Long Non Coding RNA BALR2-CDK6 Axis Influences Cell Differentiation and Metabolic Status of Acute Myeloid Leukemia

Purpose: Non-coding RNA (ncRNAs) genes are at least as frequent as protein-coding genes in the human genome; however, our knowledge on their function is still preliminary. Long ncRNAs are arbitrarily defined as transcripts that are over 200 nucleotides in length, and demonstrate great heterogeneity in their molecular mechanisms. Their emerging role as new players in cancer development and therapy response is widely supported by several scientific reports, and we recently found the LncRNA BALR-2 expression significantly correlated with a decreased response to prednisone treatment in B-cell precursor acute lymphoblastic leukemia (Bcp_ALL). Given its association with mixed lineage leukemia (MLL) rearrangements in Bcp-ALL, we investigated BALR-2 involvement in acute myeloid leukemia (AML) leukemogenic pathways.

Methods: We retrospectively analyzed by RQ-PCR bone marrow samples of 132 children with de novo AML harboring different genetic abnormalities ( CBF,MLL, and NUP98 rearrangements, FLT3-ITD , other rare translocations and patients without any recurrent molecular abnormality) diagnosed between 2002-2014 in one of the AIEOP centers and treated according to AML-2002/01 protocol. We correlated the expression of BALR-2 with patient event-free survival (EFS) and gene expression profile (GEP, HTA affymetrix 2.0). Using gene set and single sample enrichment analysis (GSEA, ssGSEA), we compared the gene expression signatures of patients with either high (4^th quartile) or low expression (1^st+2^nd+3^rd quartiles) of BALR-2. SHI-1 cell line was used to perform in vitro experiments aimed at examining the function of BALR2.

Results: The expression of BALR-2 was found higher in all AML samples as compared with those collected from healthy volunteers. Moreover, we did not find any correlation between BALR-2 expression and any specific genetic abnormality. We subdivided patients by using BALR-2 expression quartiles and found that that patients with higher BALR-2 expression (4^th quartile, n=32) had worse, although not statistically significant, EFS when compared to that of patients allocated to the 1^st+2^nd+3^rd quartiles (n=100; EFS= 53% versus 61%). We discovered that the 4^th quartile was enriched for patients who did not reach complete remission (CR) after induction therapy (28% vs 12%, p=0.03). Supervised clustering analyses of GEP performed in 58 samples showed that cases belonging to the 4^th quartile (n=18) clustered separately from the remaining quartiles (n=40), irrespectively of the underlining genetic abnormality shown by these patients. In particular, we found that 57 coding and 12 non-coding RNAs were significantly differentially expressed (Fold Change|2|, p<0.01), and CDK6 , being chromosomally adjacent to BALR2, was overexpressed in the 4^th quartile. We further investigated this putative BALR2-CDK6 relationship and found a positive correlation between BALR-2 and CDK6 expression in the whole cohort analyzed (Pearson correlation>0.7, p<0.05). We silenced BALR-2 in vitro and found this induces lower CDK6 mRNA and protein levels, as well as phospho-RB, its direct target. In line with what seen in Bcp-ALL, BALR2-CDK6 knockdown decreased cell proliferation, increased apoptosis, and sensitivity to etoposide treatment. Of note, BALR-2 depletion increased myelomonocytic differentiation, as evidenced by CD11 and PU.1 markers, both of which are Runt-related transcription factor ( RUNX1 ) targets. In agreement with studies demonstrating that CDK6 acts as RUNX1 inhibitor, we found that additional 81 RUNX1 target genes showed marked down-regulation in patients with high BALR-2-CDK6 expression. Through ssGSEA, an aberrant regulation of processes regulating mitochondrial mass and activity was found in patients with the 4^th quartile of BALR-2 expression (p<0.05).

Conclusions: Taken together, our data suggest that BALR-2 expression is associated with worse outcome in AML. Furthermore, our data suggest a cis-regulatory transcriptional relationship between BALR-2 and CDK6, which, in turn, antagonizes RUNX1 and enhances myeloid differentiation. Lastly, in vitro approaches are ongoing to define the mitochondrial signaling events involved in BALR-2's ability to alter the metabolic status in AML, which in turn could lead to novel approaches for inducing anti-leukemic effects.