., Song, E., & Zhu, M. (2026). Piceatannol attenuates retinal endothelial cell dysfunction by modulating microglial polarization via the CCL5/CCR5/NF-κB axis. Journal of Translational Medicine. https://doi.org/10.1186/s12967-026-08893-6

Image Credits: AI Generated

DOI: 10.1186/s12967-026-08893-6

Keywords: diabetic retinopathy, piceatannol, microglial polarization, CCR5, CCL5, NF-κB, retinal endothelial cells, blood-retinal barrier, neuroinflammation, tight junctions, network pharmacology, resveratrol

The original study, led by Li and colleagues, assembled an impressive array of experimental tools to implicate NIK. The team analyzed retinal transcriptomic data, exposed a human microglial cell line known as HMC3 to advanced glycation end products, the damaging sugar-derived molecules that accumulate in diabetes, and then used genetic knockdown of NIK alongside a pharmacological inhibitor called B022 to see what happened. They also set up co-culture systems in which microglia and endothelial cells, the building blocks of blood vessels, were grown together to model the crosstalk that occurs at the blood-retinal barrier. The results pointed in a consistent direction: when NIK activity was reduced, inflammatory signaling quieted down and the integrity of the barrier improved, suggesting that NIK participates in the inflammatory signaling relevant to retinal vascular injury.

But the interpretation of those results has become the flashpoint of the debate. Li and colleagues framed their findings largely through the lens of M1/M2 polarization, a long-standing model in immunology that treats macrophages and microglia as existing in two opposing states. In this binary picture, M1 cells are inflammatory warriors that pump out tumor necrosis factor-alpha, interleukin-1 beta, and interleukin-6, while M2 cells are the healers, producing anti-inflammatory mediators such as interleukin-10 and repairing tissue. According to this framework, diseases like diabetic retinopathy arise when the balance tips too far toward M1, and therapies work by pushing cells back toward M2. It is an appealingly simple story, and it has dominated the literature for nearly two decades.

The trouble, as Jing Chen of the People’s Hospital of Leshan argues in a letter to the editor, is that the study’s own data do not fit that simple story. In the AGE-treated HMC3 cells, both CD68-positive and CD206-positive cells increased, markers that under the binary model would be assigned to opposing phenotypes. At the same time, elevated levels of the inflammatory cytokines TNF-alpha, IL-1 beta, and IL-6 were accompanied by increased IL-10, the archetypal anti-inflammatory signal. Under a strict reciprocal switch model, these markers should move in opposite directions. Instead, they rose together, a pattern far more consistent with overlapping or coexisting activation programs than with cells flipping from one fixed identity to another. As Chen emphasizes, this is not a quibble over nomenclature imposed from outside; the discrepancy arises directly from the study’s own measurements.

There are also technical reasons why bulk measurements of a handful of canonical markers cannot settle the question. When a population of cells is measured in aggregate, it is impossible to determine whether apparently opposing markers are co-expressed within the same cells, whether they come from distinct cellular subsets, or whether they reflect transitions between transcriptional states. In retinal tissue, the problem is compounded because bulk measurements cannot separate resident microglia from infiltrating myeloid cells that flood in from the circulation as the barrier breaks down. And a change in a small panel of markers, however canonical, does not define a complete transcriptional state. The reduction in M1-associated markers and the increase in M2-associated markers after NIK inhibition therefore support a shift in inflammatory phenotype, but they do not by themselves establish a conversion between fixed M1 and M2 identities. What they are compatible with is something more interesting: a mixed activation landscape in which NIK influences one or more disease-associated microglial programs.

This reinterpretation is not happening in a vacuum. The microglia field has been undergoing a conceptual revolution, driven largely by single-cell RNA sequencing technologies that can profile the gene expression of thousands of individual cells at once. In 2022, a landmark consensus paper in Neuron by Paolicelli, Sierra, Stevens, and colleagues argued that the field had outgrown the M1/M2 dichotomy and should embrace a multidimensional, context-dependent view of microglial states. Under this framework, microglia do not occupy one of two poles but instead slide along continuous axes of gene expression, adopting transient, disease-specific transcriptional programs that cannot be captured by any short list of surface markers. The states observed in a dish, moreover, often bear little resemblance to the states that emerge in living tissue under disease pressure.

