The story begins with glioblastoma, the most aggressive form of adult brain cancer. Despite surgery, radiation and chemotherapy, median survival remains roughly 14.6 months, and a major reason is the persistence of glioblastoma stem cells, which drive tumor initiation, progression and resistance to therapy. Because aberrant epigenetic programs help these cells maintain their plasticity, the researchers had previously screened a library of epigenetic chemical probes and found that QC6352 suppressed glioblastoma stem cell viability at nanomolar concentrations. In the new study, the team validated that potency across six patient-derived glioblastoma stem cell lines, including RKI1, MMK1, JK2, FPW1, HW1 and GBM6, with growth-inhibition potency values ranging from 3 to 133 nanomolar. Live-cell imaging showed that nearly 80 percent of treated cells arrested in the G0/G1 phase of the cell cycle, and the drug left normal human astrocytes unharmed up to 1 micromolar, a tenfold therapeutic window.

The cellular phenotype was striking. Treated cells ramped up expression of senescence and apoptosis genes, accumulated senescence-associated beta-galactosidase and Annexin V staining, and lost expression of stemness markers such as NES, SOX2 and SOX9, shifting toward a nonproliferative, differentiated state. In mice bearing orthotopic GBM6 xenografts, QC6352 at 100 milligrams per kilogram slowed tumor growth and extended median survival from 28.5 to 34.5 days without affecting body weight or blood counts. On the surface, everything pointed to a spectacular epigenetic vulnerability. But the team, led by Lenka Munoz of the University of Sydney together with Paul Workman, applied a discipline that is recommended far more often than it is practiced: testing multiple, structurally distinct inhibitors against the presumed target.

That discipline exposed the crack. Alongside QC6352, the researchers evaluated the orthogonal KDM4 inhibitor ML324, which they confirmed inhibits KDM4A with essentially the same biochemical potency, an IC50 of 326 nanomolar versus 355 nanomolar for QC6352. Yet the two compounds behaved nothing alike in cells. While QC6352 at 100 nanomolar completely repressed glioblastoma growth and triggered apoptosis, ML324 at 500 nanomolar had no impact whatsoever, and even at 5 micromolar it failed to block colony formation. Both compounds raised the levels of the histone marks H3K9me3 and H3K36me3 that KDM4 enzymes normally remove, but the team cautioned that such changes can also reflect broader heterochromatin and senescence responses, so they went further, directly testing whether either drug actually engaged KDM4 proteins inside living cells.

It did not. Cellular thermal shift assays detected no stabilization of overexpressed KDM4A even at 100 micromolar of either compound, and single-molecule tracking of SNAP-tagged KDM4A in live HeLa cells revealed no change in the mobility or chromatin-binding behavior of the enzyme after drug treatment. Expression analysis deepened the skepticism: KDM4A and KDM4B transcripts were present at low levels in only some glioblastoma stem cell lines, KDM4C and KDM4D were largely undetectable across twelve lines, and single-cell RNA sequencing of biopsies from 62 patients showed consistently low KDM4 expression in malignant cells. Proteomics found no KDM4 protein at all in A172 glioblastoma cells. Crucially, knocking out KDM4A, KDM4B or KDM4C individually with CRISPR-Cas9 left cells just as sensitive to QC6352 as before, and genome-wide CRISPR dropout screens using the 71,090-guide TKOv3 library identified 3,334 fitness genes in RKI1 cells without a single KDM4 family member among them.

The chemogenomic screens added another layer. Treating TKOv3-transduced cells with QC6352 at its 90 percent effective concentration of 70 nanomolar and at its 20 percent effective concentration of 25 nanomolar revealed 609 sensitizing and 278 resistance genes, with RNA splicing and mRNA and tRNA processing emerging as key biological processes. Again, no KDM4 enzyme met the threshold. The evidence was converging on an uncomfortable conclusion: the celebrated KDM4 probe was doing something else entirely in glioblastoma cells.

To find the real mechanism, the team turned QC6352’s transcriptional fingerprint into a search query against the Connectivity Map, comparing its signature with more than 20,000 pharmacologic perturbations across 71 cell lines. The top matches were brequinar, a known DHODH inhibitor, and floxuridine and raltitrexed, both inhibitors of thymidylate synthase. All three converge on a single metabolic pathway: de novo pyrimidine synthesis, the assembly line that produces the nucleotide building blocks of DNA, RNA and lipids. The logic of the pathway also explained a curious secondary hit, the cardiac glycosides digoxin, strophanthidin and bufalin, which indirectly block the uridine salvage route by suppressing sodium-dependent nucleoside uptake. When the researchers simply added uridine to the culture medium, it completely rescued glioblastoma cells from QC6352-induced growth arrest, the classic signature of a blocked pyrimidine supply.

