Exploring the Expression Profile of Long Non-Coding RNA (lncRNA) in Different Acute Myeloid Leukemia (AML) Subtypes: t(8;16)(p11;p13)/MYST3-Crebbp AML Harbors a Distinctive Lncrna Signature

Introduction: Long non-coding RNAs (lncRNAs) have recently emerged as important actors in the regulation of multiple cellular processes including cancer. Acute myeloid leukemia (AML) is a heterogeneous disease; most of the main cytogenetic AML subgroups harbor a specific gene expression profile. AML with translocation t(8;16)(p11;p13) (t(8;16) AML) is a subtype with specific clinical and biological characteristics including a distinctive  gene (Camós et al, Cancer Research 2006) and microRNA (Díaz-Beyá et al, Leukemia 2013) expression profile. In this translocation, MYST3 on chromosome 8p11 fuses with CREBBP on chromosome 16p13.3. The MYST3-CREBBP fusion protein is able to interact with multiple transcription factors (TF) producing a disturbed transcriptional program. However, the lncRNA expression pattern of different cytogenetic AML subtypes, including t(8;16) AML, have not been described yet.

Aims: To examine the expression profile of lncRNAs within different AML subtypes, and to characterize the expression pattern of lncRNAs in t(8;16) AML in comparison to other AML subtypes.

Patients and Methods: 46 AML patients, 4 normal bone marrow (NBM) and 3 CD34+ NBM samples were included in the study. Samples included different AML subtypes: intermediate-risk cytogenetic AML (IR-AML, n=18), t(15;17) (APL, n=4), t(8;21) AML (n=4), inv(16) AML(n=2), t(6;9) AML (n=7), AML with monosomal karyotype (n=4), t(3;3) AML (n=1), t(9;11) AML (n=1) and t(8;16) AML (n=5). Within IR-AML patients with a different mutational profile: FLT3-ITD (n=7), NPM1 (n=5), CEBPA (n=7) and DNMT3A (n=6) were included. The lncRNA expression was studied using Affymetrix® Human Gene 2.1 ST platform which includes 9698 lncRNAs transcripts. The filtering and normalization of the array data was performed using oligo package from Bioconductor. Statistical analyses were performed with TiGR MultiExperiment Viewer, BRB tools and R. The Transcription factor Affinity Prediction Web Tool was used to determine the putative transcription factors binding to the differentially expressed lncRNAs promoters.

Results: The hierarchical cluster analysis showed that all 4 NBM as well as all 3 CD34+ NBM clustered together according to their lncRNA expression. Interestingly, all 5 t(8;16) AML samples clustered together, as well as the 3 APL, the 7 t(6;9) AML and 5 out of 7 cases with CEBPA mutations. The specific lncRNA signature of APL was composed of 79 differentially expressed lncRNA and t(6;9) AML lncRNA signature comprised of 15 differentially expressed lncRNAs. When we focused on t(8;16) AML lncRNA profile, we identified an specific 113-lncRNA signature in the supervised analysis (Figure).

Interestingly, when we analyzed which (TF) had motifs overrepresented in the promoters regions of the t(8;16) AML lncRNA signature, we identified GATA2 as the TF with significantly overrepresented motifs for GATA2 (p<0.001). Interestingly, levels of GATA2 were differentially expressed in t(8;16) AML samples in comparison with other AMLs samples (p<0.001).  GATA2 has been described to interact with CREBBP, one of the partners involved in t(8;16) AML. Of note, 4 overexpressed lncRNAs of the signature (linc-HOXA11, HOXA11-AS, HOTTIP and NR_038120) were located in the HOXA genomic region, previously found upregulated in t(8:16) AML. Since several studies suggest an active crosstalk between microRNAs and lncRNAs, we also correlated the expression of these lncRNAs with the microRNA t(8;16) AML profile. We found significant correlation between linc-HOXA11 and miR-222* (R² =0.996, p=0.003), HOXA11-AS and miR-let-7c (R²= 0.994, p=0.006), HOTTIP and miR-196b*(R²= 0.958, p=0.041), and NR_038120 with miR-486-3p (R²=0.999, p=0.0004) and miR-19a (R²= 0.953, p=0.04).

Conclusions: LncRNAs expression profile seems specific of several AML subtypes, including t(8;16) AML. Some of the lncRNAs of this distinctive signature in t(8;16) AML are located in the HOXA genomic region, and correlate with several of the characteristic microRNAs previously described in this entity. Interestingly, we have predicted in silico GATA2, which interacts with CREBBP, as the most significant TF that could potentially regulate this lncRNAs signature. Nonetheless, further investigation is warranted to determine the mechanisms leading to this lncRNA signature and to identify the specific targets of these lncRNAs.

Río Hortega CM13/00205, FIS PI13/00999