Single-cell studies in diabetic retinopathy specifically have made the case even more compelling. Wang and colleagues reported marked microglial heterogeneity and dynamic subtype changes in early experimental diabetic retinopathy, showing that the cellular landscape shifts as the disease progresses. Geng and colleagues, working in a rat model of non-proliferative diabetic retinopathy, identified four distinct microglial subtypes, including an SPP1-high population associated with oxidative stress, apoptosis, and injury to the inner blood-retinal barrier. And in human proliferative diabetic retinopathy, Gu and colleagues described MARCO-positive microglia with dual pro-angiogenic and pro-fibrotic properties, a phenotype that could contribute to both the abnormal vessel growth and the scarring that characterize the advanced disease. These studies differ in species, tissue source, and disease stage, and they do not define a single longitudinal trajectory, nor do they establish NIK as the driver of any of these states. But together they show why a binary framework may miss biologically relevant heterogeneity in diabetic retinopathy, and why the question raised by the NIK study needs sharper tools to answer.

The distinction matters because changes in the abundance of a particular microglial state and changes within a state are biologically different phenomena, and they could imply very different therapeutic effects. If a state-based model is correct, the goal of NIK inhibition would not simply be to suppress M1 and promote M2, but to reduce harmful NIK-dependent programs while preserving the homeostatic and reparative functions that microglia perform in the healthy retina. That is a far more nuanced therapeutic objective, and it demands far more precise experimental designs. Chen also raises a critical caveat about the original study’s in vivo work: because B022 was delivered intravitreally without microglia-specific targeting, the observed improvement in blood-retinal barrier integrity cannot establish that retinal microglial NIK is the cell-specific mediator of the effect. Other cell types in the eye, including endothelial cells and infiltrating immune cells, also rely on NF-κB pathway signaling, and the drug would have reached them all. The pharmacological benefit supports further evaluation of NIK as a therapeutic target, but it is not equivalent to microglia-specific causality.

The path forward, Chen suggests, is concrete and technically achievable. A practical first step would be to compare the retinal transcriptomic data already generated by the original study with published signatures of disease-associated microglial states in diabetic retinopathy, asking whether the genes altered by NIK inhibition overlap with known state markers. That could be followed by single-cell or spatial transcriptomic profiling of retinas after NIK inhibition, which would reveal whether the treatment changes the abundance of particular microglial subtypes, alters their transcriptional programs, or both. Even more decisive would be microglia-restricted manipulation of Map3k14, the gene encoding NIK, combined with lineage-resolved analysis that can distinguish resident microglia from infiltrating macrophages. Such experiments would directly test whether the enzyme acts within microglia to remodel pathogenic states, or whether its protective effects flow through other cellular targets entirely.

None of this diminishes the significance of the original findings; if anything, it sharpens them. The evidence that NIK participates in the inflammatory signaling that damages the blood-retinal barrier in diabetes remains intact, and the pharmacological data justify continued interest in NIK-directed therapy for diabetic retinopathy. What changes is the biological framing: rather than a dial that turns inflammatory cells from bad to good, NIK may be better understood as a regulator of pathogenic microglial state remodeling, a molecular lever that shifts the composition and behavior of a heterogeneous population of immune cells in the diabetic retina. For a disease that affects hundreds of millions of people worldwide and for which current treatments only slow, and do not reverse, the loss of vision, that refined framework could prove to be the difference between another failed anti-inflammatory strategy and a therapy that finally targets the right cells in the right state at the right time.

Subject of Research: The role of NF-κB-inducing kinase in microglial state remodeling and blood-retinal barrier dysfunction in diabetic retinopathy

Article Title: From M1/M2 polarization to disease-associated microglial states: refining the role of NIK in diabetic retinopathy

Article References: Chen, J. (2026). From M1/M2 polarization to disease-associated microglial states: refining the role of NIK in diabetic retinopathy. Journal of Translational Medicine, 24(1), Article 1190. https://doi.org/10.1186/s12967-026-08987-1

Image Credits: AI Generated

DOI: 10.1186/s12967-026-08987-1

Keywords: diabetic retinopathy, microglia, NIK, NF-κB-inducing kinase, M1/M2 polarization, disease-associated microglial states, blood-retinal barrier, single-cell transcriptomics, neuroinflammation, B022, retinal endothelial cells, MAP3K14