Biochemistry then delivered the decisive numbers. In a recombinant enzyme assay, QC6352 inhibited human DHODH with an IC50 of 0.3 nanomolar, making it more potent than the established DHODH inhibitors BAY2402234 at 1.1 nanomolar and brequinar at 2.6 nanomolar. ML324, by contrast, showed no DHODH inhibition at all. Metabolomics sealed the case: within 24 hours of treatment, both QC6352 and BAY2402234 drove a roughly 10,000-fold accumulation of the upstream DHODH substrates dihydroorotate and carbamoylaspartate while depleting the downstream nucleotides UMP, UDP and UTP. A 1.95-angstrom cocrystal structure of QC6352 bound to human DHODH showed the compound occupying the hydrophobic channel leading to the active site, blocking ubiquinone cofactor binding, with its isonicotinic acid head forming polar contacts with R136, Q47 and Y356 and its tail packing into a hydrophobic tunnel. The same scaffold, in the KDM4A structure, chelates the catalytic nickel ion, explaining how one molecule can moonlight in two entirely different enzyme families.

The clinical implications came into sharp focus with zavondemstat, a structural analog of QC6352 now in trials for advanced and metastatic cancers. The R-enantiomer inhibited DHODH at 0.6 nanomolar, outperforming brequinar, farudodstat and BAY2402234, while its S-enantiomer was 40-fold weaker against DHODH, a chirality-dependent potency gap that tracked far better with the compounds’ cellular efficacy than their modest twofold difference in KDM4A inhibition. Uridine fully reversed the effects of both enantiomers, and expressing an inhibitor-resistant DHODH A58T mutant rescued cells from QC6352, zavondemstat and BAY2402234 alike, while uridine could not rescue the unrelated drugs GSK-J4 or doxorubicin. Guided by the crystal structures, the team even designed three new isonicotinic acid analogs that retain potent KDM4A inhibition, with IC50 values between 41 and 208 nanomolar, while showing no DHODH activity above 20 micromolar, providing the field with genuine KDM4-selective probes for the first time in this chemical series.

Beyond glioblastoma, the study is a cautionary tale for chemical biology as a whole. Despite long-standing guidelines urging the use of at least two structurally distinct probes plus inactive controls, reliance on a single compound remains common, and QC6352 shows exactly how that shortcut can produce confident but wrong conclusions about drug mechanism. DHODH itself is a validated therapeutic target, with the inhibitors leflunomide and teriflunomide already approved for autoimmune disease, though no DHODH inhibitor has yet been approved for cancer. The authors note that zavondemstat’s KDM4-independent mechanism does not diminish its clinical promise, and may even explain it, but they urge that ongoing trials be benchmarked against prior DHODH inhibitor studies. They also flag that DHODH should now be considered a frequent hidden target in phenotypic drug discovery, since multiple unrelated chemotypes keep landing on it. For glioblastoma, a disease desperate for options, the message is paradoxically hopeful: a drug already in the clinic may work through a metabolic vulnerability that is increasingly recognized as one of cancer’s most druggable Achilles heels.

Subject of Research: Target deconvolution of the KDM4 inhibitor QC6352 revealing DHODH as the driver of antiglioblastoma activity

Article Title: DHODH is the key target of the KDM4 inhibitor QC6352 in glioblastoma stem cells

Article References: Sterling, J. R., Holtsmark, J. R., Joun, G. L., Graus, M. S., Du, T. Y., Kuchibhotla, M., Johns, T. G., deWeck, A., Loo, L., Neely, G. G., Tran, K., Shard, C., Gomez, G. A., Patil, A., Shi, D. D., Losman, J.-A., Saxton, A. J., Baker, J. R., Lei, Z., … Munoz, L. (2026). DHODH is the key target of the KDM4 inhibitor QC6352 in glioblastoma stem cells. Nature Chemical Biology. https://doi.org/10.1038/s41589-026-02306-x

Image Credits: AI Generated

DOI: 10.1038/s41589-026-02306-x

Keywords: glioblastoma, DHODH, QC6352, zavondemstat, KDM4, pyrimidine biosynthesis, chemical probes, target deconvolution, epigenetics, drug mechanism, cancer metabolism, CRISPR screens

