Effects of ZMIZ1 mutations in newly defined early T-cell precursor leukemia

Abstract

Abstract Recent large scale genomic studies of T-cell acute lymphoblastic leukemia (T-ALL) patients from the Children's Oncology Group AALL0434 clinical trial identified two new overlapping high-risk T-ALL subtypes called early T-cell precursor-like ALL (ETP-like) and bone marrow progenitor-like ETP-ALL (BMP-like). ETP-like leukemia is a poor prognosis subtype that is defined by diverse alterations in genomic drivers in hematopoietic stem cell regulators. BMP-like ETP-ALL is notable for poor prognosis and high expression of genes associated with bone marrow stem and progenitor cells. There is an urgent need to find new, more effective treatments for these ETP-type leukemias because they resist conventional T-ALL treatment. Targeting rare mutational drivers of ETP leukemia in individual patients might be tedious. In contrast, targeting the BMP-like transcriptional network that drives the shared stem cell state might be more efficient. ZMIZ1 is a poorly characterized PIAS-like coactivator that promiscuously binds several transcription factor partners. ZMIZ1 expression is highly and selectively expressed in ETP leukemia. We recently reported that ZMIZ1 induces a native transcriptional network of feedforward circuits linking together BMP-like enhancers and oncogenes. This network contributes to cardinal features of ETP cancer, including association with poor prognosis, stem cell gene expression, and opposition to T-cell development. Since ZMIZ1 helps define the BMP-like state, targeting ZMIZ1 might be more efficient than individually targeting genetic alterations. Further, since Zmiz1-deficient mice are healthy, inhibiting ZMIZ1 might also be safe. While our previous work established the importance of ZMIZ1 in ETP leukemia, there remains limited knowledge on how to inhibit it. To address this, we examined the AALL0434 genomic dataset for ZMIZ1 mutations. Interestingly, we identified rare ZMIZ1 mutations that are monoallelic and somatic. Most of these mutations are clustered in the alanine rich domain (ARD). The ARD is a low complexity domain (LCD) that has no known function. Cancer-associated mutations that cluster in hotspots generally tend to be gain-of-function. To test whether ZMIZ1 mutations might be gain-of-function, we examined the transcriptomes of ZMIZ1-mutated samples in the AALL0434 dataset. Consistently, ETP leukemia samples with ZMIZ1 mutations express higher levels of ZMIZ1 ETP signature genes compared to non-mutated samples. Additionally, the ARD is an intrinsically disordered region (IDR). IDRs are often important for protein stability. To test whether ARD mutant proteins are more stable, we transduced cells with ZMIZ1 mutants and measured ZMIZ1 expression by western blot, flow cytometry, and qRT-PCR. Consistently, T-ALL-associated mutants showed 1.6-2.7-fold increased protein expression but no differences in mRNA levels. These data suggest that the ARD contains a degron signal. In preliminary studies, we generated a ZMIZ1 mutant that lacks the ARD domain (Delta-ARD). Delta-ARD showed increased protein levels. These data suggest that the T-ALL mutations disable a degron motif. Our study identifies the ARD as a domain that regulates ZMIZ1 protein abundance. Further, we illustrate a unique example in which mutations in adjacent amino acids can have distinct functional outcomes leading to divergent diseases leukemia.