The path to the new inhibitors began not with NVL but with a mystery molecule. As part of a high-throughput viability screen of 99,599 compounds in the colorectal cancer cell line HCT116, the researchers identified MM17, a dibenzothiazepinone that killed cancer cells with a half-maximal inhibitory concentration of 0.46 micromolar. What MM17 actually targeted, however, was unknown. To find out, the team turned to forward genetics, a strategy that lets the cancer cells themselves reveal a drug’s mechanism. Using a derivative cell line in which an auxin-triggered system degrades the DNA mismatch repair protein MLH1, thereby raising the mutation rate, the researchers exposed mutagenized cells to escalating doses of MM17 and waited for resistant survivors to emerge.

The screen delivered 59 independently derived resistant clones, each carrying a unique barcode, with resistance ranging from roughly 2.5-fold to 33-fold higher than parental cells. Whole-exome sequencing of 15 representative clones pointed to recurrent mutations in just two genes: TERT in three clones and, strikingly, NVL in all 15. Targeted sequencing across the full clone collection confirmed NVL mutations in 57 of 59 clones, overwhelmingly clustered in the D1 AAA+ domain of the protein. When the team used CRISPR-Cas9 to introduce specific substitutions at residues H304 and R403, the edited cells became resistant to MM17 while remaining fully sensitive to paclitaxel, demonstrating that NVL mutations alone were sufficient to confer resistance.

Genetic evidence alone cannot prove direct target engagement, so the researchers purified recombinant NVL and measured its enzymatic activity in vitro. NVL is a hexameric AAA+ ATPase, an ATP-driven motor that remodels ribosomal RNA and ejects transient assembly factors during the maturation of the 60S subunit. MM17 inhibited the ATPase activity of wild-type recombinant NVL with a half-maximal effective concentration of 0.12 micromolar, and the R403W variant, which retains normal catalytic activity, showed a 28-fold reduction in drug sensitivity that mirrored the cellular phenotype. A pair of analogs differing only in the position of a single methoxy group produced concordant structure-activity relationships in cells and in purified protein, providing convergent biochemical confirmation that NVL is the direct target.

To see exactly where the drug binds, the team turned to cryo-electron microscopy. A truncated, catalytically inactive NVL construct formed stable hexamers, and a reconstruction of the MM17-bound complex reached 3.06-angstrom resolution. The structure revealed the staircase architecture typical of type II AAA+ ATPases, with pore loops from five subunits gripping a peptide in the central channel. MM17 density appeared in two equivalent pockets within the D1 domains of subunits E and F, hydrophobic clefts formed by the central beta-sheet and the alpha0 helix. The dibenzothiazepinone core of the drug sits buried among residues including M280, M284, P297, L324, R403 and L405, four of which are either resistance sites themselves or lie adjacent to them. Comparison with an apo structure at 2.83 angstroms showed that the pocket does not pre-exist in the unbound enzyme; instead, MM17 binding swivels the alpha0 helix and pries open the D/E subunit interface, allosterically disrupting nucleotide coordination at a key site for ATP hydrolysis.

With target engagement established, the researchers traced the downstream consequences inside cells. Polysome profiling of MM17-treated HCT116 cells showed shrinking 60S and 80S peaks, accumulating free 40S subunits and the appearance of half-mer polysomes, a classic signature of stalled 40S subunits waiting at translation start sites for partners that never arrive. Northern blots revealed accumulation of the 32S precursor RNA and loss of the 12S intermediate, pinpointing the block to late nucleolar maturation. Using SNAP-tagged ribosomal proteins and heterochronic labeling, the team showed that newly synthesized eL36, a 60S protein, failed to incorporate into ribosomes and piled up in nucleolar foci, while 40S production proceeded largely unimpaired. All of these defects vanished in the NVL(R403W) cells, confirming they stemmed from on-target inhibition.