The conventional toolbox for estimating the postmortem interval, or PMI, has long relied on physical signs of decomposition: body temperature cooling curves, the onset and resolution of rigor mortis, lividity patterns, and in later stages, insect colonization and microbial succession. Each of these approaches has well-documented weaknesses. Ambient temperature swings, morgue storage, wrappings and encasement, and individual variation in physiology can all distort the timelines inferred from these markers. Researchers have even shown that rigor mortis can persist far longer than classic models assume when bodies are kept cold, and that maggot development continues during refrigerated storage, complicating entomological estimates. The result, as the review’s authors note, is that conventional methods are often compromised by environmental factors and variable preservation conditions, restricting their accuracy and practical utility.

Vitreous humor offers a way around many of these problems, and the reason is anatomical. The fluid is sequestered behind the blood-retinal barrier, a selective gatekeeper that tightly controls which substances pass between the bloodstream and the eye. Because of this isolation, the vitreous humor exhibits relative metabolic stability after death and resists contamination and bacterial invasion far better than blood, cerebrospinal fluid or tissue samples. It is also less susceptible to the postmortem redistribution of drugs and metabolites that plagues toxicological interpretation. These properties make it, in the words of the review, an optimal biological matrix for PMI estimation, one that can extend the effective time window during which reasonably precise estimates are possible.

The scientific pedigree of vitreous humor analysis stretches back more than a century, to structural studies of the gel in the 1860s, but its forensic career truly began in the 1960s. In 1963, investigators reported that potassium concentrations in the vitreous humor rise steadily after death, a finding quickly confirmed by parallel studies showing that the correlation could serve as a clock for the postmortem interval. The mechanism is elegantly grim: after circulation stops, cells lose their ability to maintain ion gradients, and potassium leaks from intracellular stores into the surrounding fluid at a roughly predictable rate. Later work added hypoxanthine, a breakdown product of ATP metabolism that accumulates as oxygen deprivation sets in, and showed that combining hypoxanthine with potassium and ambient temperature improves estimates further.

Beyond potassium and hypoxanthine, the review catalogues a rich cast of biochemical characters whose postmortem trajectories correlate with time since death. Sodium and chloride concentrations shift in patterns that have proven useful both for PMI estimation and for distinguishing saltwater drowning from immersion deaths unrelated to drowning. Calcium and magnesium rise as cellular membranes fail. Glucose, urea nitrogen, creatinine and uric acid provide complementary information, though glucose is complicated by the fact that it can be elevated in conditions such as ethylene glycol poisoning. Ammonium accumulates as proteins degrade, and has even been measured at crime scenes using microfluidic paper-based devices designed to bring thanatochemistry out of the autopsy suite and into the field. Free amino acids, lactate, and protein fragments analyzed by high-performance liquid chromatography and mass spectrometry round out an expanding analytical repertoire.

Historically, the mathematical treatment of these signals relied on linear regression equations, most famously the potassium-based formulas refined across generations of studies. These equations correct for ambient temperature, age and other covariates, and software tools have been built to make them accessible to practicing pathologists. Yet the underlying biology is not strictly linear. Electrolyte diffusion, enzymatic degradation, temperature-dependent reaction kinetics and individual variation all interact in complicated ways, and single-marker equations carry confidence intervals that can span hours or days. Flexible regression models and chemometric approaches have pushed accuracy further, but the review argues that the real leap forward comes from artificial intelligence.

Machine learning entered the field in earnest in 2002, when researchers paired capillary zone electrophoresis measurements of vitreous electrolytes with artificial neural networks to estimate the PMI, demonstrating that nonlinear models could capture dependencies that simple equations missed. Since then, the approach has gathered momentum. Neural networks and other algorithms have been applied to sodium and potassium data, to metabolomic profiles of ocular fluids, and to microbiome succession data from animal models. Recent studies combining the human metabolome with machine learning have reported substantially improved PMI predictions, and systematic reviews now document machine learning and metabolomics applications across multiple tissue types. The review emphasizes that these models do more than enhance predictive accuracy; they enable robust nonlinear time series analysis of the complex, interacting biochemical changes that unfold after death.