The two technologies at the heart of the study capture fundamentally different views of drug activity. The first, thermal proteome profiling, or TPP, grew out of the cellular thermal shift assay introduced in 2013. The principle is elegant: when a small molecule binds a protein, it often shifts the protein’s melting temperature, either stabilizing or destabilizing it. Researchers heat intact cells to a range of temperatures before lysis, then use mass spectrometry to measure how much of each protein remains soluble at each temperature. Coupling this to whole-proteome analysis, first demonstrated by Savitski and colleagues in 2014, allows an unbiased survey of thousands of drug-protein interactions in a single experiment. A typical whole-cell TPP experiment yields tens to hundreds of proteins with perturbed thermal stability, but these include not only direct drug targets but also proteins in complexes with, or downstream of, the true target, making it difficult to separate cause from consequence.

The second technology, Cell Painting, works from the opposite direction. Instead of measuring protein behavior directly, it captures the morphological fingerprint a drug imposes on a cell. Cells are treated with compounds and then stained with a fixed panel of fluorescent dyes labeling the nucleus, endoplasmic reticulum, nucleoli and cytoplasmic RNA, Golgi apparatus and actin cytoskeleton, and mitochondria. Software such as CellProfiler then extracts thousands of features per cell, producing a rich quantitative description of each treatment’s morphological perturbation. Because the assay is inexpensive and highly multiplexable, it can be applied to thousands of compounds at once. Compounds with similar mechanisms of action tend to induce similar morphological changes, so machine learning can infer the mechanism of an unknown compound by comparing it to well-annotated neighbors. The weakness is that small molecules often have many targets, and the accuracy of predictions depends heavily on the quality of available annotations.

The Uppsala team’s insight was that these two data types are complementary in a very specific way. TPP detects target engagement directly but misses proteins, sometimes because a peptide fails to ionize in the mass spectrometer, sometimes because a protein is more thermally stable than the tested temperature range, and sometimes because membrane-associated proteins sediment during centrifugation. Cell Painting, meanwhile, can suggest targets through compounds that produce similar shapes but cannot confirm physical binding. The researchers therefore built a pipeline that starts with the list of thermally stabilized or destabilized proteins from a TPP experiment and constructs a protein-protein interaction network using the STRING database, which aggregates experimentally scored interactions, curated pathway databases, and text mining. In parallel, the compound of interest is located within a Cell Painting dataset of 5259 compounds from the SPECS drug repurposing repository, profiled in U2OS bone cancer cells, and the known targets of its closest morphological neighbors are used to fill gaps in the network.

A key technical challenge was clustering the Cell Painting data reliably. Popular algorithms such as k-means and HDBSCAN are stochastic and can produce very different cluster structures on repeated runs, particularly on high-dimensional, poorly separated data like morphological profiles. The team turned to SC3s, a consensus clustering method originally developed for single-cell RNA sequencing, which combines principal component analysis with thousands of repeated k-means runs and summarizes the results into a consensus matrix. Running the algorithm 2000 times across five different cluster numbers, between 100 and 180, produced highly reproducible clusters for most test compounds. The cluster-derived target candidates were then merged with the TPP network, and betweenness centrality scores, a graph-theoretic measure of how often a node lies on shortest paths between others, were used to filter the merged network. Because drug targets have been shown to occupy central positions in interaction networks, the researchers reasoned that true targets would rank among the highest-scoring nodes.

To validate the approach, the team used publicly available TPP datasets for five well-characterized drugs: the BET bromodomain inhibitors (+)-JQ1 and I-BET151, the BRAF inhibitor vemurafenib, the ALK and MET inhibitor crizotinib, and the histone deacetylase inhibitor panobinostat. For (+)-JQ1, 67 proteins showed dose-dependent thermal shifts, and the resulting network correctly placed the known targets BRD4, BRD3, and BRD2, along with the documented off-target HADHA, among the top-ranked nodes. For I-BET151, 40 perturbed proteins yielded a network containing BRD4, BRD3, and BRD2. Intriguingly, although the two compounds share targets, only 11 perturbed proteins overlapped between them, suggesting they regulate different pathways downstream of target binding. Gene Ontology and Reactome enrichment analysis of the network communities recovered mechanisms consistent with the known biology of (+)-JQ, including lysine-acetylated histone binding and activation of cell death.

