miR-451a abrogates treatment resistance in FLT3-ITD-positive acute myeloid leukemia

Acute myeloid leukemia (AML) is a fatal disease with up to 95% of patients remaining incurable. While supportive care has been increased in quantity and quality; substantial progress of AML therapy itself is still lacking. Primarily two predicaments challenge current therapy: first, the heterogeneity of the disease; second, the phenomenon of treatment resistance. Although many AML patients initially respond to therapy, the majority of patients relapse caused by chemoresistant clones, ultimately leading to the incurability of the disease. Approximately a quarter of AML patients present with FLT3-gain-of-function mutations. This usually entails a constitutive activation of FLT3 receptor downstream signaling pathways and changes gene expression patterns from healthy to malignant. On account of the poor prognosis of FLT3-ITD AML patients, FLT3-inhibitors are under development and in clinical testing. Nevertheless, acquired treatment resistance persists. This underlines the necessity of a greater knowledge on the origins of resistance as well as new approaches abrogating treatment resistance. Therefore, we propose to deviate from the common strategy of target inhibition and focus on the reinforcement of negatively regulated downstream targets of wellknown oncogenes such as FLT3-ITD. Because microRNA (miR) expression was shown to be altered by FLT3-ITD, this group of small non-coding RNAs is of interest in this work. By binding to the 3’UTRs of mRs, miRs post-transcriptionally regulate gene expression. Thus, miRs hold key regulatory functions in processes such as hematopoiesis and leukemogenesis. In accordance with this, various miRs have been shown to impact treatment resistance as well as sensitivity toward AML therapy. First clinical trials disclosed the potential of miR-inhibitors to be a promising therapeutic option. Nevertheless, no miR-based experimental drugs are currently available in AML. Therefore, we screened for differentially expressed miRs in bone marrow mononuclear cells from healthy donors and AML patients with either FLT3-wildtype, mutations in the FLT3-tyrosine kinase domain (TKD) or FLT3-ITD mutations (Fig. 1a, Supplementary Table 1). Here, we found miR-451a to be significantly (p= 0.002) lower expressed in FLT3-ITD AML patients compared to healthy individuals; while among the remaining AML samples screened, miR-451a levels were lowest in FLT3ITD patient samples (Fig. 1b). In agreement with this, we observed increasing miR-451a levels upon differentiation of human and murine hematopoietic cells (Supplementary Figure 1a-c). In addition, we analyzed the AML patient cohort of “The Cancer Genome Atlas” (TCGA) and discovered a similar miR-451a expression distribution (Fig. 1c). In order to confirm our observations in vitro, we assessed miR-451a levels by qPCR in FLT3-ITDtransduced U937 and 32D cells (Fig. 1d,e). Indeed, a stable overexpression of FLT3-ITD led to a decrease in miR-451a levels. In contrast, interruption of FLT3-ITD signaling by treatment of MV4;11 cells with FLT3 inhibitors quizartinib and gilteritinib (currently evaluated in clinical trials) caused an increase in miR-451a expression (Fig. 1f). As a healthy control, we treated mononuclear cells from umbilical cord blood (UCB) with quizartinib as well as gilteritinib and observed neither a significant change in


miRNA expression profiling
MicroRNA expression profiling was performed as previously described 1 . 500ng total RNA extracted using Trizol method was used for small RNA library preparation with the TruSeq™ Small RNA Sample Prep Kit v2 (Illumina) according to the manufacturer's instructions. The barcoded libraries were size restricted between 140 and 165 bp, purified and quantified using the Library Quantification Kit-Illumina/Universal (KAPA Biosystems) according to the manufacturer's instructions. A pool of up to 12 libraries was used for cluster generation per lane. Library DNA at a concentration of 10 pM was clustered using an Illumina cBot according to the SR_Amp_Lin_Block_Hybv8.0 protocol of the manufacturer. Sequencing of 50 bp was performed with an Illumina HighScan-SQ sequencer using version 3 chemistry and the version 3 flow cell according to the instructions of the manufacturer. For data analysis, Cutadapt software was used to trim adaptor sequences from raw sequences. Alignment to human mature miRNA sequences of miRBase was done from sequences with a length between 15-27 bases with bowtie aligner. Quantification of aligned reads was performed using R/Bioconductor programming environment.

miRNA detection by quantitative real-time PCR
Isolation of RNA, reverse transcription and quantitative PCR reaction of miRNAs was performed as previously described 2 . RNU44 was used for normalization of human microRNAs, snoRNA-135 for normalization of mouse microRNAs. All microRNA primers (miR-451a, sno-RNA135, RNU44) were obtained from Life Technologies.

