T-cell acute lymphoblastic leukemia (T-ALL) is an aggressive blood cancer caused by the uncontrolled transformation of immature T-cell precursors, which flood the bone marrow and shut down normal blood cell production. It affects both children and adults, and when patients relapse or fail to respond to initial therapy, outcomes remain poor.
A key driver of T-ALL is the Notch1 (N1) receptor, a transmembrane protein that normally guides cell fate decisions during development. When a ligand binds to Notch1 on a cell's surface, an enzyme called gamma-secretase cleaves the receptor, releasing its intracellular domain (NICD), which travels to the nucleus and switches on cancer-promoting genes.
Activating mutations in the Notch1 gene are found in roughly 75% of T-ALL cases, making it one of the most frequently mutated genes in this disease. This high mutation rate makes Notch1 an attractive drug target, leading researchers to develop gamma-secretase inhibitors (GSIs), small molecules that block the enzymatic cleavage step and thereby silence Notch1 signaling.
Despite scientific promise, clinical trials of GSIs have largely disappointed. The central reason is that leukemia cells develop resistance to these drugs, both before treatment begins (intrinsic resistance) and during the course of therapy (acquired resistance). Understanding the molecular basis of this resistance is critical for improving how T-ALL is treated.
This study used quantitative mass spectrometry-based phosphoproteomics to comprehensively map differences in protein activity between GSI-sensitive and GSI-resistant T-ALL cells. Phosphoproteomics measures which proteins in a cell are phosphorylated - a process where a phosphate group is added to a protein to switch it on or off - providing a snapshot of active signaling networks.
The researchers designed three complementary experimental models: a panel of six T-ALL cell lines with naturally different GSI sensitivities (Model 1); a single cell line (DND-41) that was grown in the presence of increasing GSI doses over 5-6 weeks until it became resistant, generating so-called persister cells (Model 2); and two patient-derived xenograft (PDX) models where human T-ALL cells were grown in mice until they developed resistance to an anti-Notch1 antibody (Model 3).
Across all three models, the team measured both the full proteome (all proteins present) and the phosphoproteome (all phosphorylated proteins) using liquid chromatography-tandem mass spectrometry (LC-MS/MS). In total, they quantified over 218,000 peptides and more than 16,000 phosphorylation sites - one of the most comprehensive analyses of T-ALL signaling to date.
By comparing resistant and sensitive states across multiple independent models, the researchers aimed to identify common mechanisms of resistance that are not specific to one particular cell line or laboratory artifact but represent true, shared features of how T-ALL cells escape Notch1 inhibition.
When comparing intrinsically resistant versus sensitive T-ALL cell lines at the protein level, the analysis revealed that resistant cells showed characteristic enrichment of proteins involved in small molecule metabolic processes, including fatty acid degradation and reactive oxygen species (ROS) metabolism. Sensitive cells, by contrast, were enriched in proteins connected to transcription and DNA-binding, consistent with their dependence on Notch1 transcriptional activity.
To find a unifying molecular explanation for these diverse metabolic changes, the team queried the Connectivity Map (CMap), a database of gene expression signatures in response to thousands of compounds. This analysis revealed that the sensitivity profile of T-ALL cells most closely resembled the molecular fingerprint of protein synthesis inhibitors, suggesting that resistant cells may rely less on Notch1-driven transcription and more on constitutive protein production.
Phosphoproteome analysis pointed to a convergent mechanism: mTOR signaling, a master regulator of cell growth and protein synthesis. The mTOR complex 1 (mTORC1) pathway was significantly more active in resistant cells, particularly in those lacking PTEN (a tumor suppressor that normally restrains mTOR). Validation using an external drug sensitivity database confirmed that GSI-resistant cell lines are significantly more sensitive to mTOR inhibitors such as rapamycin.
In acquired GSI-resistant persister cells, the study found that even though Notch1 signaling was fully suppressed by the drug, the oncogene cMyc (a key Notch1 target gene responsible for its cancer-promoting effects) was partially reactivated through alternative mechanisms involving the chromatin regulator Brd4. This epigenetic reactivation bypasses Notch1 inhibition, explaining why persister cells survive long-term drug exposure.
A central finding across all three resistance models was the identification of a distinct kinase signature - a pattern of altered kinase activity that distinguishes resistant from sensitive T-ALL cells. Kinases are enzymes that add phosphate groups to other proteins, and their activity profiles reveal which signaling pathways are running hot or cold inside a cell.