Blocked 60S assembly triggers a well-characterized ribosome biogenesis checkpoint. When assembly falters, free 5S ribonucleoprotein particles accumulate and bind the zinc finger domain of the E3 ubiquitin ligase MDM2, preventing MDM2 from tagging the tumor suppressor p53 for destruction. MM17 treatment raised p53 levels within six hours, followed by induction of the p53 target p21, and this response was absent in resistant cells. Crucially, cells engineered to express an MDM2 C311F mutant, which cannot bind 5S RNP but retains other p53-regulatory functions, failed to activate p53 in response to MM17 while responding normally to the DNA-damaging drug etoposide. The checkpoint, not genotoxic stress, was the driver. Cell-cycle analysis showed arrest at both G1/S and G2/M, with the G2/M block fully dependent on p53 and a delayed G1/S arrest persisting even in p53-knockout cells, hinting at a parallel, p53-independent pathway that monitors ribosome output.

To ask whether NVL is genuinely required for tumor maintenance, the team built a system for acute, drug-induced NVL degradation using an auxin-inducible degron knocked into the endogenous NVL locus. In xenograft tumors, pharmacologic NVL depletion raised p53 staining, cut the proliferation marker Ki67 and, over two weeks of treatment, inhibited tumor growth by 96.2 percent in p53-wild-type tumors and 85.2 percent in p53-knockout tumors. This established NVL as a therapeutic target regardless of p53 status and set a benchmark for what full target engagement can achieve.

MM17 itself, however, was a poor drug candidate, being metabolized by mouse liver microsomes with a half-life of just 2.68 minutes. A medicinal chemistry campaign produced MM927, which improved microsomal stability more than ninefold, showed roughly tenfold greater potency in cell viability assays (IC50 of 0.053 micromolar) and fivefold stronger inhibition of NVL ATPase activity. A 2.86-angstrom cryo-EM structure confirmed that MM927 occupies the same pocket with a near-perfectly aligned core, its benzoxazolone moiety replacing MM17’s furan ring and its rigid butynyl tail stabilizing the alpha0/beta1 loop. Photoaffinity labeling with a clickable analog identified NVL and only nine additional proteins as binding partners, and notably p97, a closely related AAA+ ATPase, was not among them. In a 21-day efficacy study, twice-daily MM927 at 35 milligrams per kilogram inhibited growth of NVL-wild-type xenografts by 56.5 percent while leaving resistant NVL(R403W) tumors untouched, and treatment was well tolerated with no notable changes in body weight, hemoglobin, platelets or liver enzymes. In a model of MOLM-13 acute myeloid leukemia, MM927 reduced leukemic burden sevenfold by bioluminescence imaging and depleted leukemia cells from bone marrow and peripheral blood, again without overt toxicity.

Perhaps the most provocative finding concerns the clinical drugs 5-fluorouracil and oxaliplatin, mainstays of colorectal cancer chemotherapy long classified as DNA-damaging agents. Both are now known to also disrupt ribosome production, but their relative contributions have been unclear. The researchers found that at fully inhibitory concentrations, 5-FU, oxaliplatin and the RNA polymerase I inhibitor CX-5461 all induced robust gammaH2AX staining, a marker of DNA double-strand breaks, whereas MM17 and MM927 did so not at all, even at concentrations far exceeding their IC50 values. A PRISM screen across 858 cancer cell lines then revealed that 5-FU and its prodrug carmofur showed the strongest correlations with MM17 sensitivity among 1,183 compounds tested, with oxaliplatin close behind. The implication is striking: the clinical efficacy of these chemotherapies may partly derive from ribosome-directed effects, and a molecularly defined NVL inhibitor could capture that vulnerability while sparing patients the genotoxic collateral damage. With MM927 as a first-in-class lead, ribosome assembly has moved from an attractive hypothesis to a druggable target, and the authors note that next-generation derivatives with improved solubility and lower protein binding may push efficacy closer to that of complete NVL depletion.