The authors’ forward-looking vision is a multimodal fusion framework that integrates multiple streams of information: vitreous electrolytes, hypoxanthine and lactate kinetics, amino acid and peptide degradation patterns, imaging data such as the postmortem changes visible in the eye on computed tomography, and contextual variables like temperature and body weight. By training machine learning models on such heterogeneous datasets, the framework would hedge against the weaknesses of any single marker while exploiting the complementary strengths of many. The review also points to enabling technologies that could make this vision practical at scale, including tandem mass spectrometry, capillary electrophoresis, electrochemical biosensors for hypoxanthine, and rapid field-deployable microfluidic devices. Such a framework, the authors argue, is poised to become an indispensable tool for PMI estimation in forensic investigations and medicolegal proceedings.

Challenges remain before the vision becomes routine practice. Machine learning models are only as good as the data they are trained on, and the forensic literature contains decades of measurements gathered with different instruments, protocols and populations. Standardization of sampling and analytical methods will be essential, as will validation across diverse climatic conditions, causes of death and antemortem physiological states, since factors such as pre-death hypoxia can independently elevate hypoxanthine levels. Interpretability is another concern; courts and defense attorneys will demand to know why a model produced a particular estimate. Still, the review makes a compelling case that the convergence of a uniquely stable biological matrix, a maturing analytical toolkit and powerful computational methods is transforming one of forensic science’s oldest quests. What began with a simple potassium measurement in the 1960s is evolving into a data-rich, algorithm-driven science of the dead, one in which the fluid of the eye may soon testify, with unprecedented precision, about the moment life ended.

The choice of vitreous humor over alternative biological matrices reflects a broader pattern in postmortem biochemistry, where researchers have explored pericardial fluid, cerebrospinal fluid, and even synovial fluid as potential sources of time-dependent markers. Each fluid has its own advantages and drawbacks. Cerebrospinal fluid, for instance, contains promising markers such as glial fibrillary acidic protein, neuron-specific enolase and S100B, but its biochemical composition can vary depending on the sampling site along the spinal column, a source of variability that complicates comparisons across studies. Pericardial fluid has proven valuable for metabolomic investigations and for detecting markers of antemortem drug exposure, yet it is more exposed to contamination and postmortem diffusion from adjacent tissues than the sequestered ocular environment.

Practical considerations also favor the eye in casework. Vitreous humor can be collected with a simple needle aspiration through the sclera, typically from both eyes, and the small sample volumes required are compatible with standard clinical biochemistry analyzers as well as more sophisticated platforms. The fluid’s clarity and low protein content reduce matrix effects in spectrophotometric and chromatographic assays, which is one reason potassium-based methods achieved early reproducibility. At the same time, forensic researchers have cautioned that sampling technique, the interval between collection and analysis, and storage temperature can all influence measured concentrations, underscoring the need for harmonized protocols.

Imaging-based approaches add a complementary, non-destructive dimension. Studies of postmortem computed tomography have documented progressive changes in the eye, and measurements of the radiodensity of vitreous humor and cerebrospinal fluid have been proposed as indicators of the time since death. Because such imaging can be performed before autopsy without altering the specimen, radiological markers could eventually be integrated alongside biochemical ones in the multimodal frameworks the review envisions.

The pursuit of a single universal formula for the postmortem interval has long been viewed with skepticism among forensic scientists, and this skepticism shapes the review’s argument for data-driven fusion. Rather than seeking one equation that fits all circumstances, the field appears to be moving toward context-aware models that weigh temperature history, individual physiology and multiple analytes simultaneously. If validated rigorously, such models would not replace the forensic pathologist’s judgment but would supply a quantified, uncertainty-aware estimate that can be communicated transparently in medicolegal settings, marking a meaningful shift in how time since death is established in practice.

Subject of Research: Postmortem interval estimation using vitreous humor biochemistry and machine learning

Article Title: From biochemical markers to machine learning: a review of postmortem interval estimation via vitreous humor analysis

Article References: Rao, M., Wang, C., Tian, Y., Tao, H., Liu, L., & Zeng, X. (2026). From biochemical markers to machine learning: a review of postmortem interval estimation via vitreous humor analysis. International Journal of Legal Medicine. https://doi.org/10.1007/s00414-026-03979-8

Image Credits: AI Generated

DOI: 10.1007/s00414-026-03979-8

Keywords: vitreous humor, postmortem interval, forensic science, machine learning, potassium, hypoxanthine, blood-retinal barrier, thanatochemistry, forensic toxicology, artificial neural networks, multimodal fusion, death investigation