In 1980 Bill Roediger proposed that ulcerative colitis was a metabolic disease. He had measured butyrate oxidation in colonocytes from UC patients and found it severely impaired. His framing was simple. The colon epithelium runs on butyrate. When it cannot oxidize butyrate, the cells starve. When the cells starve, the barrier fails. When the barrier fails, the immune response that follows is downstream of the energy lesion, not its cause.
The field largely ignored him.
For four decades the dominant model has been autoimmunity driven by dysbiosis. The drugs that emerged target the immune system. Mesalamine, corticosteroids, thiopurines, anti-TNF biologics, integrin blockers, JAK inhibitors. They work for a meaningful fraction of patients. They fail for many. Roughly thirty percent of patients on infliximab do not respond. Of those who initially respond, a third lose response within a year. The field calls these patients refractory and reaches for the next molecule.
The 2024 and 2025 literature is starting to explain why.
What the new data show
A 2025 paper in Cell Death and Disease identified CKMT1, the mitochondrial creatine kinase isoform, as markedly decreased in colon tissues of UC patients. When the authors knocked it out in intestinal epithelial cells, the cells lost mitochondrial homeostasis, barrier function broke down, and apoptosis accelerated. This is not an immune cell phenotype. It is an epithelial energy phenotype. Nature
A 2024 Frontiers in Immunology study integrated 1,137 UC colon mucosal samples from twelve multicenter cohorts and used differentially expressed mitochondria-related genes to stratify UC into intrinsic subtypes. The subtypes were not defined by immune infiltrate. They were defined by mitochondrial state. nih
A 2025 paper in IBD Journal then closed the clinical loop, demonstrating the relationship between mitochondrial dysfunction and treatment response in ulcerative colitis, and concluding that mitochondrial homeostasis is central to anti-TNF response and that current methods mask these signals. The patients whose mitochondria are most disrupted are the ones who do not respond to the immune-targeted drug. Oxford Academic
A 2022 Harvard paper from the Haigis lab found dysregulated NAD+ metabolism in UC alongside altered mitochondrial state, linking mitochondrial dysfunction, inflammation, and NAD+ metabolism. NAD+ is the cofactor every colonocyte needs to oxidize butyrate to acetyl-CoA. NAD+ is also the molecule CD38 destroys when innate immunity is chronically activated. The connection is mechanistic and direct, and I have written about that cascade at length elsewhere. nih
A 2025 paper in the International Journal of Molecular Medicine reported that intestinal epithelial cells in DSS-induced colitis exhibit concurrent Drp1-mediated mitochondrial fission and ZBP1-dependent PANoptosis. Energy failure and programmed cell death are not parallel events. They are coupled. Spandidos Publications
The mechanism, end to end
The colon epithelium is a high-turnover, energy-starved tissue. Crypt enterocytes divide every three to five days. The differentiated colonocytes lining the lumen extract roughly seventy percent of their ATP from beta-oxidation of butyrate produced by fiber-fermenting commensals.
Three upstream factors compromise this oxidation.
Hydrogen sulfide produced by sulfate-reducing bacteria binds cytochrome c oxidase and inhibits the short-chain acyl-CoA dehydrogenase step of beta-oxidation. Sodium hydrogen sulfide inhibits n-butyrate oxidation in human colonocytes along the length of the colon in a pattern that closely mirrors the metabolic abnormalities observed in active ulcerative colitis, and increased H2S production in UC suggests sulfide is involved in disease genesis. Impaired butyrate oxidation and raised counts of sulfate-reducing bacteria in the colon of UC patients indicate that the disease may be induced or aggravated by hydrogen sulfide toxicity. Production goes up. Clearance goes down. Butyrate oxidation collapses. PubMedPubMed
CD38 activation then depletes NAD+. Without NAD+ the beta-oxidation pathway cannot run. SIRT3, the mitochondrial deacetylase, cannot maintain ETC complex activity. Reverse electron transfer at complex I begins producing ROS rather than ATP. The colonocyte is now leaking electrons.
Iron-sulfur cluster biogenesis falters under this oxidative load. Complexes I, II, and III all carry FeS clusters that are uniquely vulnerable to H2S and superoxide. Once these clusters are damaged, the ETC cannot recover without dedicated mitochondrial biogenesis. The DSS-colitis model recapitulates this pattern exactly: colonocyte butyrate oxidation is significantly reduced, glucose oxidation is significantly higher than in controls, beta-hydroxybutyrate falls, and lactate rises. The Warburg shift the field associates with inflammation is actually the colonocyte’s last metabolic option. PubMed
In vivo measurements in patients with active extensive UC confirm significantly lower butyrate oxidation and increased colonic permeability, with butyrate oxidation correlating negatively with clinical activity, and patients with inactive disease but decreased oxidation relapsing within weeks. PubMed
The epithelium then dies in patches. The microbiota that normally sit on a mucus layer now contact a barrier that cannot regenerate. The immune response that follows is appropriate. The cell death it is responding to is what the field has been treating as the disease.
Why this matters for treatment
The implication is uncomfortable for the standard of care. Suppressing the immune response in a patient whose colonocytes are dying of energy failure removes the inflammatory signal without addressing why the cells died. Some patients heal anyway because their underlying bioenergetic capacity is sufficient once the inflammatory amplification is dampened. Many cannot. These are the patients labeled refractory.
A mitochondrial framing produces different questions. What is the patient’s NAD+ status. What is the H2S burden. What is the SRB load on a stool panel. Is the sulfide detoxification pathway intact. Is iron handling sufficient to support FeS cluster biogenesis. Is the patient consuming enough fermentable fiber to produce butyrate, and if so, can the colonocytes oxidize it at all. None of these questions are answered by a fecal calprotectin or a colonoscopy score.
I have written elsewhere about why butyrate is not a supplement strategy in this state, why probiotics tend to worsen presentations that are upstream of dysbiosis, and why the CD38, NAD+, SIRT3 cascade is a more accurate lens on chronic illness than any single pathway. UC is the cleanest example of all three points converging in a single tissue.
The original Roediger framing was correct. The mechanism is now resolved. The clinical implications are still being worked out, and the field is doing it slowly.
For complex UC and IBD cases I work through these layers with clients and their treating physicians. Case analysis includes targeted testing for the upstream lesions, not just the downstream inflammation.