Mesothelin Promotes Acute Myeloid Leukemia Cell Proliferation, Adhesion, and Chemoresistance through Novel Binding Partner Lyn

Current treatment options for pediatric acute myeloid leukemia (AML) have little curative success, presenting a need for novel therapy options. Mesothelin (MSLN), a glycosylphosphatidylinositol-anchored protein with limited expression in normal tissues, is expressed in 33% of pediatric AML patients and represents a validated immunotherapy target. Although the exact function of MSLN is widely unknown, it has been implicated to play a role in adhesion through interactions with its only known binding partner MUC16/CA125.

To understand the function of MSLN in AML, we generated 2 MSLN knockout (KO) clones of NOMO-1 cells using CRISPR/Cas9 mutagenesis followed by single cell cloning. Lack of MSLN protein was confirmed via western blotting and flow cytometry. Transcriptome analysis of NOMO-1 and MSLN KO cells displayed differential expression of 434 genes with a log2FC>1. Gene-set enrichment analysis using a gene ontology database revealed enrichment of pathways such as leukocyte proliferation (NES=1.3; p<0.05), G2/M checkpoint (NES=1.3; p=0.055), oxidative phosphorylation (NES=1.6; p<0.01), glycolysis (NES=1.3; p<0.05), cell-cell junction assembly (NES=2.2; p<0.01), and positive regulation of cell adhesion (NES=1.7; p<0.01). Consistent with the pathway analysis, we observed that NOMO-1 cells had a significantly shorter doubling time compared to MSLN KO (39.8 h vs 53.7 h; p<0.05). NOMO-1 also had significantly higher EdU+ cells (22.4 ± 5.9%) than NOMO-1 MSLN KO cells (17.8 ± 2.1%; p<0.05). Cell cycle progression assayed by flow cytometry revealed higher proportions of cells in G2/M (13.4 ± 1.8%) in NOMO-1 compared to MSLN KO cells (6.4 ± 0.4%; p<0.05). Mito Stress test using Seahorse xF Pro showed NOMO-1 cells had significantly higher basal respiration normalized oxygen consumption rates (OCR) (38.3 ± 5.3 vs. 22.0 ± 5.1 pmol/min; p<0.0001), maximal respiration OCR (84.9 ± 20.7 vs. 37.6 ± 12.1 pmol/min; p<0.0001), and ATP-linked respiration OCR (15.1 ± 4.6 vs. 7.7 ± 3.9 pmol/min; p<0.0001). These results suggest that MSLN KO reduces cell proliferation, delays cell cycle progression and suppresses metabolic fitness of NOMO-1 cells.

Based on enrichment of adhesion-based pathways, we evaluated adhesion to human bone marrow extracellular matrix (ECM) MaxGel™. The percentage of NOMO-1 cells adhering to MaxGel™ was greater (40.6 ± 11.0%) than MSLN KO cells (21.5 ± 10.2%; p<0.01). Because we have shown earlier that leukemia cell adhesion modulates drug sensitivity, cells were plated with or without MaxGel™ and treated with Ara-C or vehicle. In non-coated wells, the percentage of viable NOMO-1 cells (69.7 ± 2.2%) exposed to 3 µM Ara-C for 48 h was higher than MSLN KO cells (59.4 ± 1.9%; p<0.0001), suggesting that MSLN promotes chemoresistance. Upon interaction with ECM, NOMO-1 cells but not the MSLN KO cells were protected from Ara-C and had a significantly higher chemoprotection index (8.1% vs -3.6%; p<0.0001), indicating that MSLN-mediated adhesion of NOMO-1 cells to ECM promotes Ara-C resistance.

To understand the mechanism of MSLN-induced Ara-C resistance, novel binding partners of MSLN were identified via immunoprecipitation followed by mass spectrometry. A comparison of proteins pulled down by MSLN in NOMO-1 versus MSLN KO cells identified 72 proteins with a log2FC>1. We shortlisted Src family kinase (SFK) member Lyn (log2FC = 2.80; p<0.0001) based on its probable role in cell signaling. Immunoprecipitation followed by western blotting confirmed the MSLN-Lyn interaction. Plasma membrane-derived lipid rafts isolated from NOMO-1 cells confirmed the enrichment of MSLN and Lyn in the lipid rafts, suggesting that MSLN may recruit Lyn to the lipid rafts. To investigate the role of Lyn in MSLN-induced resistance to Ara-C, we used saracatinib (pan-SFK inhibitor) and bafetinib (Lyn inhibitor) in conjunction with Ara-C. Zero-interaction potential (ZIP) scores for cell viability calculated by synergyFinder showed that both saracatinib (ZIP=20.09; median=21.23) and bafetinib (ZIP=11.65; median=15.13) worked synergistically with Ara-C, suggesting that Lyn activity is involved in MSLN-mediated drug resistance.

Taken together, our data support MSLN playing an oncogenic role through increased proliferation, cell cycle progression, metabolic fitness, ECM adhesion, and Ara-C resistance. We identified a novel MSLN-Lyn signaling axis that could be used to improve targeted therapy approaches.