The work, led by Jianzhao Liu and Xin-Hua Feng of Zhejiang University together with colleagues at Soochow University and other Chinese institutions, began with a simple but technically demanding question: is METTL14, the scaffold subunit of the m6A methyltransferase complex, itself chemically modified? Using immunoprecipitation followed by immunoblotting in HEK293T and HeLa cells, the team found that METTL14, but not its catalytic partner METTL3 nor the regulatory subunit WTAP, is robustly ubiquitinated. Ubiquitin is best known as a destruction tag: chains linked through its lysine-48 residues deliver proteins to the proteasome. But ubiquitin can be assembled into eight distinct chain types, and several of them, including K27-linked chains, are thought to modulate protein interactions and localization rather than stability. When the researchers restricted cellular ubiquitin to single-linkage variants, METTL14’s modification turned out to be dominated by exactly this noncanonical K27 linkage.

Mass spectrometry then mapped the modification to specific lysine residues on METTL14, including lysines 148 and 278, positions that sit near the interfaces through which METTL14 grips its partners. To find the enzyme responsible, the team pulled down METTL14 and searched its associated proteins for candidate E3 ubiquitin ligases. The screen converged on RLIM, also known as RNF12, a RING-domain ligase first characterized as a corepressor of LIM homeodomain transcription factors. RLIM’s catalytic activity proved essential: a ligase-dead H569A/C572A mutant or a construct lacking the RING domain entirely failed to promote METTL14 ubiquitination, whereas wild-type RLIM drove robust K27-linked tagging in cells and in reconstituted in vitro reactions with defined E1 and E2 enzymes.

What makes this finding conceptually striking is what the ubiquitin mark does not do. RLIM knockout did not alter METTL14 protein abundance, and immunofluorescence showed that METTL14 still entered the nucleus normally in cells lacking RLIM. Instead, the modification acted on function. In vitro methylation assays demonstrated that K27-ubiquitinated METTL3–METTL14 complexes were markedly impaired in their ability to methylate RNA substrates. Two mechanisms accounted for the deficit. First, structural modeling and biochemical assays indicated that ubiquitin conjugated at lysine 278 distorts the loop regions at the METTL3–METTL14 interface, weakening the heterodimer that constitutes the active writer. Second, electrophoretic mobility shift assays and microscale thermophoresis showed that ubiquitinated or ubiquitination-mimicking complexes bound RNA less tightly, reducing substrate engagement. A split Nano-Luciferase reporter confirmed in living cells that RLIM expression disrupts METTL3–METTL14 association. In other words, RLIM acts as a rheostat on the writer’s activity rather than a trigger for its disposal.

The physiological consequence was visible across the transcriptome. MeRIP-seq and RNA-seq comparisons of control and RLIM-knockout cells revealed widespread hypermethylation of mRNA m6A peaks when RLIM was absent, with the affected transcripts enriched for genes governing the G2/M transition of the cell cycle. Consistent with known links between m6A and mRNA stability, hypermethylated transcripts in RLIM-knockout cells showed shortened half-lives, and many overlapped with binding targets of the m6A reader YTHDF2, which routes methylated transcripts toward degradation. Representative loci such as E4F1 and RHOB, both implicated in cell-cycle control, gained m6A and lost stability when RLIM was removed, and MeRIP-qPCR validated these changes independently.

The cell-cycle connection then became explicit. When the researchers synchronized cells by double thymidine block and released them, they observed coordinated periodic oscillations in three quantities: transcriptome-wide m6A levels, the intensity of K27-linked ubiquitin on METTL14, and RLIM protein abundance. RLIM knockout arrested cells at the G2/M boundary, an effect that could be rescued by knocking down METTL14, placing RLIM upstream of the writer in a pathway that governs mitotic entry. Measurements of global m6A-to-A ratios by LC-MS/MS across G1/S, S, G2, M and G1 phases confirmed that the methylation landscape rises and falls with the cycle, and immunofluorescence for phospho-histone H3 showed RLIM levels dropping precisely in mitotic cells.

The mechanism behind RLIM’s own oscillation turned out to be a classic mitotic kinase. The CDK1–cyclin B1 complex, the master switch for mitotic entry, physically associated with RLIM and phosphorylated it. Phosphatase treatments and Ser/Thr-to-Ala mutant analysis identified the phosphorylation sites responsible, and these modifications triggered RLIM to ubiquitinate itself with K48-linked chains, condemning it to proteasomal degradation. Blocking CDK1 or cyclin B1 by RNA interference stabilized RLIM, while proteasome inhibition with MG132, but not lysosomal inhibition with chloroquine, rescued the protein. This arrangement creates an elegant feedback loop: as CDK1–cyclin B1 surges at G2/M, it destroys RLIM, which releases METTL14 from inhibitory ubiquitination, allowing the writer complex to reassemble and retune the m6A landscape for the next phase of the cycle.

