ASPP2 k, a Dominant-Negative Splicing Variant of the Apoptosis-Stimulating Protein of p53-2 (ASPP2), Modulates Treatment Response Towards BCL-Signaling Inhibitors in Acute Myeloid Leukemia

In recent years, it has become increasingly apparent that BCL-2 inhibition in AML is a clinically highly effective approach - and the first BCL-2 inhibitor, venetoclax, has gained FDA and EMA approval. However, drug resistance frequently occurs and mechanisms or biological markers to predict response to BCL-2 inhibition are urgently needed. ASPP2 plays a central role to orchestrate induction of apoptosis via binding of p53 - but also via binding of antiapoptotic BCL-2, allowing release of proapoptotic proteins. We recently described an alternatively spliced oncogenic ASPP2 isoform (ASPP2k) (Schittenhelm et al., 2019), which is characterized by truncation of the C-terminus allocating the BCL2- as well as the p53 binding sites (in analogy to major TP53 mutations lacking the ASPP2 binding motifs). We therefore hypothesized that expression of ASPP2k attenuates efficacy of pro-apoptotic compounds:

AML cell line models, including MOLM13/14 (FLT3 ITD+), OCI-AML3 (NPM1 A+) and HL60, as well as freshly isolated native leukemia blasts (n=40) are used in dose-dilution assays to assess for pro-apoptotic efficacy in annexin V-based assays. Bone marrow donors served as a control population. A lentiviral approach was used for isoform-specific ASPP2k-shRNA transduction. A HisMax vector was used to forcely express ASPP2k. Several compounds targeting BCL-2 signaling, which are under clinical investigation, were tested (BCL-2: venetoclax, BCL-2/Xl: AZD4320, MCL-1: AZD5991 and CDK9: AZD4573).

To summarize, leukemia cells demonstrated variable and preferential sensitivity profiles towards the tested compounds: MOLM cells were sensitive towards all tested compounds. In contrast, OCI-AML3 proofed reduced sensitivity towards BCL-2/Xl inhibition - whereas MCL-1 and CDK9 inhibition (indirectly targeting MCL-1 signaling) showed IC50s in the nanomolar range (AZD5991: 307nM, AZD4573: 16nM). Furthermore, HL60 were relatively resistant towards both BCL-2/Xl and MCL-1 inhibition - however remained high sensitivity towards CDK9 inhibition (IC50 25nM). Priming with a hypomethylating agent (HMA, decitabine) resulted in additive (CI close to 1) to synergistic (CI 0.25 - 0.7) proapoptotic effects in isobologram analysis and led to a release of drug resistance in primary resistant cell lines.

Exposure of native leukemia cells towards all inhibitors confirmed (individually differing) sensitivity in the nanomolar ranges - whereas bone marrow donor controls were relatively resistant towards the tested compounds, which argues for a therapeutic clinical window.

In a last step, we addressed, whether ASPP2k functionally impedes the proapoptotic efficacy observed for the tested compounds. Indeed, we demonstrate that isoform-specific ASPP2k interference similarly results in a significant increase of pro-apoptotic capacities of all tested BCL-2, BCL-2/XL, MCL-1 and CDK9 inhibitors, (resp. attenuation thereof after forcely expressing the dominant-negative splicing variant).

To summarize, we show that inhibition of BCL-2 signaling is a promising approach to target acute leukemia - as a monoagent as well as in combination with HMA: Further, we provide a path for exploration of ASPP2k as a predictive tool as well as a therapeutic sensitizer of pro-apoptotic drugs, which will be addressed in future studies.