Plant Compound Ivangustin Pins Neuroinflammation on Covalent Hit at UBE2L3 Cys86

A sesquiterpene lactone drawn from the traditional Chinese medicine herb Xuanfuhua, Inula japonica Thunb., has been shown in a new study to quiet inflamed microglia by locking onto a single cysteine residue of a ubiquitin-carrying enzyme. The compound, ivangustin (IVA), covalently modifies the catalytic cysteine 86 of the ubiquitin-conjugating enzyme UBE2L3, choking off the ubiquitination events that the NF-κB pathway needs to switch on inflammatory genes. Reporting in Chinese Medicine, the research team combined cell biology, chemical proteomics, Mendelian randomization, and a mouse model of acute cognitive impairment to build what they describe as the first identification of a direct functional target for this small molecule in neuroinflammation.

The authors frame neuroinflammation as both a shared pathological hallmark and a therapeutic target across neurodegenerative disorders, including Alzheimer’s disease, Parkinson’s disease, amyotrophic lateral sclerosis, and Huntington’s disease. Microglia, the resident immune cells of the central nervous system, respond to inflammatory stimuli by releasing nitric oxide, reactive oxygen species, and pro-inflammatory cytokines through signaling cascades such as NF-κB and the MAPKs. Because activation of the IKK complex depends on polyubiquitination of the NF-κB essential modulator (NEMO) subunit, the team reasoned that intercepting that ubiquitination step could offer a route to suppressing microglial inflammation that differs from existing small molecules aimed at upstream kinases such as IKKβ.

In vitro, the evidence was consistent across two microglial systems. IVA dose-dependently inhibited lipopolysaccharide (LPS)-induced nitric oxide production in BV2 cells and in primary microglia isolated from postnatal day 1 C57BL/6 mouse pups, without significantly affecting cell viability in CCK-8 assays. Quantitative PCR and western blotting showed parallel suppression of LPS-driven mRNA and protein expression of the inflammatory enzymes iNOS and COX-2. The compound also dampened the transcription of NF-κB-targeted cytokines IL-1β, IL-6, and TNF-α and of the chemokines MCP-1, CXCL1, and CXCL9, mirroring the effect of the known NF-κB pathway inhibitor BAY11-7082 used as a positive control.

Signaling assays pinpointed where the blockade occurred. IVA significantly suppressed LPS-induced phosphorylation of IKKα/β, prevented phosphorylation and degradation of IκBα, reduced phosphorylation of the p65 subunit, and substantially prevented p65’s nuclear translocation in BV2 cells. Notably, the compound did not significantly inhibit LPS-induced JNK or p38 phosphorylation; a slight upward trend in the phospho-to-total ratios at higher concentrations was not statistically significant, leading the authors to conclude that the anti-neuroinflammatory effect is primarily associated with NF-κB suppression rather than MAPK modulation under their experimental conditions.

The electrophilic chemistry of the molecule proved essential. The α-methylene-γ-lactone moiety of IVA acts as a Michael acceptor, and liquid chromatography-mass spectrometry using β-mercaptoethanol as a thiol surrogate revealed a specific IVA-BME adduct at m/z 349.14, confirming selective thiol reactivity. Functionally, pre-incubation with the thiol donors BME or glutathione completely abrogated IVA’s inhibition of nitric oxide production and erased its suppression of IKKα/β and IκBα phosphorylation and IκBα degradation. In other words, the covalent reactivity that makes the compound reactive is also what makes it pharmacologically active in these cells.

To find the protein on the receiving end, the team used a cysteine-reactive biotin-iodoacetamide probe in competitive pull-down experiments on BV2 lysates, followed by LC-MS/MS. Among proteins showing fold changes greater than four, UBE2L3 stood out with a fold change of 7.04, and pull-down confirmed reduced probe labeling of UBE2L3 in the presence of IVA. Cellular thermal shift and DARTS assays further supported direct binding, with IVA lowering both the thermal stability and protease resistance of the enzyme. Critically, the functional relevance went both ways: knocking down UBE2L3 with siRNA markedly attenuated IVA’s inhibition of nitric oxide release and of IKKα/β and IκBα phosphorylation, indicating that the enzyme is required for the compound’s anti-neuroinflammatory action.

Mass spectrometry then localized the chemistry to a single residue. A diagnostic peptide from recombinant human UBE2L3 showed a +248.05 Da mass shift after IVA incubation, localized to fragment ions flanking the catalytic cysteine 86. Covalent docking modeled a Michael addition between the C11=C13 double bond of IVA and the Cys86 thiol, with a docking score of −5.8 kcal/mol and a carbon-sulfur distance of 2.1 Å, stabilized by a hydrogen bond between the compound’s hydroxyl group and the carbonyl of Glu118. One hundred nanosecond molecular dynamics simulations showed ligand-induced conformational rearrangement, heightened dynamics around Cys86, two to seven persistent hydrogen bonds, and a single dominant conformational state. Point mutation experiments sealed the case: a C86S mutation reduced IVA binding in pull-down assays, and in knockdown BV2 cells wild-type UBE2L3 restored the drug’s inhibitory effect on nitric oxide production while the C86S mutant did not. Co-immunoprecipitation showed that IVA substantially attenuated LPS-induced K63- and M1-linked ubiquitination of NEMO, the events crucial for IKK complex assembly.