Flow cytometry and fluorescence activated cell sorting (FACS)
For flow cytometry analysis, 1x10 6 cells were washed with PBS and stained for 20 min at 4°C with anti-MDR1 antibody (Clone UIC2, e Bioscience) or isotype control. Subsequently, cells were washed in PBS and measured with BD LSR II cytometer using CellQuest software (BD Biosciences). For Annexin V staining, PE Annexin V Apoptosis Detection Kit I (BD Pharmingen) was used according to the manufacturer's instructions. Final evaluation of data was done using NovoExpress 1.0.2 (ACEA Biosciences).

Primary human and murine cell samples
AML patient samples as well as samples from healthy donors were obtained from Leipzig University Hospital (Leipzig, Germany). The study protocols employed were approved by Ethical Committee at the Medical Faculty, Leipzig University and all patients provided informed consent. Primary human bone marrow mononuclear cells were isolated by means of Ficoll-Hypaque density gradient centrifugation using human Pancol (Pan-Biotech) according to the manufacturer's instructions. Hematopoietic CD34 + cells were isolated from fresh umbilical cord blood using a CD34 + MicroBead kit (Miltenyi Biotec) as previously described 2 . Primary murine samples were separated by the Lineage Depletion Kit (Miltenyi Biotec) according to the manufacturer's instructions.

Transient transfection and lentiviral transduction
Transfection of MV4;11 cells was performed using Lipofectamine 3000 (Thermo Fisher) according to the manufacturer's instructions. Transfection efficiency was calculated using flow cytometry for GFP and was between 20-40%. PGhU6 lentiviral vector was kindly provided by Dr. Alberich-Jorda and described previously 4 . Pseudo-viral particles were produced as previously described 4 . MV4;11 and MV4;11-R cells were transduced by spin infection for 90min at 2200 rpm and 35°C with at least 5 MOIs, incubated for 24 hours, washed two times with PBS and sorted for GFP by FACS another 48 hours later.

Cell treatment and viability assessment
Parental and resistant MV4;11 cells were plated at a density of 5*10 5 cells per 1 mL in 12 well plates. Solvent of therapeutic compounds were always kept at a dilution of 1:1000. After 24 h, fresh media and treatment were added and after 48 h of treatment, cell viability was measured with the MTS cell Proliferation Assay Kit (Abcam #197010) according to the manufacturer's protocol. Probes were analyzed at a wavelength of 450 nm. For the growth curves, cells were plated at a density of 100.000 cells/mL and treated. Cell count was performed daily.

DNA constructs and cloning
Cloning of human precursor miR-451a into pcDNA6.2 and pGhU6 has been done using the appropriate sequence provided by miRBase: 5'-CTTGGGAATGGCAAGGAAACCGTTACCATTACTGAGTTTAGTAATGGTAATGGTT CTCTTGCTATACCCAGA-3' and performed as previously described for pGhU6 4 and pcDNA6 3 . For luciferase assay, the 3´UTR of MDR1 amplified from genomic DNA of U937 cells was cloned into the pmiR-GLO vector (Promega). Mutagenenesis was done with the Q5® site directed mutagenesis kit by NEB.

Statistical Analysis
To determine the statistical significance of the results, the student´s t test as well as the Wilcoxon test for patient samples was conducted. A p-value of 0.05 or less was considered significant (*), while a p-value of 0.01 or less were considered as highly significant (**). The results were depicted as the median ± SD from three independent experiments. All graphs were generated with either Microsoft Excel or Graph Pad Prism 6.