Proteomic Analysis Directs Effective Drug Selection in Relapsed AML By Quantifying Drug Targets

Introduction:

Standard therapy for acute myeloid leukaemia (AML) generally includes intense induction with daunorubicin (D) on days 1-3 and cytarabine (A) on days 1-7, followed by consolidation should complete remission (CR) be achieved. Assessment of bone marrow morphology, including percentage of blasts, remains the standard approach to gauge treatment response, however more sensitive molecular based approaches are capable of detecting subclinical levels of leukemic blasts (minimal residual disease, MRD). MRD often remains during and after standard treatment and is the main cause of relapse, a major problem in the management of AML. Resistance of the residual blasts to treatment can be attributed to the activity of pro-survival enzymes, some of which can be pharmacologically inhibited, however, finding the right inhibitor for the right patient presents a major challenge due to the plethora of different enzymes and combinations thereof. Liquid chromatography - tandem mass spectrometry (LC-MS/MS) proteomics enables global and unbiased quantification of protein expression and enzymatic activity in samples. We applied this technology to AML blasts at relapse compared to diagnosis, and in cell lines treated with standard chemotherapy to detect modulated biochemical pathways that contribute to resistance. Thorough investigation into the expression and activity of the protein drug targets enabled selection of inhibitors which proved effective when cells were treated in culture. This approach represents an effective way to better understand the biochemistry of cells following chemotherapy and identify suitable drug targets in biopsies to guide effective inhibitor selection.

Methods:

LC-MS/MS proteomics and phosphoproteomics was used to investigate global protein expression and kinase activity in primary AML samples at diagnosis and matched relapse (18 cases), and in 3 AML/APL cell lines before and after chemotherapy. Briefly, we collected frozen biopsy specimens from the Barts tissue bank and after thawing the AML blasts were incubated in media for 2 hr at 37^oC. Cell lines (HL60, MV411 and P31/FUJ) were treated ± D and/or A (2, 6, or 24 hr). After incubation, cells were centrifuged and washed in PBS, then proteins extracted in urea lysis buffer. Proteins were digested with trypsin, and resulting peptides analysed directly by LC-MS/MS for proteomics or subjected to phosphopeptide enrichment using TiO₂ for phosphoproteomics. Commercial (Mascot) and in-house (Pescal, KSEA) software were utilised to identify and quantify proteins, determine kinase activities and investigate intracellular signalling. Cell Viability of blasts ± treatments were recorded using the Guava ViaCount Reagent and Cytometer.

Results:

On average, >3000 proteins and >9,000 phosphorylation sites were identified per sample. One of the drug targets that correlated strongest with % blasts was CD99 (r=0.79). Blasts showed high abundance & activity of enzymes involved in DNA repair (e.g. PARP1, ATR and PRKDC) at diagnosis and relapse, several significantly increasing in relapse (e.g. PLK3 and APEX1). We observed significant increase in phosphorylation of signalling proteins, such as KIT and STAT5, in relapse. Other signalling pathways regulating survival, apoptosis and metabolism were modulated after relapse but these were patient specific. AML cell lines were more sensitive to D than A. HL60 was the most sensitive cell line while P31/FUJ were least sensitive. Chemotherapy significantly increased the activity of ATM, ATR, PRKDC and MAPKAPK2. Phosphorylation of HSPB1 increased significantly in the presence of D and/or A, and inversely correlated with sensitivity of cells to these drugs. Simultaneous inhibition of ATM & ATR significantly reduced P31/FUJ & MV411 cell viability ± A, while MAPKAPK2 inhibition increased sensitivity of MV411 cells to A.

Conclusion:

We identified the most abundant and active protein drug targets in AML primary samples and cell lines. Investigating primary AML at diagnosis and relapse uncovered changes in biochemical pathways that regulate DNA repair, survival, apoptosis and metabolism, some of them being modulated by chemotherapy in AML cell lines. These changes were often patient specific, suggesting that to effectively implement targeted therapies, a personalised approach is required and we demonstrate drug selection can be directed by LC-MS/MS proteomics.