Systematic profiling of protein function and drug response mechanisms in lymphoid leukemias

Cancer biology is characterized by complexity, because disease progression depends on the interplay of molecular components that shape phenotypes collectively. To advance our understanding of cancer biology and overcome remaining therapeutic limitations, it is necessary to deeply and systematically account for this molecular plasticity.

For the most prevalent lymphoid leukemia malignancies in children and adults, respectively, acute lymphoblastic leukemia (ALL) and chronic lymphocytic leukemia (CLL), most patients can access relatively successful treatments. But standard therapies fall short for many patients, who continue to experience both poor treatment outcomes and complications of over-treatment. These under-served patients have a wide range of phenotypic and treatment-linked variability, and traditional approaches fall short of the power needed to systematically apply these insights productively. Applications that improve our understanding of protein variation specifically are critical, because proteins impact not only biology but also drug affinity and targeting.

The overall aim of the thesis was to investigate the functional and molecular plasticity of the proteome in lymphoid leukemia, with the goal of expanding precision therapy opportunities across the disease spectrum through proteogenomic and functional proteomics approaches.

Study I: We assembled a biobank of 49 commercially available childhood ALL cell lines (and 2 Epstein-Barr virus transformed B-cell lines). These were profiled using proteomics and transcriptomics, and drug responses were quantified for 528 oncology-related drugs suitable for repurposing applications. Molecular phenotypes were linked with drug sensitivity measurements and evaluated within the context of the other cell lines to identify therapeutic correlations and lineage-dependent effects. The primary aim was to perform multi-omic analyses of childhood ALL cell lines, identifying potential therapeutic candidates and their mechanistic associations. The outcome was a resource of cell line models with associated molecular phenotypes that can be used to infer potential sensitivities to targeted drugs.

Study II: We conducted thermal proteome profiling with deep peptide coverage in 20 B-cell precursor ALL cell lines. Using a graph-based model and the Leiden community detection algorithm, we clustered peptides according to their physical properties, which was inferred from thermal stability. This approach enabled the detection of functional proteoform groups, in which peptides from the same gene could be distinguished as separate proteoforms. The primary aim was to infer proteoforms from bottom-up proteomics data using thermal proteome profiling and to link differential results to disease biology and drug sensitivity.

Study III: Using the approach from Study Il for functional proteoform group detection, we systematically profiled ibrutinib target proteins. We distinguished between functional proteoform groups to better understand biologically relevant aspects of drug response. In a CLL patient cohort, profiled using proteomics, we linked specific peptides indicative of ibrutinib target proteins to annotate treatment associated phenotypes and ex vivo drug response. The primary aim was to refine the target landscape of ibrutinib, a key treatment in CLL known to have polypharmacology, and to explore the implications of proteoform variability in treatment response.

Study IV: To better understand what leads to bryostatin-1 sensitivity in MEF2D-rearranged (MEF2Dr) ALL, following up on a drug sensitivity finding observed in Study I, we profiled bryostatin-1 interactions and signaling in responsive and resistant B-cell precursor ALL cell lines. We tested the hypotheses built from broad multi-omics data with specific assays and chemical proteomics approaches, quantifying the effects of PKCδ activation, ERK phosphorylation, histone remodeling processes, and calcium signaling. The primary aim was to investigate the rationale of this selective vulnerability and to explore how pre-B-cell receptor signaling thresholds may differ in ALL.

By leveraging proteomics technologies, we systematically detected associations with drug responsiveness and enhanced the functional interpretation of this data by enabling the identification of functional proteoform groups. Together, we identified correlations between molecular phenotypes and drug responses in ALL cell lines, identifying bryostatin-1 as a therapeutic candidate in the MEF2Dr subtype and associating mediators of pre-B cell negative selection with drug susceptibility. Functional proteoform groups were resolved from bottom-up proteomics data using thermal proteome profiling, and could be linked to disease biology through differential co-aggregation and associations with drug sensitivity. Investigation of ibrutinib's target landscape implicated additional targets that perform immunomodulation and cellular processes such as Golgi and endosomal trafficking and glycosylation. Variability in proteoform profiles among CLL patients could be linked with treatment status and ex vivo response, suggesting complex biological effects linked to off-target engagement. Collectively, this work empowers nuanced detection of proteome features and introduces an approach that streamlines prioritization of therapeutic targets. Our findings emphasize that recognizing the spectrum of drug affinities and targets modulated by proteoform variability is crucial for improving the precision and efficacy of targeted therapies. Ultimately, this research contributes to efforts in developing more effective and personalized treatment strategies for leukemia patients.

List of scientific papers

I. Integrative multi-omics and drug response profiling of childhood acute lymphoblastic leukemia cell lines. Leo, I. R .* , Aswad, L .* , Stahl, M .* , Kunold, E., Post, F., Erkers, T., Struyf, N., Mermelakas, G., Joshi, R. N., Gracia-Villacampa, E., Östling, P., Kallioniemi, O. P., Pokrovskaja-Tamm, K., Siavelis, I., Lehtio, J., Vesterlund, M., Jafari, R .. et al. Nature Communications 13, 1691 (2022). * Equal contribution https://doi.org/10.1038/s41467-022-29224-5

II. Deep thermal profiling for detection of functional proteoform groups. Kurzawa, N., Leo, I. R., Stahl, M., Kunold, E., Becher, I., Audrey, A., Mermelakas, G., Huber, W., Mateus, A., Savitski, M., Jafari, R.. Nature Chemical Biology 19, 962–971 (2023). https://doi.org/10.1038/s41589-023-01284-8

III. Functional proteoform group deconvolution reveals a broader spectrum of ibrutinib off-targets. Leo, I. R., Kunold, E., Audrey, A., Tampere, M., Eirich, J., Lehtiö, J., Jafari, R.. Nature Communications 16, 1948 (2025). https://doi.org/10.1038/s41467-024-54654-8

IV. Epigenomic mechanisms link MEF2D fusions to toxicity of pre-BCR signaling in leukemia. Leo, I. R., Post, F., Struyf, N., Mermelekas, G., Qi, X., Pokrovskaja-Tamm, K., Joshi, R., Erkers, T., Lehtio, J., Jafari, R .. (2025) [Manuscript]