Bioinformatic analyses placed UBE2L3 in a disease context, though with caveats the authors themselves emphasize. Mendelian randomization using peripheral blood eQTL data from the eQTLGen Consortium and UK Biobank outcomes found that each unit increase in genetically predicted UBE2L3 expression was associated with a 13.9% higher risk of Parkinson’s disease (OR 1.139, 95% CI 1.052–1.234, FDR-adjusted p = 0.017) and increased risk of cognitive impairment (OR 1.0004, 95% CI 1.0001–1.0006, FDR-adjusted p = 0.027), with sensitivity analyses showing no significant heterogeneity or pleiotropy. For Alzheimer’s disease the association was only nominally significant (OR 1.005, FDR-adjusted p = 0.676) and unstable in leave-one-out testing with just three SNPs, so the authors call that signal preliminary. GEO microarray datasets showed UBE2L3 mRNA upregulated in Alzheimer’s and Parkinson’s brain tissue, and its expression correlated positively with neuroinflammatory markers including NOS2, PTGS2, CCL2, CXCL1, and CXCL9, while IL-6 and TNF-α showed non-significant trends.

In mice, the compound showed behavioral and histological protection. Ten-week-old male C57BL/6 J mice received daily intraperitoneal IVA at 5, 10, or 20 mg/kg for four days, with dimethyl fumarate at 100 mg/kg as a mechanistically relevant positive control, before an acute LPS challenge of 2.5 mg/kg. IVA inhibited LPS-induced weight loss and elevations in spleen and brain tissue coefficients, ameliorated reductions in open-field movement, and markedly improved the Y-maze spontaneous alternation rate that LPS had depressed. Histology revealed that LPS reduced neuronal density and increased damaged neurons in hippocampal CA1 and CA3 regions, changes that IVA mitigated, while immunofluorescence showed suppressed LPS-induced Iba1 elevation, a microglial activation marker. ELISA showed reduced TNF-α and IL-6 in serum and cortex, and western blotting and immunohistochemistry confirmed attenuated IκBα phosphorylation and degradation and reduced hippocampal p65 phosphorylation, with cortical UBE2L3 levels unchanged across groups. IVA at 20 mg/kg produced protective effects comparable to dimethyl fumarate at 100 mg/kg in this model.

The authors are explicit about the limits. The acute LPS model does not recapitulate the chronic pathophysiology of Alzheimer’s or Parkinson’s disease, so the protective effects warrant validation in disease-specific models. Pharmacokinetic data, including brain penetration and bioavailability, and systematic toxicology are lacking; an in silico Deep-B3 prediction suggests IVA is blood-brain barrier permeable, but experimental confirmation is needed. Direct evidence of UBE2L3 enzymatic activity and NEMO ubiquitination in brain tissue after in vivo dosing is also missing, the eQTL data come from blood rather than brain, and as an electrophilic compound IVA carries potential off-target risks from its inherent cysteine reactivity. Within those boundaries, the study positions UBE2L3’s Cys86 as a druggable site, identifies UBE2L3 as a candidate therapeutic target for neuroinflammation, and presents IVA as a lead compound for further preclinical investigation rather than a treatment ready for any clinical use.

Subject of Research: Covalent inhibition of the ubiquitin-conjugating enzyme UBE2L3 by the natural compound ivangustin as a mechanism for suppressing neuroinflammation

Article Title: Ivangustin, a sesquiterpene lactone from Inula japonica Thunb., attenuates neuroinflammation through covalent inhibition of ubiquitin-conjugating enzyme UBE2L3

Article References: Qiao, M., Gan, H., Shi, Y., Yang, S., Liu, T., Shen, X., & Wu, W. (2026). Ivangustin, a sesquiterpene lactone from Inula japonica Thunb., attenuates neuroinflammation through covalent inhibition of ubiquitin-conjugating enzyme UBE2L3. Chinese Medicine, 21(1), Article 217. https://doi.org/10.1186/s13020-026-01488-9

Image Credits: AI Generated

DOI: 10.1186/s13020-026-01488-9

Keywords: ivangustin, Inula japonica, sesquiterpene lactone, UBE2L3, neuroinflammation, NF-κB, microglia, NEMO ubiquitination, Cys86, Mendelian randomization, LPS mouse model, natural products