Differentiation Status Revealed By Shotgun Phosphoproteomics Determines Sensitivity of Primary AML Cells to Kinase Inhibitors

Introduction

Protein kinases play a key role in how cells respond and adapt to intra and extracellular stimuli. By the addition of phosphate groups to serine, threonine or tyrosine residues, these enzymes modify the activity and properties of the targeted proteins which in turn modulate biological processes like proliferation, differentiation and cell death. Kinase signalling pathways are deregulated in most cancer types including haematological malignancies. Indeed, the kinases FLt-3, c-Kit and JAK2 as well as the up-stream kinase signalling regulators KRAS and NRAS are among the most frequently mutated genes in acute myeloid leukaemia (AML). Consequently, protein kinases have attracted the attention of the pharmaceutical and biotechnology companies and inhibitors have been found for one fifth of human kinases. In the case of AML, midostaurin, a multi-kinase inhibitor that targets, among others, the tyrosine kinase Flt-3, has granted a breakthrough therapy designation by the FDA and several other kinase inhibitors are in clinical trials or under preclinical investigation. Molecular profiling of patient samples will play a pivotal role for the development and implementation of personalized therapies including those based on kinase inhibitors. We used a molecular profile generated by a phosphoproteomics approach to rationalize why some primary AML cells respond to treatment with different kinase inhibitors while others are resistant to the same treatments.

Methods

Label free phosphoproteomics based on trypsin digestion and TiO₂ phosphoenrichment was used to quantify > 5,000 phosphorylation sites in mononuclear cells extracted from the peripheral blood of 36 AML patients. KSEA technology was applied to infer kinase activity from the phosphoproteomics data and DAVID software was used to determine gene ontology enrichments based on the genes that code for the proteins where the phosphorylation sites were detected. Guava EasyCyte Flow Cytometry was used to determine cell viability after the treatment of the same patient samples with different kinase inhibitors. Mass cytometry was used to measure the expression at the plasma membrane of 17 surface markers in 30 of the previously analysed AML primary samples.

Results

The FAB classification subdivide AML cases depending on cytomorphological features. We compared the phosphoproteomes of M1 and M4 classes that are associated with early and late states of differentiation. Based on the 150 phosphopeptides more significantly regulated between FAB-M1 and FAB-M4 groups, hierarchical clustering analysis was used to stratify AML patient samples into two subsets named M1-Like and M4-Like. Phosphoproteome reanalysis showed that the M4-Like set upregulated 1255 phosphopeptides and downregulated 446 when compared with the M1-Like set. The upregulated group comprised regulatory phosphorylation sites in several kinases including PAK1 and PCK delta. Kinase activity analysis using KSEA (Kinase Substrate Enrichment Analysis) also showed an increased activity of PAK, PKCδ and other kinases like P38 alpha in the M4-Like group. Interestingly, the PAK inhibitor PF03758309 reduced more efficiently the viability in M4-Like group than in the M1-Like group (average reduction after a 72h treatment with 1µM of 55.2% for M4-Like compared to 33.8% for M1-Like, p-value = 0.0078). This difference was not observed for other inhibitors such as those targeting CK2 or p38. CyTOF analysis showed that the M4-Like group upregulated the surface expression of several differentiation markers.

Discussion

Predicting the effectiveness of a drug for a particular patient is a major goal of personalized medicine. In the case of kinase inhibitors, responses may be influenced by several factors including the activity of the targeted kinase as well as the activity of other kinases that act in parallel pro-survival pathways. In this work, we have found that differentiation leads to a particular activation pattern of the signalling networks, a phenomenon that determines the response to signalling inhibitors.

Conclusion

We found phosphoproteomics signatures in primary AML that are associated with distinct haematopoietic differentiation stages. These signatures are in turn associated with how AML cells respond to kinase inhibitors.