Discerning the Landscape of Menin Inhibitor Resistance

Background: The Phase 1 AUGMENT-101 trial demonstrated that the menin inhibitor (MI) revumenib has promising activity in KMT2A-rearranged (KMT2A-r) and nucleophosmin mutant (NPM1m) acute leukemias, however 40% of patients with available DNA samples developed menin (MEN1) mutations and some acquired resistance without MEN1 mutations (Issa et al., Nature, 2023; Perner et al., Nature, 2023). As more MIs enter clinical trials, open questions remain as to what extent MEN1 mutation-driven and MEN1 WT resistance will arise. Results: In a KMT2A-r PDX model, we demonstrate that continuous MI treatment generated MEN1 mutations in 4 out of 5 mice treated with lower dose MI therapy. With higher dose MI therapy, only 1 of 5 mice developed a MEN1 mutation and in the remaining mice MEN1 WT cells persisted and slowly expanded over the course of ≥6 months of therapy despite on-target gene expression changes: decreased MEIS1 and HOXA cluster genes, increased CD11b, CD13, CD14. Given that CRISPR-Cas9 base editing previously predicted some of the MEN1 mutations that arose with revumenib, we utilized improved base editing technology to help discern the landscape of MEN1 acquired mutations for four additional MIs currently in clinical trials (DS-1594, DSP-5336, JNJ-75276617, ziftomenib). MV4;11 (KMT2A-r AML) cells were transduced with spCas9 base editing variants that mediate either C->T or A->G DNA substitutions (Sangree et al., Nat. Comm., 2022) and a specific pooled guide RNA (sgRNA) library designed to cover all MEN1 exons. We conducted a 21-day positive selection screen, treating with the four MIs. After selection, 41 individual sgRNAs enriched ≥ +3.5 Log2FC relative to DMSO control. We validated these results by transducing individual sgRNAs in SpCas9 C->T and A->G MV4;11 cells and carried out a 9-day competition assay under single dose MI treatments, starting with a population of 20% edited cells mixed with 80% unedited cells. 33 sgRNAs enriched ≥3x with ≥2 MIs or ≥4.25x with ≥1 MI. Cell lines with each of these 33 sgRNAs were next subjected to viability assays and absolute IC50s were calculated. 11 different sgRNAs targeting 8 different amino acid residues shifted the absolute IC50 of at least one MI ≥10x or ≥7.5x relative to another MI. Sanger sequencing revealed these sgRNAs produced amino acid substitutions at residues previously found in patients: M327, G331, and T349, as well as previously unknown residues such as A247, D290, C334, E368, and V372. Some of these mutations confer substantial, differential sensitivity amongst MIs. For example, in MV4;11 cells a specific mutation could be found that shifts the absolute IC50 >75x for DSP-5336 (>1000 nM), 2.5x for DS-1594 (25.8 nM), 1.6x for ziftomenib (22.2 nM), and 0.3x for JNJ-75276617 (3.3 nM). In contrast, a MEN1_M327 mutation shifted the absolute IC50 ≥8x for all four MIs. Comparison of newly identified mutations with crystal structures of each MI bound to menin suggested a mechanism of action for most mutations. Orthogonal biological validation was undertaken in MOLM13 (KMT2A-r AML) cells to screen for spontaneous MEN1 mutation generation. We treated an initial culture of 15 million cells in triplicate with 25 nM of each MI. After ≤28 days of treatment, we found outgrowth of a MEN1_M327V mutation following treatment with JNJ-75276617 and a MEN1_E368V mutation with DSP-5336, with each emerging from one replicate. Remarkably, these mutations occurred at the amino acid residues that conferred the most resistance for each compound following validation of the base editor screen. Conclusions: PDX models demonstrated differences in resistance mechanisms dependent on MI dose. A MEN1 base-editor screen utilizing optimized SpCas9 variants found both shared and novel acquired mutations that affect four additional MIs currently in clinical trials: DS-1594, DSP-5336, JNJ-75276617, and ziftomenib. Some MEN1 mutations have pan-class effect (M327 mutations) and others are selective for specific MIs. An in vitro technique leveraging large starting cell numbers validates this base editor screening approach, which resulted in rapid development of resistant MEN1 mutant clones in less than 4 weeks of treatment. Identifying MEN1 mutations in patients receiving MIs is important because there may be opportunities for patients with select MEN1 mutations to derive benefit from an alternate MI-based therapy, provided that the precise mutational profile of each inhibitor is known.