Development of a Novel Class of Agents Targeting the RNA-Splicing Machinery in Myeloid Malignancies

SF3B1 is a splicing factor gene whose mutations are pathognomonic of MDS with ring sideroblasts. Because of the ubiquitous importance of splicing, a major barrier in targeting cells with spliceosomal mutations is the discovery of agents decreasing the competitiveness of mutant cells while preserving the integrity of wild type cells. To date no specific therapies are FDA approved for SF3B1 mutant (SF3B1^MT) MDS and few agents are in early clinical testing. We describe a novel targeted approach to drug development for SF3B1^MT malignancies.

Our investigative strategy started with a high throughput drug screen. We introduced K700E mutation into myeloid cells using CRISPR/Cas9. We then subjected K562^+/K700E and matched-parental K562 cells to high throughput drug screen of a library of 3,000 mechanistically annotated, non-redundant, bioactive compounds. Top hits were validated by dose response testing (8 concentrations in half-log dilutions). Our interest focused on compounds with cytostatic activity towards K562^+/K700E cells. Among these, a 4-pyridyl-2-anilinothiazole (PAT) showed preferential inhibition of growth in K562^+/K700E cells with an IC50 of 3uM. Dose response showed that K562^+/K700E cells were significantly sensitive to PAT with a growth inhibition of 20%, 32%, 51%, 65%, 95% at .3uM (P=.01), 1uM (P=.002), 3uM (P<.0001), 10uM (P<.0001), and 20uM (P=.01). High doses (10, 20uM) were toxic to parental cells. PAT treatment did not induce growth arrest in other myeloid cells (THP1, MOLM13^FLT3, OCI-AML3^DNMT3A, SIG-M5^TET2/DNMT3A, K562^PHF6) including cells with mutations in other splicing factors (K562^U2AF1, K562^LUC7L2). PAT induced similar effects in primary SF3B1^MT MDS cells at 3uM while it did not induce significant growth inhibition in primary MDS cells with other mutations, including other spliceosomal mutations (e.g.,U2AF1) (N=6). In normal bone marrow cells (N=6), complete growth arrest of erythroid and myeloid cells was observed at high doses (20uM).

Using PAT as our lead, we employed a fragment based reiterative medicinal chemistry approach to synthesize selective compounds and improve therapeutic index. Libraries were constructed following Lipinski rules with ease of synthetic construction in mind. Pilot libraries were constructed via the classic Hantsch thiazole synthesis which involves condensation of α-halo ketones with substituted thioureas. This enterprise investigated SAR modifications of PAT by considering features such as regiochemistry and basicity of the nitrogen of the pyridine ring in the head region; replacement of the spacer 2,4-disubstituted thiazole ring with heteroatom containing rings (5,6,7 membered aromatic or aliphatic ring structures); alternatives for the aniline of the tail region e.g., sulfonamide, amide, and substituted amine linkers and substituted aromatic and aliphatic ring structures for the phenyl substituent. A major challenge in drug development of agents targeting spliceosomal mutations is the identification of key clusters of aberrantly spliced genes restored by drug treatment, e.g., identification of a pharmacodynamic endpoint that could be used to prove that that the drug reached its target. SF3B1 mutations induce aberrant 3′-ss selection by promoting use of alternative branch points. To identify biomarkers of splicing changes to screen our libraries, K562^+/K700E and parental cells were treated with PAT (3uM) and RNA libraries were subjected to RNA Seq. The splicing pattern of parental cells was used as reference. We identified 328 cassette exons (±25% splicing inclusion difference, pFDR<.05) in K562^+/K700E cells of whom the splicing of 22 exons was partially or completely restored upon treatment. Among these, PAT treatment restored the splicing of exons in sensitive 3′-ss sequences (ENOSF1), RNA binding SR-related factors (ACIN1), zing fingers (ZC3H7A) already associated with SF3B1 downregulation, histone modifiers (HMGN3), mitochondrial genes (TMEM126B), proto-oncogenes (CREB1) and heat shock proteins (DNAJC24). Ongoing experiments include tests of the ability of PATs to restore the splicing of misspliced exons and its efficacy in reducing the percentage of SF3B1^MT cells in xenografts of K562^+/K700E and primary SF3B1^MT cells and Sf3b1^+/K700E mice.

In sum, we described novel classes of compounds that inhibit the expansion of SF3B1^MT cells by restoring the splicing defects intrinsically associated with SF3B1.