Ets1 Enhances Context-Dependent Notch1 Activity in T-Cell Leukemia

The discovery of NOTCH1 as the most frequently mutated oncogene in T-ALL patients raised hopes for targeted therapy in this cancer. Unfortunately, in clinical trials, the pan-Notch inhibitor GSI caused excessive GI toxicity. Mice treated continuously with GSI die from intestinal stem cell loss and severe intestinal secretory cell metaplasia. Intermittent dosing of GSI is tolerable, but has weak anti-cancer effects. Thus, the challenge has been to find ways to selectively disable Notch in T-ALL. Our idea to meet this challenge stems from work by others showing that Notch cannot activate enhancers by itself. Notch requires a favorable "chromatin context" at its enhancers that is created by cooperating transcription factors. In theory, one could target cell-specific factors at these enhancers in order to avoid the intolerable effects of pan-Notch inhibition. In support of this, others showed that ubiquitous deletion of the T-cell specific Notch-dependent Myc enhancer in mice impairs T-ALL proliferation and thymopoiesis, but has no effect on other tissues. We previously showed that the transcriptional coactivator Zmiz1 is a direct cofactor of Notch1 that selectively promotes Notch activity at the T-cell Myc enhancer. However, it was unclear what other factors promote context-dependent Notch activity.

Ets1 is an attractive candidate. It can bind nucleosome-occupied regions in T-cell precursors and most Notch response elements in T-ALL cells, including the T-cell MYC enhancer. To investigate its importance, we generated conditional Ets1 knockout mice. Deletion of Ets1 in hematopoietic cells using the VavCre transgene caused a 21-fold loss of thymocytes starting at the earliest stage. This was 4-fold more severe than the loss of thymocytes in Notch-deficient mice. Deletion of Ets1 using a ubiquitous tamoxifen-inducible Cre caused a Notch loss-of-function phenotype in the intestine with a 1.4 to 2.3-fold increase in goblet cells. This was milder than the effects of GSI (3.3 to 4.2-fold increase). ~64% of the Ets1-deleted mice died from unclear causes. In vivo deletion of Ets1 in Notch1-induced murine T-ALLs reduced blast counts by 30-fold and prolonged survival. In a panel of human T-ALL cell lines, on average, knockdown with two different shEts1 reduced proliferation by 2 and 9-fold respectively over ~1.5 weeks of culture. This was superior to the effects of GSI (up to 2-fold inhibition). A small molecule inhibitor of Usp9x, the deubiquitinase of Ets1, induced Ets1 protein degradation and impaired T-ALL cell proliferation with submicromolar GI50. In PDX models, shEts1 reduced circulating blasts by 44-fold and prolonged survival.

To identify the mechanism by which Ets1 promotes T-ALL, we performed endogenous co-IP assays, which showed that Ets1 interacts with Notch1 and its cofactor Zmiz1. Further, Ets1 binding by ChIP correlated with Zmiz1 binding (R²=0.93). Knockdown of Ets1 reduced Zmiz1, Ets1, and Notch1 binding to enhancers of major T-ALL oncogenes, MYC and IL7R. RNA-Seq showed that Ets1 co-regulates the expression of ~30% of Notch1 target genes. Multiple MSigDB enrichment analyses of both Ets1 and Notch-regulated genes showed that the MYC and MTORC pathways were the #1 or #2 most enriched list. Enforced expression of Myc partially rescued the proliferation of human T-ALL cell lines deprived of Ets1. Based on these data, we predicted that Ets1 inhibition would sensitize enhancers to Notch inhibition. Accordingly, Ets1 withdrawal promoted the effects of GSI in repressing Myc expression and cell proliferation. Further, in our mouse model of Notch-induced T-ALL, Ets1 deletion in combination with intermittent doses of GSI reduced blast counts and prolonged survival more effectively than either treatment alone.

Our data support an emerging model in which cofactors like Ets1 create a favorable chromatin context for Notch1 to activate a subset of response elements. The context dependence of Ets1 action, which promotes certain oncogenic signals of Notch1 in T cells, might be clinically relevant. Ets1 deprivation inhibited thymopoiesis and leukemic proliferation more effectively and with less intestinal toxicity than Notch deprivation. Our data suggest that inhibiting Ets1, possibly through targeted protein degradation, would combat important drivers of the Notch pathway with reduced adverse effects linked to pan-Notch inhibition.