Subject of Research: Selective small-molecule inhibition of the NVL AAA+ ATPase to disrupt ribosome biogenesis and suppress tumor growth

Article Title: A selective NVL inhibitor suppresses tumor growth by targeting ribosome biogenesis

Article References: A selective NVL inhibitor suppresses tumor growth by targeting ribosome biogenesis. (n.d.). https://doi.org/10.1038/s41589-026-02324-9

Image Credits: AI Generated

DOI: 10.1038/s41589-026-02324-9

Keywords: NVL, ribosome biogenesis, cancer therapy, AAA+ ATPase, 60S subunit, p53, nucleolar stress, cryo-EM, MM927, forward genetics, drug discovery, colorectal cancer

A team led by Timo Jenne and Axel Mogk at Heidelberg University has now taken a significant step toward drugging this machine in S. aureus. Writing in the open-access journal iScience, the researchers report the first examples of small-molecule activators of the S. aureus ClpC unfoldase, identified through a high-throughput screen of roughly 110,000 chemically diverse compounds. The work demonstrates that the protease, normally held in a tightly regulated resting state, can be chemically forced into persistent, uncontrolled activity — a strategy that in principle turns the bacterium’s own protein-quality-control machinery against it.

The logic of the approach rests on how ClpC works. The protein is an AAA+ ATPase, a ring-forming molecular motor that, in its active hexameric state, unfolds substrate proteins and threads them through a central channel into the barrel-shaped serine peptidase ClpP, where they are chopped into fragments. In its basal state, however, ClpC is inactive: it forms an inactive resting state composed of two half-spirals, stabilized by head-to-head interactions of its coiled-coil middle domains and by contacts between its N-terminal domain (NTD) and the ATPase domains. Activation normally requires adapter proteins such as MecA or McsB, or the binding of substrates tagged with phosphoarginine, a degron recognized by dedicated pockets in the NTD. Crucially, the adapters themselves are degraded by the protease, creating a negative feedback loop that restricts activity to times when substrates are present.

Because deregulating this machinery is toxic — as shown by acyldepsipeptide antibiotics that hijack ClpP directly, and by natural cyclic peptides such as cyclomarin A that deregulate the mycobacterial ClpC1 homolog — the Heidelberg team reasoned that small molecules capable of locking S. aureus ClpC into its active state could similarly disrupt protein homeostasis. They screened their library at a single concentration of 40 micromolar, monitoring degradation of the fluorescent model substrate FITC-casein by ClpC and ClpP. Around 900 preliminary hits emerged, which the team prioritized by chemical clustering, keeping only compound families with multiple active analogs. Fifty-seven representative compounds were repurchased for rigorous biochemical validation.

About half of those failed to reproduce their initial activity and were discarded as false positives, but eight bona fide activators survived. The most potent, compound 8, boosted ClpC/ClpP proteolysis of FITC-casein nearly eightfold, approaching the activation achieved by the natural adapter MecA. Dose-response curves placed the compounds’ potencies in the low-to-moderate micromolar range, with EC50 values as low as 4.4 micromolar for compound 10. Importantly, the compounds activated degradation of a native substrate as well: the bacterial cell division protein FtsZ was broken down 1.7- to 4.2-fold faster in their presence, with compound 30 driving essentially complete FtsZ degradation within two hours. All eight compounds also stimulated ClpC’s ATPase activity, and the degree of ATPase stimulation correlated with proteolytic activation, indicating that the molecules work by overriding the protease’s built-in activity control rather than by an assay artifact.

Structural and biophysical experiments then revealed what the compounds do to the machine. Using an ATPase-deficient ClpC variant to trap complexes, analytical size-exclusion chromatography showed that compounds 8, 10, and 41 promote the formation of ClpC/ClpP assemblies. Negative-stain electron microscopy revealed that these assemblies are strikingly large and heterogeneous: instead of the canonical single-hexamer-on-ClpP complex seen with MecA, the compounds drive the formation of ring dimers, tetrahedral arrangements of four ClpC hexamers, and even larger networks of ClpC rings interconnected by ClpP, which offers two docking sites for unfoldase hexamers. These architectures mirror those previously observed with the toxic cyclic peptide cyclomarin A and with phosphoarginine-tagged substrates, suggesting that different deregulating ligands all funnel ClpC through a common activation pathway linked to higher-order assembly.

