NPM1-ALK Overexpression-Driven Toxicity in ALCL Is Partner Dependent and Mediated By STAT1 Antagonism of STAT3 Survival Signaling, Revealing a Novel Therapeutic Strategy

Activation of the tyrosine-kinase domain (TKD) of anaplastic lymphoma kinase (ALK) drives malignant growth in subsets of several cancers as part of oncogenic fusions resulting from chromosomal rearrangements. Clinically, 70% of anaplastic large-cell lymphoma (ALCL) and ~5% of non-small-cell lung cancer (NSCLC) are ALK+. ALK tyrosine-kinase inhibitors (TKIs) are approved for treatment of NSCLC and are under investigation in ALCL. Clinical resistance in lung cancer typically develops within 10-12 months and is due most commonly to activation of alternate pathways or second-site kinase-domain mutations. In contrast, in ALK + ALCL, we and others have shown that over-expression of the NPM1-ALK fusion kinase is the major mechanism of TKI resistance. While this permits cells to survive TKI treatment, it results in a toxic signaling overdose upon inhibitor withdrawal both in vitro and in vivo. Therefore, NPM1-ALK is both essential to the survival of ALK+ ALCL and able to induce apoptosis when activated above a certain threshold. We now report the mechanism underlying NPM1-ALK signaling overdose.

We find that all generations of ALK TKIs drive NPM1-ALK overexpression as the most frequent resistance mechanism in ALCL cells, but increased ALK activity upon inhibitor withdrawal is universally toxic. ALK stimulates several oncogenic pathways, including JAK/STAT3, MEK/ERK, and PI3K/AKT, but we find none of these canonical downstream pathways is a key driver of overdose toxicity. Novel analysis of viability data using a normalized weighted area under the curve to quantify level of rescue showed that only ALK inhibitors can rescue from ALK overdose signaling. We also observed that NPM1-ALK over-expression has a significantly more detrimental effect on cell fitness than the lung cancer fusion EML4-ALK, suggesting that the toxicity seen upon overdose is fusion-partner dependent. For confirmation, we generated constitutively active mutants of the ALK TKD with no fusion partner that are potently transforming but do not impair cell fitness upon ALK overdose. Phosphoproteomics analysis of resistant sub-clones during overdose as compared to parental lines in vehicle show an overlap of over 1,700 targets driven by ALK signaling. Furthermore, comparison of the parental line in drug to the resistant line undergoing overdose, 178 genes group together in the overall death state. These analyses revealed 21 phospho targets uniquely upregulated during ALK overdose (p <0.05). Annotation of these genes align with RNA-seq data showing increased phosphorylation of proteins driving the same pathways.

STAT1 was a prominent target further pursued as the driver of apoptosis during NPMI-ALK overdose. Normally activated by INF-γ, STAT1 antagonizes the transcriptional output of STAT3, the core survival pathway of ALCL. It previously was shown to be a direct phospho target of NPM1-ALK, which at steady state results in its proteasomal degradation permitting predominance of the STAT3 transcriptional program. Upon NPM-ALK overdose, however, we see rapid and dramatic upregulation of phosphorylated STAT1 (Y705), in both the cytoplasm and nucleus. RNA-Seq analysis at the gene level shows loss of STAT3 targets but up-regulation of the STAT1 program, suggesting that during NPM1-ALK overdose STAT1 activation antagonizes STAT3 survival output required for ALCL cell viability and driving them to apoptosis. IFN-γ, already known to be detrimental to ALK+ ALCL cells as a single agent, combined with ALK inhibitors therefore represents a novel treatment strategy able to synergistically deactivate the core survival pathway of ALK+ ALCL.