To test whether this axis matters beyond cultured cells, the team turned to animal and clinical models. In mice with Rlim deleted specifically in the myeloid lineage, bone marrow cells showed elevated m6A methylation on cell-cycle and differentiation genes, and flow cytometry revealed perturbed hematopoietic populations, effects that intensified after vesicular stomatitis virus infection. Endogenous METTL14 in these marrow cells lost its K27-linked ubiquitination, confirming the pathway operates in primary tissue. The researchers also examined MOLM-13, an acute myeloid leukemia cell line, where RLIM knockdown altered cell-cycle distribution and myeloid differentiation markers and reduced colony formation. Most compellingly, samples from patients with acute monocytic leukemia, collected through collaborating hospitals in Suzhou, displayed dysregulation of the RLIM–METTL14 axis, linking the molecular circuit to human disease.

The study’s broader significance lies in expanding the vocabulary of epitranscriptomic regulation. For a decade, the m6A writer has been viewed largely as a stable assembly whose output is tuned by accessory proteins, substrate availability and chromatin state. This work adds a temporal dimension: a cell-cycle-programmed, noncanonical ubiquitin code that switches the writer’s activity on and off without touching its concentration. Because K27-linked chains are increasingly recognized as scaffolds for protein–protein interactions in cell-cycle and quality-control pathways, the finding suggests that other RNA-processing enzymes may carry similar non-degradative marks awaiting discovery. It also reframes ubiquitin ligases, traditionally studied in the context of protein turnover, as direct modulators of the epitranscriptome.

Therapeutically, the implications are tantalizing but preliminary. Acute myeloid leukemia has already emerged as a disease exquisitely sensitive to m6A dosage, with the writer METTL3, the erasers FTO and ALKBH5, and the reader YTHDF2 each implicated in leukemic self-renewal. The new axis adds RLIM, and the CDK1–cyclin B1 switch that controls it, to the list of potential intervention points. CDK1 inhibitors are already in clinical use for other indications, raising the possibility that modulating RLIM stability could tune m6A levels in leukemia cells. For now, the study stands as a demonstration that the cell’s division clock reaches deep into its RNA chemistry, and that a ubiquitin tag once dismissed as obscure may be one of the hands on that clock.

Subject of Research: Cell-cycle-regulated noncanonical K27-linked ubiquitination of METTL14 controlling mRNA m6A methylation

Article Title: Cell-cycle-driven noncanonical ubiquitination of METTL14 orchestrates RNA methylation

Article References: Cao, J., Ying, X., Zhou, L., Zhang, Q., Huang, C., Dai, T., Shu, X., Gao, M., Mi, Y., Wu, X., Wang, F., Zhang, L., Zhou, F., Feng, X.-H., & Liu, J. (2026). Cell-cycle-driven noncanonical ubiquitination of METTL14 orchestrates RNA methylation. Nature Chemical Biology. https://doi.org/10.1038/s41589-026-02328-5

Image Credits: AI Generated

DOI: 10.1038/s41589-026-02328-5

Keywords: m6A, METTL14, RLIM, ubiquitination, K27-linked ubiquitin, cell cycle, CDK1, cyclin B1, epitranscriptomics, RNA methylation, acute myeloid leukemia, post-translational modification

The biological rationale for targeting CDK1 is compelling. CDK1 is the master regulator of the G2-to-M transition in the cell cycle, the moment when a cell commits to dividing. Paired with cyclin B, it forms the M-phase promoting factor that phosphorylates the proteins responsible for chromosome condensation and spindle assembly. Unlike CDK2, CDK4, and CDK6, which show functional redundancy, CDK1 has no substitute at this checkpoint. Many tumors, particularly those carrying p53 mutations that cripple the G1 checkpoint, become dangerously dependent on CDK1 to enter mitosis even when DNA is damaged. Blocking CDK1 can therefore trigger mitotic catastrophe and apoptosis selectively in cancer cells, while normal cells with intact checkpoints pause and repair instead.

Despite this therapeutic logic, no CDK1 inhibitor has yet been approved by the FDA. Early pan-CDK inhibitors such as flavopiridel, now known as alvocidib, showed activity against leukemia and multiple myeloma but failed on selectivity, producing toxicity, diarrhea, nausea, and hepatotoxicity. Roniciclib trials in small-cell lung cancer were halted after severe adverse effects. Dinaciclib, which targets CDK1, CDK2, CDK5, and CDK9, remains the most clinically advanced agent, but the field continues to search for molecules with cleaner safety profiles. Natural products, which often carry lower intrinsic toxicity and can engage multiple pathways simultaneously, have emerged as an attractive hunting ground.

The research team turned to the NPACT database, a curated collection of plant-derived anticancer compounds, and focused on benzofuranoid phytochemicals from Morus alba. Rather than screening millions of synthetic molecules, they performed a targeted virtual screen of compounds with documented anticancer relevance. The CDK1 crystal structure, resolved at 2.33 angstroms and retrieved from the Protein Data Bank under the identifier 6GU6, was prepared in UCSF Chimera, and ligands were energy-minimized with the MMFF94 force field before docking with AutoDock Vina through the PyRx interface.

