Feedback Regulation of STAT5 Is Critical to Balance MYC and BCL6-Dependent Transcriptional Programs That Regulate Cell Size and Glucose Metabolism

Background: The BCR-ABL1 tyrosine kinase in Ph⁺ ALL and other ABL1-kinase fusions and JAK-kinase lesions that are found in the emerging subset of Ph-like ALL cause malignant transformation through constitutive phosphorylation of STAT5 at Y694. In addition, genetic lesions of genes encoding IL7R, CRLF2, EPOR are frequently found in Ph-like ALL and cause constitutive STAT5 Y694-phosphorylation. Ph⁺ ALL represents the most frequent subtype of human ALL in adults and accounts for ~30% of adult B-lineage ALL cases. Ph-like ALL is frequent among children (~10%) and young adults (~25%). Both Ph⁺ and Ph-like ALL represent highly aggressive diseases and define the group of patients with the worst clinical outcome.

STAT5 induces its own negative feedback through transcriptional activation of CISH and SOCS proteins. The cytokine inducible SH2-containing protein (CISH) binds to tyrosine kinase with its SH2 domain and induces its degradation through poly-ubiquitinylation and recruitment of cullins and other ubiquitin-ligases to its SOCS-box. Suppressor of cytokine signaling (SOCS) proteins can interfere with oncogenic tyrosine kinase signaling directly through its kinase inhibitory region (KIR). Interestingly, SOCS2, SOCS3 and CISH are highly expressed in Ph⁺ ALL and represent the focus of this study. Results: Gene expression analyses showed that mRNA levels of SOCS2 and CISH were 10-fold (p<0.0001) and 3-fold (p<0.0001) upregulated in Ph⁺ ALL cells (n=15) compared to normal pre-B cells (n=8) respectively. As measured by qRT-PCR and Western blot, tyrosine kinase inhibitor (TKI)-treatment rapidly decreased the expression of SOCS2, SOCS3 and CISH in Ph⁺ ALL cells, indicating that high expression levels of SOCS2, SOCS3 and CISH are induced by tyrosine kinase activity.

Importantly, we found that high expression levels of SOCS and CISH feedback inhibitors are predictive of poor outcome in the COG trial (P9906; n=207) for children with high-risk ALL: Patients with early minimal residual disease (MRD) findings on day 29 had much higher mRNA levels of SOCS2, SOCS3 or CISH (n=207; p<0.04), and patients with higher than median expression of SOCS2, SOCS3 or CISH had decreased overall survival rate (p<0.05) and relapse free survival rate (p<0.01) compared to patients with lower than median expression of SOCS2, SOCS3 or CISH. These results collectively indicate that inhibitory signaling from SOCS2, SOCS3 and CISH molecules may contribute to disease progression of pre-B ALL.

To study the function of Socs2, Socs3 and Cish in genetic experiments, we transformed pre-B cells from mutant mice. Deletion of Socs2, Socs3 or Cish resulted in increased STAT5-activation, increased ROS (reactive oxygen species), accumulation of Arf, p53 and p21, cellular senescence and subsequent cell death of leukemia cells. Studying BCR-ABL1 transformed Cish^fl/fl ALL cells in vivo, we found that genetic deletion of Cish resulted in rapid leukemia regression and prolonged survival of recipient NOD-SCID mice (13 days vs 30 days; n=7; p<0.001).

To study the mechanism of Cish-dependency of pre-B ALL cells, we studied the consequences of inducible expression of a constitutively tyrosine-phosphorylated form (pY) or inactive form that cannot be tyrosine-phosphorylated (YF) of Stat5 in BCR-ABL1 ALL cells. While both pY and YF forms of Stat5 were highly toxic, the phenotypic outcome was dramatically different. While Stat5-YF caused rapid cell shrinkage, Stat5-pY increased cell size by nearly 2-fold. Stat5-YF upregulated Bcl6 expression, decreased Myc expression, glucose consumption and ATP production. In contrast, Stat5-pY increased Myc expression, glucose consumption and ATP production. Metabolomics and RNA-seq results indicated that Stat5-pY increased glycolytic and TCA metabolism through mTOR activation. Inhibition of mTOR by rapamycin or knocking down Myc largely rescued leukemia cells that were transduced with Stat5-pY, while overexpression of Myc rescued Stat5-YF transduced cells.

Conclusion: Our results suggest that STAT5 feedback control is required to keep a balance between p-STAT5 (pY) and non-p-STAT5 (YF), which promote divergent, MYC (pY) and BCL6 (YF)-dependent transcriptional programs. Even minor imbalances between these two programs are toxic to ALL cells. These findings also identify SOCS proteins as members of a fundamentally novel class of therapeutic targets for ALL.