Systematic Evaluation of GAPs and Gefs Identifies ARHGAP45 As a Targetable Leukemia-Specific Dependency

GTPase-activating proteins (GAPs) and guanine nucleotide exchange factors (GEFs) function as negative and positive regulators of small GTPases, respectively, and play key roles in cell fate determination. The aberrant function of GAPs and GEFs has long been implicated in cancer development but their vast number and potential redundancy have hindered comprehensive understanding of their roles. Here we performed functional genomic screens of GAPs and GEFs in primary AML specimens to uncover an unexpected and selective role for ARHGAP45 in AML. Depletion of ARHGAP45 impedes AML growth without affecting normal human CD34⁺ cells in vivo. Epistasis screens revealed RhoA as the major substrate of ARHGAP45 and that depletion of the Rho GTPase CDC42 synergized with ARHGAP45 deletion. Consistent with this, pharmacologic CDC42 inhibition enhanced the effect of ARHGAP45 knockout (KO). ARHGAP45 gives rise to a minor histocompatibility antigen (termed HA-1), which can be recognized by antigen-specific T cells. CDC42 inhibition not only sensitized AML cells to ARHGAP45 depletion but also promoted the cytolytic activity of HA-1 antigen-specific T cells by upregulating ARHGAP45-derived epitope in AML cells.

We generated a domain-focused CRISPR library of 1,320 sgRNAs targeting 187 GAPs, 103 GEFs, as well as negative and pan-essential controls, and conducted CRISPR KO screens in PDX models in vitro and in vivo from 2 AML patients and 1 AML cell line (MOLM-13). ARHGAP45 emerged as the top-ranking gene in vitro and vivo. Competition assays across cancer cell lines, including AML, B-ALL, myeloma, and solid tumors, validated ARHGAP45 as a lineage-specific dependency in AML. CRISPR-scanning of ARHGAP45’s domains revealed the RhoGAP domain as essential in AML. Consistent with these findings, ARHGAP45 depletion in 3 AML PDX models and 1 cell line extended survival of recipient NSGS mice compared to controls. In contrast, deletion of ARHGAP45 in normal human cord blood CD34⁺ cells had no impact on engraftment or differentiation in vivo. CITE-seq across 5 CRISPR-edited primary AML samples revealed that ARHGAP45 expression is associated with leukemia stemness as ARHGAP45 depletion decreased immature leukemia progenitor subpopulations while increasing differentiated cDC and erythroid subpopulations.

GAPs function as the signaling terminator of the GTPase-modulated pathways; however, specific pathway(s) that ARHGAP45 suppresses is unknown. We therefore performed CRISPR epistasis screens to identify effectors whose depletion rescues the growth defect of ARHGAP45 KO. Deletion of one particular GTPase RhoA and ARHGEF1, a RhoA GAP, conferred resistance to ARHGAP45 depletion. Consistent with this, ARHGAP45 KO elevated active GTP-bound RhoA, and expression of constitutively active RhoA Q63L impaired AML cell growth. The epistasis screens nominated CDC42 as another gene whose KO synergized with ARHGAP45 depletion to promote leukemia cell death. Pharmacological inhibition of CDC42 with the CDC42/Rac1 inhibitor MBQ-167 further enhanced survival following ARHGAP45 KO in AML models.

ARHGAP45 harbors a SNP encoding a peptide (VLHDDLLEA, “HA1-H”) stably presented on HLA-A*02:01, whereas the allelic counterpart VLRDDLLEA (“HA1-R”) has insufficient HLA class I binding. Engineered T cell receptor (TCR) cells against HA-1H are under clinical investigation in AML. Here we find that a novel CAR-like TCR targeting HA-1H, consisting of a single-chain variable TCR derived from the variable regions of TCRα and β chain and signaling through a CD28-based co-stimulatory domain is as effective as an existing full-length TCR T cell approach. Specificity of the HA-1H targeting CAR was confirmed by dual color peptide/MHC dextramers against HA-1H peptide. Furthermore, MBQ-167 augmented protein expression of ARHGAP45, enhanced HA-1H antigen presentation, and increased the cytolytic activity of HA-1H-targeting TCR-T cells.

These findings provide a resource for probing oncogenic and potentially druggable regulators of GTPase activities and highlight rational therapeutic strategies for modulating GTPase activity in AML. We find that ARHGAP45 is a bona fide GAP for RhoA that is selectively required in AML cells but dispensable in normal hematopoietic cells. Furthermore, we propose a novel immunotherapeutic strategy to target AML by augmenting new and existing TCR-T therapies against ARHGAP45-derived antigens.