The Direct Notch1 Cofactor Zmiz1 Differentially Regulates Notch1 Signals in a Stage-Specific Manner to Preserve Early T-Cell Precursors and Expand Committed T Cells

When stem cells first enter the thymus and become early T-cell precursor (ETP) cells, they are exposed to high levels of Notch1 ligand. Notch1 signal strength must be tightly regulated because on one hand, excessive Notch1 signals drive premature T-cell commitment, resulting in loss of ETP cells and alternative cell fates. On the other hand, complete loss of Notch1 signals impairs ETP proliferation, also resulting in loss of ETP cells. Thus, keeping Notch signals finely balanced in ETP cells preserves "stemness". However, after ETP cells commit to the T-cell lineage by the DN3 cell stage, Notch1 signals ramp up dramatically to drive proliferation. It is unclear how Notch1 signals are initially restrained and then amplified. We previously showed that the PIAS-like coregulator Zmiz1 is a direct, context-dependent cofactor of Notch1 in T-cell leukemia. In contrast to drosophila Zmiz1, mammalian Zmiz1 evolved a tetratricopeptide repeat (TPR) domain that binds directly to Notch1 and selectively induces oncogenic target genes such as Myc through its transactivation domain (TAD).

To understand the role of Zmiz1 in T-cell development, we bred conditional Zmiz1 and Notch1 knockout mice to Cre strains that delete floxed genes in hematopoietic cells (VavCre/MxCre), in early T cells (LckCre), and in late T-cells (CD4Cre). Like deletion of Notch1, deletion of Zmiz1 caused an early defect at the ETP stage and a late defect at the DN3 stage. The defect with Zmiz1 deletion was less severe than with Notch1 deletion (~4-fold reduction for Zmiz1 deletion versus ~8-fold reduction for Notch1 deletion). Unexpectedly, the ETP defect in Zmiz1-deficient mice partially phenocopied excessive Notch1 activation with increased differentiation to DN2 cells and loss of ETP cells and alternative cell fates (myeloid and NK). To confirm this effect, we plated Zmiz1-deficient hematopoietic stem and progenitor cells (HSPCs) directly on OP9-DL stromal cells. Accordingly, these cultures recapitulated the in vivo phenotype, including suppression of myeloid cells. Reducing Notch signals slightly with modest doses of Notch inhibitors restored myeloid differentiation. In contrast to the ETP defect, the DN3 defect resembled Notch1 loss-of function. Accordingly, overexpression of activated Notch1 or a fusion protein containing only the Notch-interacting domain (TPR) and the TAD was sufficient to rescue the DN3 block.

To determine mechanism, we performed RNA-Seq in sorted ETP and DN3 cells from Zmiz1-deficient mice and mice treated with the anti-NRR Notch1 antibody. Zmiz1 coregulated ~16% of Notch1 target genes in ETP cells and ~24% in DN3 cells. In ETP cells, Zmiz1 primarily acted as a repressor of Notch1 target genes. Enrichment analyses showed that Notch1 promoted changes associated with T-lineage commitment, such as induction of Dtx1, Notch3, and Ptcra. In contrast, Zmiz1 reversed these changes. Although generally antagonistic with each other, Zmiz1 and Notch1 concordantly activated a minority of target genes important for proliferation, such as Myc. Upon differentiation to the DN3 cell stage, Zmiz1 switched from primarily a repressor of Notch1 target genes to primarily an activator, inducing Myc and Wnt pathway genes. Accordingly, overexpression of Myc in Zmiz1-deficient DN3 cells was sufficient to rescue the DN3 block in OP9-DL culture. To determine whether Zmiz1 needed to bind Notch1 in order activate or repress Notch1 target genes, we used HSQC NMR to identify amino acids in the TPR that were required for Notch1 binding. Two amino acids, R14 and E34, were confirmed by reporter and co-IP assays to be critical for the Zmiz1-Notch1 interaction. The Zmiz1(R14A+E34A) mutant, which was incapable of binding Notch1, failed to induce Myc. In contrast, this mutant retained full ability to repress Notch target genes associated with T-cell commitment.

These data suggest that in ETP cells, Zmiz1 preferentially restrains Notch1 T-cell commitment genes. However, after T-lineage commitment, Zmiz1 switches primarily into a Notch1-interaction mode that preferentially promotes Notch1 signals. It has been puzzling how Notch1 can drive seemingly conflicting biological processes of self-renewal and commitment. Zmiz1 appears to be one solution that evolved to differentially regulate Notch1 signals at a given Notch dosage strength in order to preserve ETP "stemness" while at the same time expanding committed T cells.