The IL-1-Like Cytokine IL-33 Is Constitutively Expressed in the Nucleus of Endothelial Cells and Epithelial Cells In VivoA Novel ‘Alarmin’?

Here’s the puzzle that grabbed immunologists by the collar: IL-33 lives a double life. Outside the cell, it behaves like a classic cytokine, talking to the ST2 and IL-1 receptor accessory protein complex on mast cells and type two helper T cells, kicking off a wave of inflammatory and tissue-repair signals. Inside the cell, though, it looks nothing like a roving messenger. It parks in the nucleus and binds chromatin, much like the danger-signal proteins IL-1 alpha and HMGB1. The first clue to that second identity goes back to its discovery as NF-HEV — a nuclear factor abundant in high endothelial venules, the specialized vessels that usher lymphocytes into lymph nodes. So which is it: a secreted alarm or a nuclear sentinel? Moussion, Ortega, and Girard set out to answer that by asking a simple, sweeping question: where is IL-33 actually sitting in human tissues, in health and in tumors? They didn’t nibble around the edges. They built a tissue atlas. Using two independent human tissue microarray collections — one with 45 normal tissues sampled in duplicate, and a second combining 31 normal and 10 tumor types — they stained thin, five-micrometer sections with a mouse monoclonal antibody to IL-33, called Nessy-1, and two independent polyclonals. Then they cross-checked identity with the usual landmarks: CD31 or von Willebrand factor to mark endothelium, desmin and alpha-smooth muscle actin for stromal cells, and immune markers like CD3, fascin, and CD68 to see who did not light up. The critical control was baked in. When they pre-absorbed the IL-33 antibodies with recombinant IL-33 peptide, the signal vanished. When they left the peptide out, the signal came roaring back. That simple swap told them the antibodies were seeing what they claimed to see. And when they counterstained nuclei with DAPI, which binds DNA, IL-33 piled up in the densest DNA regions — a strong hint that it was riding chromatin rather than diffusing in the nucleoplasm. Across this atlas, the headline is surprisingly clear: IL-33 is a constitutive nuclear resident in endothelial cells. Not induced and fleeting, but present at baseline in the nuclei of CD31 or von Willebrand factor positive endothelium from large vessels down to capillary beds in most organs they examined. In human tonsil, high endothelial venules glowed with nuclear IL-33, and that same pattern repeated across lymph nodes and the appendix. Widen the lens to non-lymphoid tissues and the theme holds: liver sinusoids, skeletal muscle capillaries, kidney peritubular capillaries, prostate microvessels, and the dermal vasculature carry nuclear IL-33. There are exceptions that prove the rule. Brain endothelium and kidney glomerular microvessels did not show nuclear IL-33, underscoring that not every bed plays by the same rules. Another negative is just as informative: vascular smooth muscle cells — the contractile layer in arteries — lacked nuclear IL-33 in vivo, even though cultured arterial smooth muscle has been reported to express IL-33 messenger RNA. That mismatch between petri dish and person matters because it tells us where the in-tissue reservoir really is. Now flip to tumors. You might expect chaos there, but one element is familiar: tumor vessels carry nuclear IL-33 too. In adenocarcinomas of kidney, stomach, liver, and pancreas — and in additional examples from lung, breast, and colon — CD31 or von Willebrand factor positive endothelial nuclei were IL-33 bright. Same nuclear confinement, same peptide-blockable signal. That consistency across malignant and non-malignant vasculature suggests IL-33 isn’t a quirky byproduct of inflammation; it’s part of endothelial identity in many contexts. Endothelium isn’t the only address, though. Think about where the body meets the world — skin, gut, upper airways. Those barrier epithelia also store IL-33 in their nuclei. Skin keratinocytes were a textbook example. So were epithelial cells lining mucosal surfaces and the secretory cells in stomach glands and salivary tissue. Here the story adds texture: the intensity varied a lot. Cell to cell, region to region, and even between individuals, nuclear IL-33 rose and fell, as if local cues were turning the dial. That variability doesn’t muddy the message; it tells you IL-33 is part of how barrier cells tune their readiness. Inside lymph nodes, there’s a second structural reservoir. Fibroblastic reticular cells — the desmin positive stromal cells that build the T-cell zones and guide traffic — showed nuclear IL-33. Many alpha-smooth muscle actin positive reticular cells did as well, though the two stromal markers didn’t overlap perfectly, and not every fibroblastic reticular cell was positive. Translation: the reticular network that organizes immune encounters also harbors a nuclear cache of IL-33, but it’s a mosaic, not a monolith. In contrast, the usual mobile immune suspects were quiet. CD3 positive T cells, CD68 positive macrophages, and fascin positive dendritic cells did not show constitutive IL-33 in vivo. Most fibroblasts in non-lymphoid tissues were negative too. One niche stood