Inflammation writes a durable memory into pouch organoid stem cells

What the work claims

This is a primary study from the Cadwell group at the University of Pennsylvania, with clinical recruitment at NYU Langone, first-authored by Chaoting Zhou. The team built what they say are the first organoid models of pouchitis, the de novo inflammation that arises in the surgically constructed ileal J-pouch used to restore continence after colectomy for ulcerative colitis. Up to half of IBD patients develop pouchitis after this ileal pouch-anal anastomosis surgery, most within the first year, and its cause is unknown.¹

The claim has three layers. First, pouchitis organoids remember the inflamed state: they form organoids less efficiently than organoids from matched non-inflamed pouch tissue, that deficit persists through multiple passages and freeze-thaw cycles, and it traces to elevated apoptosis in the intestinal stem cell (ISC) compartment. Second, the memory is epigenetic: ISCs from inflamed pouch organoids show inflammation-driven chromatin remodeling dominated by AP-1 family factors such as c-Jun and by excessive STAT1 activity, and knocking down JUN in healthy organoids partially phenocopies the injury. Third, the memory is partly reversible: tofacitinib, an FDA-approved JAK inhibitor, rescues organoid formation and shifts chromatin back toward the non-inflamed state, though 42.6 percent of the remodeled regions stay open despite the drug.¹

How it works

The design is the strength. Twenty consenting J-pouch patients contributed paired biopsies from the pouch body (exposed to fecal stasis) and the pouch inlet (closer to native ileum), stratified by blinded endoscopic and histopathologic assessment, with Crohn's-like disease of the pouch explicitly excluded. Because eligibility for the surgery requires quiet ileum at the time of pouch construction, the authors can be confident the inflamed ISCs arose in previously healthy tissue.¹ Organoids were passaged three generations before any experiment, so what persists is stable, not acute stress.

The phenotype is internally consistent. Inflamed organoids showed impaired formation efficiency, and single-cell suspensions revealed that the CD24-high ISC population (the fraction competent to form organoids) was enriched for Annexin V-positive apoptotic cells. The blockade was specific to the intrinsic apoptosis pathway: the pan-caspase inhibitor Q-VD-OPh rescued formation, as did caspase-3 and caspase-9 inhibitors, while a caspase-8 inhibitor and the ER-stress reliever TUDCA did nothing. Differentiation was simultaneously skewed: more CHGA-positive enteroendocrine and MUC2-positive goblet cells, fewer APOA1-positive absorptive enterocytes, and the same skew was visible in the matched primary pouch tissue in vivo.¹

The molecular layer combines ATAC-seq, RNA-seq, and H3K27ac and H3K4me1 CUT&RUN on flow-sorted ISCs (3 to 4 samples per group), analyzed with a linear mixed-effects model that treats donor as a random effect, so matched-pair structure is respected. AP-1 motifs, c-Jun, and STAT1 dominate the differentially accessible landscape. On the intervention side, tofacitinib at 1 to 10 micromolar restored organoid formation efficiency and normalized differentiation (upadacitinib behaved similarly), and of the 7,508 chromatin regions that gained accessibility in inflamed ISCs, 57.4 percent closed again after the drug while 42.6 percent resisted it.¹

Where a skeptic should push

The most load-bearing assumption is that the persistent chromatin state is cell-intrinsic reprogramming of an unchanged stem-cell population. There is a competing explanation the paper cannot fully exclude: survival selection. Inflamed tissue loses ISC viability, so the organoids that do grow are descended from the subset of ISCs that survived, and the "memory" could partly be the fingerprint of that selected survivor population rather than of reprogramming in an identical cell. Sorted-ISC chromatin profiling argues for a cell-intrinsic component, and the JUN knockdown phenocopy is genuine causal evidence that AP-1 tone controls the phenotype; but the two explanations are not mutually exclusive, and the study does not quantify their relative contributions.

Second, the sample sizes are honest but small. The multi-omics rest on 3 to 4 samples per group drawn from a 20-patient cohort, and no power calculation was performed. The authors themselves list the gaps: organoids exclude the immune, stromal, and microbial compartments that drive real pouchitis, so the epithelial-intrinsic story is necessary but not sufficient; linking any single regulatory region to function is not established, and the AP-1 and JAK/STAT axis almost certainly acts at many loci in concert; and the tofacitinib reversal is incomplete, with over 40 percent of gained regions tofa-resistant.¹ A further caution: tofacitinib was applied chronically in culture at micromolar concentrations, which models sustained pathway blockade rather than the intermittent exposure a patient receives.

What this means for gut organoid disease models

The non-obvious implication cuts in both directions, and both matter for drug discovery built on patient-derived intestinal organoids. As an opportunity, this paper is a template for ex vivo reversal assays: take an inflamed-tissue organoid, apply a candidate drug, and read out whether the disease state (formation efficiency, lineage skew, chromatin state) normalizes. That is a genuinely personalized pharmacology experiment, and the fact that a clinically approved JAK inhibitor passes it while the readout is only partial is exactly the kind of honest, bounded result that distinguishes a predictive model from a cheerleading one. If organoid memory is this durable, biobanks of IBD and pouchitis organoids become longitudinal records of patient state, not just genotype repositories.

As a threat, the same finding means every screen run on organoids derived from previously inflamed tissue is standing on a hidden variable. Resting organoids, passaging them, even banking and reviving them does not erase the inflammatory imprint, so a nominally "healthy control" line from a patient with an inflammatory history may behave like an injured one, generating false disease associations and muddying compound effects. The field's habit of treating passage number as a proxy for stabilization assumes the tissue started neutral, and this study is direct evidence it does not. There is also a subtler hazard: partial reversal. If 42.6 percent of remodeled chromatin persists through a treatment that fully rescues the gross phenotype, then ex vivo "normalization" readouts can declare victory while leaving a durable, drug-resistant state in place, a setup that could overstate the depth of remission any therapy achieves.

The bottom line

Established: pouchitis organoids stably recapitulate an apoptosis-prone, secretory-skewed ISC state with AP-1 and STAT1-dominated chromatin remodeling; c-Jun is causally involved; and JAK inhibition rescues function and partially resets the epigenome.¹ Hypothesis: that this epithelial-intrinsic memory is the driver of human pouchitis rather than a consequence or survivor artifact of the inflammatory milieu, and that reversing it durably in patients is possible. What would confirm it: lineage-resolved evidence separating reprogramming from selection, co-culture or organoid-on-chip versions that reintroduce immune and microbial signals, and patient-level correlation between ex vivo tofacitinib reversal and clinical response. What would break it: demonstration that the chromatin state tracks a pre-existing ISC subtype rather than injury, or that the phenotype vanishes once the missing compartments are restored.