The validation extended further. For panobinostat, TPP alone detected only three of eleven known targets, but Cell Painting cluster analysis contributed eight more, and the combined network placed six targets among the top-ranked proteins. For vemurafenib, the primary target BRAF was entirely absent from the TPP data, while the off-target FECH was missing from the Cell Painting cluster, so only the integrated network captured both. The method correctly identified known targets and mechanisms for four of the five compounds. The exception, crizotinib, proved instructive rather than disappointing: its true targets ALK, ROS1, and MET are transmembrane proteins that classical TPP protocols rarely detect, and neither U2OS nor K-562 cells expressed ALK or ROS1. Instead, the analysis highlighted the Bcr-Abl pathway through the off-targets ABL1 and BCR, a biologically plausible finding given the high expression of these proteins in the K-562 leukemia cells used for TPP. The team then scaled the validation to a public Proteome Integral Solubility Alteration, or PISA, dataset covering 49 compounds, a high-throughput TPP variant in which all temperature treatments are pooled before mass spectrometry. The combined model achieved target prediction with a ROC AUC of at least 0.7 for 21 compounds, compared with 15 for PISA alone and 14 for Cell Painting alone, and outperformed the individual methods in the majority of cases.

As a proof of principle, the researchers applied the pipeline to sinomenine, a plant-derived alkaloid used in China to treat rheumatoid arthritis and pain, whose mechanism of action has remained largely obscure. They generated their own PISA data in U2OS cells treated with 10 and 30 micromolar sinomenine, finding 175 and 121 significant thermal shifts respectively, with 94 proteins affected at both doses. The integrated network revealed a strikingly multimodal profile. Among the highest centrality scores were the beta-2 adrenergic receptor ADRB2 and the NMDA receptor subunit GRIN2C, both endogenously expressed in U2OS cells, supporting an action on nervous system receptors consistent with the compound’s analgesic use and with previous mouse studies showing elevated GABA signaling. Enrichment analysis also flagged neurotransmitter receptor activity and chemical synaptic transmission. Equally intriguing was the appearance of cyclin-dependent kinase 2, CDK2, alongside pathways governing RNA binding, DNA replication, chromosome organization, and double-strand break repair, pointing to potential anti-tumor effects that corroborate earlier sinomenine studies. Additional communities implicated actin cytoskeleton organization, membrane trafficking, RHO GTPase signaling, and metabolic processes including the citric acid cycle, with the structural protein ACTB and the mitochondrial transporter SFXN1 scoring highly.

The authors are careful to note the method’s limits. It cannot detect a target unless that target shows thermal shifts or is annotated for morphologically similar compounds, and the workflow deliberately trusts TPP-derived nodes over Cell Painting-derived ones, which meant that in 7 of 15 cases a target seen only in imaging data failed to enter the final network. Cell line choice matters enormously, as the crizotinib case demonstrated, and the team recommends using the same cell line for both assays when investigating a compound with unknown targets. Clustering reproducibility also faltered for vemurafenib, the compound with the weakest morphological signal. Even so, the payoff is substantial: a typical mass spectrometry run covers only about half the proteome, roughly 10,000 of 20,000 canonical proteins, and TPP alone can leave researchers with 35 to 64 candidate targets to chase. The integrated workflow narrows that list to ten or fewer, promising to slash the time and cost of follow-up validation experiments. With the full pipeline written in Python and released on GitHub, the Uppsala team has handed chemical biologists a practical tool for turning two imperfect lenses on drug action into one considerably sharper picture, one that could accelerate both drug development and drug repurposing for compounds whose secrets have resisted decryption for decades.

Subject of Research: Integration of Cell Painting morphological profiling and thermal proteome profiling for drug target identification and mechanism of action inference

Article Title: Cell painting and thermal proteome profiling for inference of drug targets and mechanism of action

Article References: Johansson, C., Johansson, M., Kasi, P. B., Larsson, M., Jakobsson, P.-J., Göransson, U., Carreras Puigvert, J., Spjuth, O., & Jansson, E. T. (2026). Cell painting and thermal proteome profiling for inference of drug targets and mechanism of action. Molecular Systems Biology, 22(8), 1270-1291. https://doi.org/10.1038/s44320-026-00214-9

Image Credits: AI Generated

DOI: 10.1038/s44320-026-00214-9

Keywords: Cell Painting, thermal proteome profiling, drug discovery, mechanism of action, target deconvolution, protein-protein interaction networks, proteomics, sinomenine, PISA, mass spectrometry, chemical biology, Uppsala University