Among the kinases enriched in resistant cells, protein kinase C delta (PKC-delta) emerged as a shared feature across intrinsic and acquired resistance models, including the PDX animal models. PKC-delta is a serine/threonine kinase involved in regulating cell survival, differentiation, and stress responses, and its elevated activity in resistant cells suggested it might be keeping cancer cells alive despite Notch1 inhibition.
The researchers found that sotrastaurin, a clinical-grade PKC inhibitor originally developed for transplant rejection, could enhance the anti-leukemic activity of GSIs when both drugs were combined. In PDX mouse models, the combination showed superior effects compared to either drug alone, providing proof-of-concept for this therapeutic strategy.
Critically, sotrastaurin combined with GSI treatment completely prevented the development of acquired resistance in cell culture experiments. Cells that would normally evolve resistance over weeks of single-drug treatment were unable to do so when PKC was simultaneously blocked. This suggests that co-targeting PKC-delta alongside Notch1 inhibition could potentially close the resistance escape route.
Patient-derived xenografts (PDX) are a powerful tool for studying cancer therapy: actual human leukemia cells from a patient are transplanted into immunodeficient mice, where they grow and can be treated with drugs. Unlike standard cell lines, PDX models preserve much of the genetic diversity and clinical behavior of the original tumor.
This study used two Notch1-addicted T-ALL PDX models (PDTALL11 and PDTALL19), which were treated with an anti-Notch1 neutralizing antibody (OMP52M51) weekly until resistance emerged. Resistance appeared at different rates in the two models - after 43 days in PDTALL19 and 80 days in PDTALL11 - closely mimicking the variable timescales of clinical resistance observed in patients.
Leukemia cells were harvested from mouse spleens at the point of resistance and subjected to the same proteomic analysis used for the cell line models. Despite the biological differences between the two PDX models, the phosphoproteome signatures of resistance were strikingly similar, including the shared PKC-delta kinase signature, validating the clinical relevance of the findings from simpler in vitro experiments.
The convergence of resistance signatures across cell lines, acquired resistance models, and patient-derived xenografts represents strong evidence that the molecular pathways identified are genuine, biologically meaningful features of T-ALL resistance - not laboratory artifacts. This kind of multi-model validation is considered a gold standard in translational cancer biology.
This study demonstrates the power of quantitative proteomics and phosphoproteomics as tools for understanding cancer drug resistance. Unlike genomic approaches that reveal mutations at the DNA level, proteomics captures the actual functional state of signaling networks - what pathways are actively running at any given moment and which kinases are in control.
The study's approach of comparing multiple resistance models is particularly valuable because resistance is almost never caused by a single mechanism. By identifying the intersection of resistance features across diverse models, researchers can home in on the pathways that matter most and avoid pursuing strategies that only work in artificial systems.
The identification of PKC-delta and mTOR as actionable targets in resistant T-ALL is immediately clinically relevant because inhibitors of both kinases already exist as approved or clinical-stage drugs. Sotrastaurin (PKC inhibitor) and rapamycin or its analogs (mTOR inhibitors) have established safety profiles, meaning clinical trials testing these combinations with Notch inhibitors could move forward relatively quickly.
More broadly, the framework established here - using phosphoproteomics to map kinase activity signatures in drug-resistant cancers and then matching those signatures to available inhibitors - offers a systematic roadmap for overcoming treatment resistance in other cancer types where targeted therapies face similar challenges.
The central conclusion of this work is that Notch1 inhibition alone is insufficient to control T-ALL because leukemia cells reliably rewire their signaling networks to survive. The mTOR and PKC-delta pathways represent two key survival routes that resistant cells exploit, and both are vulnerable to existing clinical drugs.
The demonstration that combining a GSI with sotrastaurin can completely prevent resistance from emerging in laboratory models is an especially striking finding. Most cancer combination therapies are designed to kill more cells; preventing resistance is a different and arguably more important goal, as it could extend the durability of response in patients who initially respond to treatment.
Future work will need to test whether these combinations are safe and effective in patients, and whether the biomarkers identified here (such as PKC-delta activity levels or mTOR pathway activation) can be used to select patients most likely to benefit from specific combinations. If successful, this could mark a step toward genuinely precision medicine for T-ALL.