Identification of Novel Lncrnas That Predict Survival in AML Patients and Modulate Leukemic Cells

Long noncoding RNAs (lncRNAs) have complex, mainly chromatin-associated functions and their expression is highly coordinated and cell-type specific. Based on their tight regulation in normal differentiation, we set out to investigate whether lncRNAs are dysregulated in diseases where differentiation is impaired, such as in acute myeloid leukemia (AML). To identify lncRNAs that are essential for both normal hematopoiesis as well as AML maintenance, we sequenced the long polyA- and non-polyA-tagged transcriptome from successive stages of human myelopoiesis (myeloblasts, promyelocytes, metamyelocytes, and neutrophils) isolated from bone marrow of healthy donors (n=3). Applying a high-dimensional data portraying approach (OposSOM, Löffler-Wirth et al., BMC Bioinformatics, 2015), we identified functional expression modules of lncRNAs that are either positively or negatively associated with myeloid lineage commitment in our dataset. Seven out of the top15 differentiation-associated lncRNAs exhibit significant prognostic relevance in overall and event-free survival analyses of independent AML patient datasets and improve the predictive power of the current prognosis standards (cytogenetic risk/age/TP53-status). In particular, a combination of 3 transcripts, PROMYS (Promoter of Myelopoiesis, annotated as uncharacterized ncRNA LOC107985167), ANTAMY (Antagonist of Myelopoiesis, uncharacterized ncRNA LOC101927745) and LINC00677, outperformed the recently reported prognostic benefit of the LSC17^high score (Ng et al, Nature, 2016) by a factor of Ø 22.7 based on concordance index score increase (Ø 4.8% vs. 0.21%).

All three lncRNAs are highly conserved, expressed in 10 tested human AML cell lines as well as significantly differentially expressed in distinct cytogenetic patient subgroups of The Cancer Genome Atlas (TCGA) LAML cohort (n=171). PROMYS is downregulated in t(15;17) and t(8;21) cases, supporting its strong association with worse OS in the TCGA-LAML dataset (p=0.0001). In contrast, ANTAMY shows high expression in AML with t(8;21), and LINC00677 in NPM1+/FLT3- mutated AML patient samples with normal karyotype (CN-AML) and in core Binding factor (CBF) AMLs. Accordingly, high expression levels of both lncRNAs associate with a significantly better OS in the TCGA LAML dataset (p=0.01 and 0.02, respectively).

To investigate their function in vitro, we knocked out each lncRNA individually in the human OCI/AML-5 AML cell line using CRISPR/Cas9. Loss of ANTAMY impaired proliferation (p=0.04) and increased both monocytic differentiation upon treatment with 2-0-tetradecanoylphorbol-13-acetate (TPA) (p=0.0001) and granulocytic differentiation with all-trans retinoic acid (ATRA) (p=0.0002) compared to the empty vector control. Loss of LINC00677 in OCI/AML-5 cells specifically increased granulocytic differentiation through ATRA (p=0.0002). In contrast, inactivation of PROMYS led to reduced differentiation induced by ATRA (p=0.00004) and TPA (p=0.002). Furthermore, we found that PROMYS is involved in the regulation of the Macrophage colony-stimulating factor 1 (CSF1), which is deregulated in ATRA- and TPA-induced differentiation in PROMYS knockout but not in control cells (p<0.002 and <0.00002, respectively), explaining its negative impact on differentiation. Through screening of human myelopoiesis, we identified three unexplored lncRNAs: LINC00677, PROMYS, and ANTAMY, which play a role in myeloid differentiation and have an impact on patient prognosis. Our in vitro findings confirm that ANTAMY, LINC00677, and PROMYS are active modulators of leukemic cells, which influence their proliferation, morphology, myeloid marker expression as well as apoptosis rate. These transcripts and their interaction partners add an additional layer of regulation to the understanding of differentiation and might represent previously unknown vulnerabilities of AML cells, which warrants their further investigation in vivo.