RNA Polymerase II Is a Therapeutic Target to Overcome FLT3 Inhibitor Resistance Mediated By the Bone Marrow Microenvironment

Background: Gilteritinib is a clinically active FLT3 tyrosine kinase inhibitor (TKI) approved for relapsed/refractory FLT3-mutant AML, but nearly all patients treated with gilteritinib and other FLT3 TKIs eventually develop clinical resistance. Activating RAS/MAPK pathway mutations are a predominant non-FLT3 dependent resistance mechanism in patients treated with gilteritinib. AML blasts can also develop FLT3 TKI resistance secondary to paracrine MAPK activation stimulated by FLT3 Ligand, FGF2, or other protective cytokines within the bone marrow microenvironment (BME). To identify potential targets that sensitize AML cells to gilteritinib-induced apoptosis in a model of the BME, we performed a genome-wide CRISPR/Cas9 death screen in MOLM-14 FLT3-ITD+ human AML cells cultured in bone marrow stromal cell conditioned media. We hypothesize that identified genes represent promising combinatorial therapeutic targets that can enhance clinical efficacy of FLT3 TKIs in AML.

Methods: To model stroma-mediated TKI resistance, we used the HS5 human bone marrow stromal cell line that secretes multiple cytokines (G-CSF, GM-CSF, FGF2) and supports myeloid progenitor proliferation in co-culture. MOLM-14 CRISPRi cells transduced with CRISPRi-v2 genome-wide sgRNA library were cultured in HS5 conditioned media for 24 hours and then treated with gilteritinib 250 nM. Cells were stained with a fluorogenic caspase 3/7 reagent and then fixed after 24 hours of drug treatment. Caspase-3 positive cells were sorted from the entire drug-treated cell population by FACS and guide RNAs enriched or depleted in this sample as compared to an untreated T0 sample were determined by NGS.

Results: We identified several gene-level hits that were enriched in the apoptotic population (FDR <0.2). Among these, we identified multiple transcription factors or regulators of transcriptional activation. The latter included multiple components of RNA pol II machinery (POLR2G, RTF1) and multiple subunits of Mediator (MED12, MED30, MED21, MED11), a complex that regulates RNA pol II activity and has been shown to modulate super-enhancer-associated genes in AML cells. To validate select hits, we transduced MOLM-14 and MV411 CRISPRi cells with a tetracycline-inducible sgRNA expression vector. Using this system, we found that conditional knockdown of MED12, the top scoring Mediator subunit in our screen, significantly sensitized MOLM-14 (FgH1) cells to gilteritinib while modestly augmenting the cytotoxicity of gilteritinib in MV411 (FgH1) cells when compared to a non-targeting sgRNA. Functionally, MED12 associates with MED13, Cyclin C, and CDK8 to form the CDK8 kinase module of Mediator. MED12 knockdown, as expected, led to suppression of STAT1 S7272 and STAT5 S726 phosphorylation, known targets of CDK8. Based on these results, we hypothesized that CDK8 inhibition would augment apoptosis induced by gilteritinib in HS5 conditioned media. Using SEL120, a novel CDK8 inhibitor already in early phase AML clinical trials, we performed an 8 x 8 dose matrix drug synergy analysis of gilteritinib and SEL120 in multiple FLT3-mutant AML cell lines using Bliss independence modeling. We found that SEL120 was synergistic with gilteritinib in inducing apoptosis in MOLM-14 and MV411 cells in HS5 conditioned media (Bliss synergy scores of 2.44 and 11.45 with most synergistic area scores of 10.91 and 26.85, respectively). We also found combinatorial activity against MOLM-14 cells harboring secondary NRAS activating mutations (G12C and Q61K), suggesting the therapeutic combination could potentially overcome cell intrinsic and extrinsic MAPK-activating resistance mechanisms. Lastly, we found that gilteritinib and SEL120 combined to impart greater cytotoxicity than either drug alone in a primary sample (AML #1) from a patient with newly diagnosed AML possessing a FLT3-ITD mutation at high mutant allele ratio.

Conclusions: The results and validation of our CRISPRi screen suggest that combined CDK8 and FLT3 inhibition is a novel strategy for augmenting gilteritinib cytotoxicity. Assessment of the activity of the combination in additional primary AML samples and in vivo murine models of AML is planned. Additional candidate targets already described and other Mediator and RNA pol II subunits from our screen are also being further evaluated to precisely define the transcriptional programs that influence FLT3 inhibitor resistance.