Menin-MLL inhibitors: docking a protein-protein interface

Menin-MLL inhibitors represent a clinically validated strategy for disrupting a protein-protein interaction central to acute leukemias driven by KMT2A rearrangements. Unlike traditional kinase or enzyme active-site inhibitors, these molecules target a shallow, hydrophobic pocket at the menin-MLL interface—posing unique challenges and opportunities for structure-based docking.

The menin-KMT2A interaction in leukemia

Menin is a scaffold protein encoded by MEN1 that recruits the histone methyltransferase KMT2A (also known as MLL1) to chromatin. In KMT2A-rearranged (KMT2Ar) acute leukemias—which result from chromosomal translocations involving the KMT2A gene—the resulting fusion proteins retain their menin-binding motif. This preserved interaction is essential for leukemic transformation: the menin–KMT2A fusion complex drives aberrant transcription of oncogenes like HOXA9 and MEIS1, sustaining leukemic stem-cell self-renewal.

Menin also plays a role in NPM1-mutant acute myeloid leukemia (AML), where the same transcriptional program is hijacked. Blocking the menin-MLL interface therefore offers a mechanism-based therapeutic approach that does not depend on drugging a catalytic site.

The menin-MLL binding pocket

The binding site for the KMT2A N-terminal fragment on menin is a shallow, predominantly hydrophobic cavity located between menin's central and C-terminal domains. Crystal structures reveal that a short MLL peptide binds as an extended strand with key hydrophobic residues—Phe9 and Pro10 of MLL—inserting into narrow pockets on menin. The interface spans roughly 1100 Ų and is stabilized by a combination of van der Waals contacts, aromatic stacking, and a handful of hydrogen bonds involving the peptide backbone.

Importantly, this is not a deep, well-defined cleft like a kinase ATP pocket. The menin pocket is relatively exposed and lacks the pronounced shape complementarity typical of enzyme active sites. This shallowness makes it challenging to achieve high-affinity binding with small molecules, and historically protein-protein interfaces were considered difficult targets for traditional drug discovery.

First-generation inhibitors and revumenib

Early menin-MLL inhibitors emerged from fragment-based screening and structure-guided optimization. Compounds such as MI-2 and MI-503 established proof of concept by demonstrating micromolar to nanomolar potency in biochemical and cellular assays, with X-ray structures confirming occupancy of the MLL-binding site.

Revumenib (SNDX-5613) is a clinical-stage menin inhibitor currently in trials for relapsed or refractory KMT2Ar and NPM1-mutant AML. It binds competitively in the MLL pocket with low-nanomolar affinity, inducing differentiation of leukemic blasts and downregulating MEIS1 and HOXA cluster genes. Clinical responses have been observed in heavily pretreated patients, validating menin as a druggable target despite the absence of a classic catalytic or allosteric site.

Docking challenges at protein-protein interfaces

Docking a ligand into a protein-protein interaction site differs in several respects from docking into an enzyme active site:

• Shallow, solvent-exposed binding sites: PPI interfaces often lack the enclosed geometry of active sites, reducing shape selectivity and increasing the entropic cost of ligand binding. Scoring functions tuned for deep pockets may underestimate binding affinity.
• Large, flat contact areas: The menin-MLL interface is broad relative to the size of small-molecule inhibitors. Ligands typically occupy a "hot spot" sub-region where a few key residues contribute disproportionately to binding energy. Identifying this hot spot is essential for docking box placement.
• Induced fit and backbone flexibility: PPI pockets may exhibit greater conformational plasticity than pre-organized catalytic clefts. The menin binding site undergoes modest rearrangement upon ligand binding; apo and ligand-bound structures can differ in side-chain and loop positioning.
• Polar and hydrophobic balance: While the menin pocket is hydrophobic, backbone hydrogen bonds to the MLL peptide are mimicked by some inhibitors through polar groups. Docking must capture both lipophilic contacts and discrete H-bond networks.

These factors mean that rigid-receptor docking may miss critical poses, and ensemble or flexible-residue docking becomes more valuable.

Setting up menin-MLL docking in Liganx

To dock a candidate inhibitor or perform virtual screening against the menin pocket, start with a high-resolution crystal structure of menin in complex with a known inhibitor. Structures co-crystallized with revumenib or earlier tool compounds provide the best representation of the ligand-binding conformation.