Where do the compounds bind? Every piece of evidence pointed to the N-terminal domain, the protease’s central allosteric hub. A ClpC variant lacking the NTD was completely unresponsive to all eight compounds, and adding an excess of isolated NTD sequestered the molecules and abolished their stimulatory effect. Computational docking with the AI-based tool DiffDock-L, followed by physics-based HADDOCK simulations, mapped the compounds onto two chemically ligandable regulatory sites: the hydrophobic substrate-binding groove, which is the target of all known ClpC1-directed cyclic peptides in mycobacteria, and the pArg1 pocket, which normally recognizes phosphoarginine degrons. Mutational analysis supported this split: compounds 10 and 41 lost their activity against a mutant disrupting the pArg1 site, while several groove-targeting compounds showed reduced stimulation of a mutant at Met92, a residue lining the groove.

The team went further, obtaining co-crystal structures of the NTD bound to compound 8 at 1.90 angstrom resolution. The structure revealed that only part of the ligand, a pyrazoline-containing fragment the authors call Cpd-8a, makes direct protein contacts, mostly hydrophobic, with just a single hydrogen bond to Gly4. Structure-activity studies with commercial derivatives confirmed the importance of this fragment, and superposition with the cryo-EM structure of the resting state suggested a mechanism: ligand binding reorients the protein’s N-terminal tail, which would sterically clash with the AAA1 domain and destabilize the inactive conformation. For compound 10, docking and derivative testing converged on the pArg1 pocket as the binding site, with predicted clashes between the ligand and the tip of the regulatory middle domain — the same region whose disruption by mutation unleashes autonomous proteolysis. Notably, this is the first demonstration that the pArg1 pocket can be engaged by classical small molecules, extending the site’s known repertoire beyond phosphoarginine-bearing ligands and BacPROTAC degraders.

There is, however, an important caveat. When the compounds were tested against live S. aureus, all eight impaired growth to some degree, and compound 10 fully prevented growth at 200 micromolar with a 90-fold drop in viability and a minimum inhibitory concentration of 64 micrograms per milliliter. Yet experiments with clpC and clpP knockout strains showed that this toxicity does not depend on ClpC/ClpP at all — the knockouts were inhibited just as strongly, or more so. The authors suggest that compound 10’s coumarin-derived aesculetin moiety, related to aminocoumarin antibiotics that inhibit DNA gyrase, may drive off-target effects, and they caution that low affinity, poor permeability, or unknown secondary binding sites could all contribute to the disconnect between potent in vitro activation and ClpC-independent cellular toxicity.

Even so, the study marks a conceptual advance. It establishes that S. aureus ClpC, like its mycobacterial counterpart, is chemically addressable, and it maps two distinct ligandable sites within the NTD whose engagement produces a defined functional outcome: sustained, adaptor-independent proteolysis. Unlike natural adapters, which are consumed in the process, the small molecules are not degraded, enabling longer-lasting protease activity — a fundamental difference between biological regulation and chemical deregulation with clear implications for manipulating bacterial proteostasis. The researchers also note that mycobacteria deploy decoy proteins, ClpC2 and ClpC3, that sequester cyclic peptides and protect ClpC1; no such shield is expected for compounds targeting S. aureus ClpC. With structural blueprints for compound 8 and compound 10 in hand, the path forward is structure-guided optimization: raising affinity, confirming direct binding, and engineering the cellular engagement needed to convert biochemical deregulation into genuine, ClpC-dependent antibacterial lethality.

Subject of Research: Small-molecule activation of the Staphylococcus aureus ClpC/ClpP AAA+ protease as a potential antibiotic strategy

Article Title: Small-molecule activators of the Staphylococcus aureus ClpC/ClpP AAA+ protease

Article References: Jenne, T., Viliuga, V., Uhrig, U., Jehle, B., Schwan, M., Kopp, J., Flemming, D., Seebach, E., Sinning, I., Bukau, B., & Mogk, A. (2026). Small-molecule activators of the Staphylococcus aureus ClpC/ClpP AAA+ protease. iScience, 29(10), Article 117738. https://doi.org/10.1016/j.isci.2026.117738

Image Credits: AI Generated

DOI: 10.1016/j.isci.2026.117738

Keywords: Staphylococcus aureus, ClpC/ClpP protease, AAA+ ATPase, antibiotic resistance, high-throughput screening, allosteric regulation, protein homeostasis, crystallography, molecular docking, acyldepsipeptides, MRSA, antibacterial drug discovery