Unravelling the Mirnome of MLL-Rearranged Acute Lymphoblastic Leukemia

BACKGROUND: MLL-rearranged acute lymphoblastic leukemia (MLLr-ALL) in infants (<1 year of age) represents an aggressive malignancy associated with highly unfavorable outcome. It is characterized by chromosomal translocations involving the MLL gene, with the majority (~80%) of these translocations resulting in the fusion of MLL to either AF4 in t(4;11)(q21;22), ENL in t(11;19)(q23;p13.3) or AF9 in t(9;11)(p22;q23). The respective MLL fusion genes, MLL/AF4, MLL/ENL and MLL/AF9, code for strong oncogenic drivers which rewrite the epigenetic landscape of the cell and profoundly alter gene expression. In recent years, several groups, including ours, have defined MLLr-ALL associated gene expression, DNA methylation and histone modification signatures, and these results have laid important corner stones for novel treatment rationales. However, a relevant regulatory mechanism of the cell, microRNAs (miRNA), has only been sparsely investigated in MLLr-ALL. Hence, in order to gain insight of the molecular pathobiology of this disease in all its aspects, elucidation of the MLLr-ALL associated miRNome is pivotal.

AIMS: This study aims at defining MLLr-associated miRNA expression patterns using high-throughput miRNA profiling of a comprehensive MLLr- and non-MLL B-cell precursor-ALL patient panel. Furthermore, in order to identify driver miRNAs directly regulated by MLL fusions, we use a gain-of-function model where the most common MLL fusion, MLL/AF4, is ectopically expressed in hematopoietic progenitor cells (HPC) derived from human embryonic stem cells (ESC). High-throughput miRNA profiling of these modulated cells and comparison with patient-derived miRNA signatures will allow to discern which miRNAs are directly involved in MLLr-driven ALL.

METHODS: In order to define MLLr-ALL miRNA patterns, we have used the Agilent miRNA microarray platform, which covers 1344 miRNAs. High-throughput profiling of primary patients samples (n=80) and healthy bone marrow (BM) controls (n=5) was performed. The patient cohort comprised 56 MLLr infant ALL patient samples, including the most common translocations MLL/AF4 (n=28), MLL/ENL (n=19) and MLL/AF9 (n=9). As a reference, non-MLL infant (n=12) and non-MLL pediatric B-cell precursor ALL patient samples (n=12) were assayed. Additionally, we also profiled ESC-derived HPCs transduced to express MLL/AF4 and corresponding backbone controls (n=3 independent pools). Statistically significant differential expression was defined as FDR-adjusted p<0.05.

RESULTS: We found expression of 344 miRNAs within the patient cohort. Unsupervised principle component analysis was able to separate MLLr-ALL from non-MLL ALL patients and healthy BM. Expression analysis between MLLr-ALL and non-MLL ALL identified 69 significantly differentially expressed miRNAs. Similarly, comparison of MLLr-ALL to healthy BM resulted in 97 differentials. Analysing each MLLr-ALL subtype separately against age-matched infant non-MLL ALL revealed n=44 differentially expressed miRNAs for t(4;11)+ and n=29 differentials for t(11;19)+ MLLr-ALL. Interestingly, comparing t(9;11)+ MLLr-ALL vs non-MLL ALL did not identify any significantly differential expression. Equally, patients with t(9;11) clustered away from the other MLLr-ALL patients and showed greater similarity to non-MLL infant ALL samples. This observation corroborates previous findings, suggesting t(9;11) to be an entity distinct from the other MLLr-ALL subtypes. Conversely, t(4;11)- and t(11;19)-specific miRNA signatures showed an overlap of 45%, indicating a more related pathobiology.

In addition to the patient panel, we profiled ESC-derived HPCs expressing MLL/AF4 and controls; 213 miRNAs were found to be differentially expressed. Comparison of this MLL/AF4+ HPC miRNA signature with patient MLLr - and MLL/AF4-specific patterns showed an overlap of >42% for each of the separate analyses, identifying a MLLr-specific miRNA core signature.

CONCLUSIONS: We have defined an MLLr-ALL core miRNA signature directly driven by MLL fusions, and are currently validating this link using RNA interference against MLL fusion transcripts. Moreover, ongoing functional analysis of these core miRNAs by overexpression or inhibition will identify which miRNAs play a role in MLL leukemogenesis, unravelling associated pathways and thereby providing rationales for urgently needed novel treatment strategies.