Define the docking box to encompass the hydrophobic sub-pockets occupied by MLL Phe9 and Pro10, as well as adjacent regions contacted by current inhibitors. A box centered on the bound ligand with dimensions of approximately 20–24 Å per side typically suffices. Because the pocket is shallow, expanding the box too far risks including irrelevant surface regions and diluting scoring discrimination.

Consider the protonation states of nearby histidine and lysine residues at physiological pH, and ensure that metal ions or structured waters observed in multiple complexes are retained if they mediate key interactions.

Liganx Studio allows you to upload PDB structures, set flexible residues at the binding interface, and run AutoDock Vina or Gnina docking with customized exhaustiveness. For PPI targets, increasing exhaustiveness and enabling a degree of receptor flexibility—particularly for side chains lining the pocket—can improve pose quality and score correlation.

Interpreting docking results

When evaluating docked poses of menin inhibitors, prioritize the following:

• Occupancy of the MLL hot spot: The ligand should insert hydrophobic groups into the pockets that normally accommodate MLL Phe9 and Pro10. Poses that float on the surface without engaging these sub-sites are unlikely to be productive.
• Hydrogen bonding to menin backbone or conserved side chains: Many menin inhibitors form H-bonds to the carbonyl or amide of menin residues that hydrogen-bond to the MLL peptide. Check for recapitulation of these interactions.
• Comparison to known binders: Overlay the docked pose with a co-crystal ligand. Significant deviations in core scaffold placement or key pharmacophore positioning warrant scrutiny.
• Score trends, not absolute values: Docking scores for PPI inhibitors may be less negative than for active-site binders of similar affinity due to the shallower pocket and reduced burial. Relative ranking within a congeneric series is often more informative than raw kcal/mol values.

Molecular dynamics simulations or free-energy perturbation calculations can refine predictions, especially when optimizing analogs within a scaffold series.

Opportunities for mutation-aware docking

While resistance mutations are not yet a widespread clinical challenge for menin inhibitors, structural surveillance of the binding pocket is prudent. Mutations in menin residues that contact inhibitors directly—or that alter pocket shape or dynamics—could emerge under selective pressure. In other PPI systems, interface mutations have been shown to confer resistance by reducing inhibitor affinity without abrogating the native protein-protein interaction.

Using Liganx mutation-aware docking, you can model point mutations in menin pocket residues and predict their impact on inhibitor binding. This workflow is particularly relevant for rational design of second-generation compounds with resilience to potential escape variants.

Outlook

Menin-MLL inhibitors exemplify how structure-based drug design can succeed at protein-protein interfaces once considered intractable. The clinical promise of revumenib and related molecules validates the menin pocket as a druggable site and provides a roadmap for targeting other PPI-driven oncogenes.

For computational chemists, docking at PPI interfaces demands careful attention to receptor preparation, box definition, and pose validation against experimental structures. The shallow, extended geometry of the menin-MLL binding site means that subtle differences in pose or scoring can distinguish active from inactive compounds—and flexible-receptor or ensemble docking can be critical for success.

References

• Grembecka J, He S, Shi A, et al. Menin-MLL inhibitors reverse oncogenic activity of MLL fusion proteins in leukemia. Nat Chem Biol. 2012;8(3):277–284. doi:10.1038/nchembio.773
• Borkin D, He S, Miao H, et al. Pharmacologic inhibition of the Menin-MLL interaction blocks progression of MLL leukemia in vivo. Cancer Cell. 2015;27(4):589–602. doi:10.1016/j.ccell.2015.02.016
• Kühn MWM, Song E, Feng Z, et al. Targeting chromatin regulators inhibits leukemogenic gene expression in NPM1 mutant leukemia. Cancer Discov. 2016;6(10):1166–1181. doi:10.1158/2159-8290.CD-16-0237
• Klossowski S, Miao H, Kempinska K, et al. Menin inhibitor MI-3454 induces remission in MLL1-rearranged and NPM1-mutated models of leukemia. J Clin Invest. 2020;130(2):981–997. doi:10.1172/JCI129126
• Issa GC, Ravandi F, DiNardo CD, et al. The menin inhibitor revumenib in KMT2A-rearranged or NPM1-mutant leukaemia. Nature. 2023;615(7954):920–924. doi:10.1038/s41586-023-05812-3