A common protein target for an uncommon subtype of AML

In this issue of Blood, Schmoellerl and colleagues have found that the commonly expressed cell cycle regulatory protein, CDK6, is a target for therapy in NUP98 FP induced acute myeloid leukemia.¹ 

It has become paradigmatic to describe acute myeloid leukemia (AML) as heterogeneous. In fact, one of the rules of my laboratory is that, “AML is complicated.” However, the work of leukemia researchers is to simplify the complexity in order to improve survival for patients diagnosed with AML. Molecular profiling has begun to clarify the necessary and sufficient events required for human AML and lead to new targeted therapies. Thus, Fms-like tyrosine kinase (FLT3) mutated relapsed AML is now treated with FLT3 inhibition based on recent trials.²  AML mutated in the isocitrate dehydrogenases, IDH1 or IDH2, can also now be targeted with oral, nontoxic but active targeted drugs.³  This approach has transformed care for AML patients, but these are relatively common mutations; recent studies have shown that a large number of genes can be mutated in AML.⁴  Many of these are recurrently but infrequently mutated. The challenge for leukemia researchers is to find molecularly targeted therapies that can be used in less common forms of the disease.

Schmoellerl and colleagues have used a series of powerful techniques to answer this question for AMLs with fusion genes involving the Nucleoporin 98 (NUP98) gene and found that such leukemias are dependent on a widely expressed protein, cyclin-dependent kinase 6 (CDK6). This represents an exciting translational advance as inhibitors of CDK6 are already Food and Drug Administration (FDA) approved. Recurrent translocations involving chromosome 11 lead to fusion of the NUP98 gene to >25 different recipient loci in leukemia.⁵  The exact mechanism of transformation by NUP98 fusion proteins (NUP98-FP) is incompletely defined but involves dysregulation of the HOXA cluster of proteins that likely act to maintain hematopoietic cells in a stem cell-like state. Schmoellerl and colleagues took 3 steps that are likely critical at arriving at a novel target protein. First, they generated murine models of 3 distinct NUP98-FP leukemias driven by inducible promoters. The use of multiple models is critical for sorting signal from noise in interpreting molecular studies using powerful omics methods. Second, they performed both transcriptional profiling and chromatin immunoprecipitation (ChIP) to identify targets of NUP98 fusions and compared data from the 3 different models to refine the gene list. Here, as is typical, gene expression analysis alone led to a list of hundreds of candidate target genes. However, after sorting for genes that are directly regulated by NUP98-JARID1A binding to promoter regions (as defined by ChIP-Seq), the authors identified 12 candidate genes as common direct transcriptional target of NUP98 fusion proteins. This included genes in the HOXA cluster, providing an internal control for robustness, and identified CDK6 as a NUP98-FP target. Finally, the authors validated using multiple models in murine and human cells that CDK6, one of the 2 kinase targets of the CDK4/6 inhibitor, palbociclib, is a bona fide target. Inhibition of CDK6 leads to a decrease in leukemogenesis in these models with induction of myeloid differentiation.

The curious biologic observation here is that a widely expressed cell-cycle regulatory protein, CDK6, has a distinct role in NUP98-FP–induced leukemia. CDK6 is a well-studied regulator of the G1-to-S transition and more recently defined roles in cellular development, DNA damage repair, centrosome stability, and metabolic hemeostasis. Given this wide array of functions, there was concern that targeting CDK6 would have toxicity.⁶  However, 3 CDK4/6 inhibitors have been FDA approved for use in breast cancer patients and have been well tolerated, although patients have an increased risk of cytopenias.⁷  This is 1 example in which human tumors have shown oncogene-determined selective dependencies for commonly expressed proteins and is proving to be a powerful way to develop novel therapies that are targeted not at the oncogene per se but at the downstream dependency. For example, the Der group recently identified CDK9 as a potential target for KRAS-mutated pancreatic ductal adenocarcinoma.⁸ 

An important question in evaluating these results is asking whether they are robust enough to justify clinical testing of CDK4/6 inhibitors for cases of relapsed or refractory NUP98-FP AML. My answer would be not yet. The authors demonstrate impressive efficacy of single-agent palbociclib in prolonging survival in 1 AML patient-derived xenotransplantation (PDX), the preferred model for preclinical studies. However, additional data using the PDX model would greatly inform any clinical use of this class of drugs, including additional studies using only a CDK4/6 inhibitor, combination studies with other agents, and, if prolonged survival is seen, secondary transplant experiments. These studies will be critical in determining if the identification of CDK6 as a potential target of NUP98-FP–induced leukemia will translate into a new therapeutic approach. Even without this preclinical data, however, the authors have provided a fascinating approach to identifying such targets that should be amenable to use with other, uncommon or rare molecularly defined subtypes of AML.