The docking results were striking. Mulberrofuran L bound CDK1 with a score of minus 8.5 kilocalories per mole, followed by mulberrofuran X at minus 8.2, mulberrofuran Y at minus 8.0, and euparin and 6-methoxy-tremetone at minus 7.7 each. Dinaciclib, the reference drug, scored only minus 6.6 kilocalories per mole. The docking protocol was validated by redocking the co-crystallized dinaciclib into the binding pocket, yielding a root-mean-square deviation of 1.38 angstroms, well below the 2-angstrom threshold that confirms a docking method reproduces known binding poses reliably.

Drug-likeness and pharmacokinetic filtering narrowed the field further. All five top compounds largely satisfied Lipinski’s Rule of Five, with the single exception of elevated lipophilicity: mulberrofuran L, X, and Y each showed logP values above 5, a violation the authors note may compromise aqueous solubility and oral bioavailability but does not automatically disqualify a compound. Predicted potency, estimated with the CODD-PRED server using machine learning models trained on curated bioactivity data, ranged from 5.28 to 5.48 in pIC50 terms, with mulberrofuran L scoring highest. The decisive filter was toxicity. ADMET profiling with the Deep-PK server flagged mulberrofuran Y, euparin, and 6-methoxy-tremetone for carcinogenicity and hepatotoxicity concerns, leaving mulberrofuran L and mulberrofuran X as the only candidates with clean toxicity profiles.

Those two survivors, along with dinaciclib as a benchmark, were then subjected to 200-nanosecond all-atom molecular dynamics simulations in GROMACS using the CHARMM36 force field. The mulberrofuran X complex proved the most stable of the three, showing the lowest and most consistent root-mean-square deviation at roughly 0.25 to 0.32 nanometers, the tightest radius of gyration at 2.32 to 2.36 nanometers, and the lowest solvent-accessible surface area, all signs of a compact, well-behaved protein-ligand complex. Mulberrofuran L fluctuated more, with RMSD values between 0.35 and 0.55 nanometers, though its excursions settled over time. Principal component analysis reinforced the picture: mulberrofuran X formed a tight conformational cluster, while mulberrofuran L sampled a broader range of motion along the first two principal components, and the free energy landscape confirmed that both compounds occupied well-defined low-energy minima rather than drifting into unstable states.

The energetic verdict was perhaps the study’s most eye-catching result. MM-GBSA binding free energy calculations, performed on frames extracted from the equilibrated trajectories, gave minus 15.66 kilocalories per mole for mulberrofuran X and minus 11.82 for mulberrofuran L, compared with just minus 2.31 for dinaciclib. Since more negative values indicate stronger binding, both mulberry compounds appeared to outperform the clinical reference by a wide margin in this metric. Density functional theory calculations added an electronic dimension: dinaciclib showed the smallest HOMO-LUMO gap at 2.894 electron volts, indicating the highest chemical softness and reactivity, while mulberrofuran L, at 2.943 electron volts, sat close behind, and mulberrofuran X, at 3.832 electron volts, was the most electronically stable of the three.

The authors are careful to frame these findings as hypothesis-generating rather than proof of therapeutic value. Every result in the study is computational, and the pipeline’s predictions, from pIC50 estimates to CYP450 inhibition flags for the mulberrofuran compounds, all require experimental confirmation. The team recommends in vitro kinase inhibition assays and cell viability tests such as MTT or CCK-8 to verify that mulberrofuran L and X actually suppress CDK1 in living cells, followed by cytotoxicity, selectivity, and mechanistic studies in cancer cell lines and eventually animal models. Predicted CYP inhibition in particular raises potential drug-drug interaction risks that only dedicated laboratory pharmacokinetics can resolve.

Even with those caveats, the study adds mulberry phytochemicals to a growing list of plant-derived CDK modulators, joining curcumin, wogonin, honokiol, and atractylenolide I, all of which have shown cell cycle arrest effects in preclinical cancer models. What distinguishes this work is the depth of the computational validation: a multi-level workflow spanning docking, dynamics, free energy, and quantum chemistry that converged on the same two candidates from different angles. If laboratory experiments bear out the simulations, the humble mulberry, a tree better known for feeding silkworms than fighting tumors, could yield lead structures for a class of cancer drugs that has long eluded medicinal chemists: selective, tolerable inhibitors of CDK1.

Subject of Research: Computational screening of Morus alba phytochemicals as CDK1 inhibitors for cancer therapy

Article Title: In silico investigation of mulberry phytochemicals targeting CDK1 in cancer using molecular docking and computational approaches

Article References: Abdullah, S., Nisar, S., Irshad, M., Saini, R. S., Alam, M., Ansari, I., & Junaid, M. (2026). In silico investigation of mulberry phytochemicals targeting CDK1 in cancer using molecular docking and computational approaches. Discover Chemistry, 3(1), Article 562. https://doi.org/10.1007/s44371-026-01020-w

Image Credits: AI Generated

DOI: 10.1007/s44371-026-01020-w

Keywords: CDK1, Morus alba, molecular docking, molecular dynamics simulation, phytochemicals, cancer, drug discovery, MM-GBSA, ADMET, DFT, mulberrofuran, cell cycle