New Insights into Deregulated Gene Expression Pathways in MLL- and AF10-Rearranged T-Lineage Acute Lymphoblastic Leukemia

For children, adolescents and young adults with T-lineage acute lymphoblastic leukemia, event free survival (EFS) following relapse is <10%. We recently showed that MLL-AF6, del3'MLL and other re-arrangements of the mixed lineage leukemia (MLL-R; KMT2A) molecular repertoire are associated with induction failure (IF) and an inferior EFS (Matlawska-Wasowska et al., Leukemia, 2016). However, we found that AF10 (MLLT10) gene structural alterations (AF10-R) did not confer a worse prognosis. While deregulation of HOXA9/10 is a hallmark of all MLL-R and AF10 lesions, its over-expression does not predict outcome or guide therapeutic decisions. Because there are currently no molecular features that inform treatment strategies in T-ALL, we hypothesized that supervised profiling of well-characterized cases might identify biologically relevant genes and signaling pathways for targeted T-ALL therapy. Using Affymetrix U133 Plus 2.0 microarray we performed a retrospective study of 100 T-ALL patients enrolled onto COG ALL0434. We performed a 3-way analysis of MLL-R Cases (n = 12), AF10-R Cases (n = 9), and Other T-ALL Cases (n = 79), and identified 330 probe sets that could discriminate between these groups. For the MLL-R Cases and AF10-R Cases, 258 and 40 probe sets, respectively, were uniquely different compared to Other T-ALL Cases. Within the MLL-R Cases group, the most deregulated genes included HOXA genes, MYO6, which encodesATP-dependent motor protein, SKAP2, a substrate of Src family kinases, transcriptional/translational regulators RUNX2, TCF4, SMAD1, CPEB2 (> 3-fold, P < 0.05) and interestingly WHAMMP2/WHAMMP3, and GOLGA8I pseudogenes (> 3-fold, P < 0.05). For the AF10-R cases, we found that MEIS1, HOXA3, 5, 7, 9/10, SPAG6, BMI1, and COMMD3 expression were deregulated, similar to the findings of others. We next identified genes that could discriminate between MLL-R and AF10-R cases. Among the most overexpressed genes, we identified MYO6, RUNX2, CPEB2, ZNF827 and TCF4 (> 3-fold, FDR < 0.05) indicating that MLL-R encompass a specific biological subset, which collectively drive a unique oncogenic transcriptional program. The most downregulated gene in both MLL- and AF10-R cases was QKI, which encodes RNA-binding protein involved in alternative splicing (> 3-fold, P < 0.05). Since MLL-AF6 confers an inferior outcome, we sought to determine which genes discriminate between MLL-AF6 (n = 5) and MLL-ENL (n = 5). Among the 26 discriminatory probes sets, we found two, MLLT4 and uncharacterized RP11-38P22 that were over-expressed and 24 were underexpressed, including SMAD1, CHI3L2, MYOM2 between AF6 and ENL, respectively (> 10 fold change, FDR < 0.05). As an extension of the study we performed Gene Set Enrichment Analyses to assess functional networks. Genes that were differentially expressed in MLL-R cases were enriched in regulators of secretion and endocytosis (NES = 1.84, FDR = 0.06), actin binding and cytoskeleton function (NES = 1.7, FDR ≤ 0.25), embryonic development (NES = 1.98, FDR ≤ 0.25), and genes upregulated by hedgehog signaling pathways (NES = 1.73, FDR ≤ 0.25). Genes that were differentially expressed in AF10-R include those that activate cytokine production (NES = 1.90, FDR ≤ 0.25). Taken together, these pathways are commonly targeted by inhibitors of kinase signaling pathways and the proteosome. In summary, we have identified an extended repertoire of aberrant gene expression in MLL-R and AF10-R T-ALL. Our findings provide a mechanistic basis for additional pre-clinical testing of in classes of therapeutic agents that may hold promise for high-risk T-ALL.