Phospho-Proteomic Profiling of T-Cell Acute Lymphoblastic Leukemia Identifies Specific Kinase Activation Signatures That Can Predict Response to Targeted Therapy

Introduction

In the last decade, intensive multi-agent combination treatment has boosted survival and cure to approximately 80% of pediatric T-cell acute lymphoblastic leukemia (T-ALL) patients. Nevertheless, patients that relapse have a very poor prognosis due to acquired therapy resistance. The current highly intensive chemotherapy is accompanied by severe toxicity, resulting in either death or frequent detrimental long-term side effects for survivors, with a relevant impairment of their quality of life. Therefore, there is an urgent need for new targeted therapies as well as reliable biomarkers that can predict treatment outcome and therapy resistance.

Protein kinase inhibitors are amongst the most successful small-molecule drugs used for cancer treatment. However, for T-ALL patients the use of protein kinase inhibitors is still limited to clinical trials. Furthermore, targetable kinases activated by gene fusions are rare in T-ALL in contrast to other types of leukemia, and are often present at the subclonal level (e.g. episomal amplification of the NUP214-ABL1 fusion in 6% of patients at diagnosis). Nevertheless, leukemic blasts rely on enhanced kinase signaling to sustain dysregulated proliferation. Therefore, protein kinases can be activated even in the absence of genetic mutations in their coding sequences. In addition to screening for disease-associated genomic alterations in patient biopsies, profiling of aberrant protein kinase activity may offer opportunities for targeted therapeutic approaches.

Aim

High-throughput phospho-proteomics can provide direct information on pathway activation, kinase signaling networks and may therefore be used to identify novel therapy targets. Here, we aimed to identify kinase activation patterns in established T-ALL cell lines and to test their differential sensitivity to targeted kinase inhibitors in order to develop a strategy to predict leukemia dependencies.

Methods

We performed unbiased, mass spectrometry-based phospho-proteomic profiling of 11 established T-ALL cell lines. By combining titanium dioxide-based enrichment with selective phospho-tyrosine immunoprecipitation, we identified about 3700 tyrosine phospho-sites and more than 13300 serine/threonine phosphorylation sites. Multiple approaches have been developed to infer kinase activity from phospho-proteomic data. We applied the Inferred Kinase Activity (INKA) scoring to rank kinase activation in our cell lines. This ranking is not only based on the phosphorylation of kinases but also integrates information on phosphorylated substrates (Beekhof et al, Mol Syst Biol. 2019).

Results

We found SRC-family members as most activated kinases in T-ALL cell lines, with a major role for LCK, SRC, FYN, and YES1. Certain cell lines also revealed high activity of ABL1, ZAP70, LYN, and FGR. Additionally, other active kinases identified include INSR, CLK1, CDK1/2/7 and PAK1/2. To test the dependency of the cell lines to the predicted SRC-family kinases activation, cellular response levels were measured towards dasatinib, a SRC/ABL multi-kinase inhibitor. Surprisingly, only the cell lines with known genetic kinases aberrations-e.g. HSB-2 (TCRB-LCK translocated) and ALL-SIL (NUP214-ABL1 rearranged)-showed strong sensitivity to dasatinib. The remaining 9 lines were resistant to dasatinib treatment and seemed not to be solely dependent on the predicted kinase activities for their survival, indicating the need for therapeutic combinations that target additional parallel kinase activities. To unravel actual kinase dependencies rather than kinase activities, analysis of the overall phosphorylation profiles yielded specific 'phospho'-signatures that associate with dasatinib responsiveness. Therefore, individual T-ALL cell lines can be used as calibrator to predict and quantify signaling dependencies that associate with response to targeted kinase inhibitors in primary T-ALL biopsies.

Conclusion

The identification of oncogenic dependencies by ranking kinases activities and their signaling networks from phospho-proteomic profiling data can guide the assignment of T-ALL patients to specific kinase inhibitors treatment. Furthermore, these phospho-signatures may provide important drug response biomarkers as well as explain possible compensatory mechanisms for therapy resistance.