apart: some visceral smooth muscle in the gastrointestinal and urogenital tracts had weaker nuclear IL-33, a reminder that smooth muscle is not one thing. If you pause here and just take stock, a pattern jumps out. IL-33 marks structural cells that define tissue architecture — endothelium that lines vessels, epithelia that face the environment, stromal cells that scaffold lymphoid organs. And almost always, it’s in the nucleus. Moussion and colleagues looked hard for cytoplasmic or extracellular IL-33 in these cells under normal conditions and did not see it. They also didn’t see the telltale signs of the protein being chopped into a mature secreted form by the inflammatory protease caspase-1. That matters because other IL-1 family cytokines rely on caspase-1 to become active. Here, the absence of a clear caspase-1 processing signature and the nuclear confinement point to a different operating model. This is where the alarmin idea pays off. Proteins like HMGB1 and IL-1 alpha sit in the nucleus and bind chromatin. When cells die abruptly — due to trauma, necrosis, or severe infection — they spill their contents, and those nuclear proteins become danger signals outside the cell. HMGB1 can also be actively exported from some immune cells after heavy acetylation, but its most potent signal often follows passive release. The evidence here says IL-33 behaves more like that. Stored in the nucleus at baseline, it’s well placed to pour out of damaged endothelial or epithelial cells and suddenly become a cytokine that engages ST2 on mast cells and type two helper T cells. You can think of the nucleus as a lidded jar; break the jar, and the alarm floods the room. Does the body ever pull that alarm in real life? We don’t have a live replay in these tissue slices, but there’s a suggestive systemic clue. The soluble form of ST2 — essentially a decoy receptor that soaks up IL-33 — tends to spike in the bloodstream during severe physiological stress. Elevated soluble ST2 has been reported in sepsis and major trauma, and also in acute myocardial infarction and chronic heart failure, with signals noted in lung fibrosis, asthma, rheumatoid arthritis, and lupus. That’s not proof that nuclear IL-33 was just released in each case, but it sketches a consistent picture: when tissues are hurt, the IL-33–ST2 axis wakes up. Let’s come back to how we know all this is really IL-33 in nuclei, not a staining artifact. The team leaned on three pillars. First, specificity: block the antibodies with recombinant IL-33 peptide and the signal disappears; remove the block and it returns. Second, identity: co-staining with endothelial markers pins the nuclear signal squarely inside vessel-lining cells, not perivascular bystanders. And third, location within the nucleus: overlap with DAPI dense chromatin says IL-33 isn’t just drifting through; it’s associated with DNA rich domains. That trifecta supports the central claim without drowning us in protocol. One nuance worth sitting with is the heterogeneity. Brain endothelium stands out as IL-33 negative, as do the capillary tufts of kidney glomeruli. Meanwhile, kidney peritubular capillaries are positive. In stroma, not every fibroblastic reticular cell is on board, and barrier epithelia show patchiness. Biology loves gradients, and IL-33’s nuclear residency seems to follow them, perhaps reflecting differences in baseline turnover, microbial exposure, or mechanical stress across tissues. It’s a reminder to be careful when generalizing endothelium or epithelium as if they were single cell types. Stack the pieces and a cohesive picture emerges. As Moussion, Ortega, and Girard showed across more than 50 human tissues using two independent microarray cohorts, IL-33 is a constitutive, chromatin-associated nuclear factor in structural cells of the vasculature and barriers. In the quiet of normal physiology, it stays put. When the peace is broken, it’s perfectly positioned to act as an alarmin, leaving the nucleus of injured cells to engage ST2 and recruit type two skewed responses — including mast cell activation and T helper two chemotaxis — that can drive repair and, in some contexts, pathology. Where does that leave us? Two short steps beyond the data are tempting but tractable. First, causality. If nuclear IL-33 is the reservoir, then selective injury to IL-33 rich endothelium or epithelium in a controlled model should unleash ST2 dependent signaling measurable in minutes. Catching that early burst — before new protein is made — would clinch the alarmin role. Second, context. The atlas tells us who stocks IL-33, but not who uses it when. Pairing spatial transcriptomics for ST2 with acute injury could map the immediate conversation partners, from perivascular mast cells to tissue resident lymphocytes. Even without those next experiments, the core message is already useful. When you think about tissue injury — a heart attack, a burned airway, an inflamed gut — picture IL-33 not as a molecule summoned to the scene, but as one already there, tucked into the nuclei of the cells that form the tissue’s front line. Break the barrier, and the signal is instantaneous. That’s an elegant design for a body that has to move fast when things go wrong.