Targeting the RAS/MAPK pathway with miR-181a in acute myeloid leukemia

Deregulation of microRNAs' expression frequently occurs in acute myeloid leukemia (AML). Lower miR-181a expression is associated with worse outcomes, but the exact mechanisms by which miR-181a mediates this effect remain elusive. Aberrant activation of the RAS pathway contributes to myeloid leukemogenesis. Here, we report that miR-181a directly binds to 3′-untranslated regions (UTRs); downregulates KRAS, NRAS and MAPK1; and decreases AML growth. The delivery of miR-181a mimics to target AML cells using transferrin-targeting lipopolyplex nanoparticles (NP) increased mature miR-181a; downregulated KRAS, NRAS and MAPK1; and resulted in decreased phosphorylation of the downstream RAS effectors. NP-mediated upregulation of miR-181a led to reduced proliferation, impaired colony formation and increased sensitivity to chemotherapy. Ectopic expression of KRAS, NRAS and MAPK1 attenuated the anti-leukemic activity of miR-181a mimics, thereby validating the relevance of the deregulated miR-181a-RAS network in AML. Finally, treatment with miR-181a-NP in a murine AML model resulted in longer survival compared to mice treated with scramble-NP control. These data support that targeting the RAS-MAPK-pathway by miR-181a mimics represents a novel promising therapeutic approach for AML and possibly for other RAS-driven cancers.

RAS proto-oncogenes encode small GTPase proteins, that is, KRAS, NRAS and HRAS, that are involved in homeostatic mechanisms of proliferation, differentiation and apoptosis of normal cells [29]. Whereas KRAS and NRAS are frequently mutated and activated in AML, HRAS mutations are rare, and HRAS wild-type expression is the lowest with respect to the other RAS isoforms in the hematopoietic system [29]. Aberrant activation of RAS signal transduction is often found in human neoplasia [30][31][32][33][34][35][36][37][38][39][40][41][42][43]. In hematopoietic malignancies, including AML, activating oncogenic RAS mutations contribute to malignant phenotypes by phosphorylating and activating downstream effectors such as the mitogen-activated protein kinase kinase (MAPKK, also known as MEK), mitogen-activated protein kinase (MAPK), and the PI3K-AKT downstream effectors, thereby promoting aberrant cell proliferation and survival [29]. However, to date, an effective therapeutic approach targeting RAS directly remains to be developed.
Recently, KRAS was shown to be a direct miR-181a target in oral squamous cell carcinoma [44]. Additionally, NRAS and the RAS-downstream effector MAPK1 are in silico predicted to be putative miR-181a targets. We hypothesized that higher miR-181a levels attenuate AML aggressiveness by targeting RAS and/or its downstream effectors in myeloid blasts, thereby reducing proliferation and decreasing the apoptotic threshold. Therefore, we reasoned that the delivery of synthetic miR-181a mimics may increase the low endogenous levels of miR-181a in AML blasts and lead to anti-leukemic activity.

Anti-leukemic activity of miR-181a
We previously reported that chemotherapy-treated patients with AML with higher miR-181a expression achieved complete remission (CR) more frequently and had longer survival compared to lower miR-181a expressing patients [2,9]. In line with these clinical observations, we and others showed that miR-181a expression is associated with a higher sensitivity to cytarabine in AML cell lines [45,46].
To further support the putative tumor suppressor activity of miR-181a, we engrafted NSG mice with virally transduced MV4-11 cells through a tail vein. The median survival for the animals engrafted with the lenti-miR-181a, lenti-sc and lenti-anti-181a transduced cells were 43, 33.5 and 28.5 days, respectively ( Figure 1D). Compared to the control group (lenti-sc), the lenti-anti-181a mice lived significantly shorter (P = 0.002, log-rank test) and lenti-miR-181a mice significantly longer (P = 0.02). Though the mice in three groups showed survival time differences, they all died from AML-like disease (Supplementary Figure S1).
We concluded that higher miR-181a expression leads to a less aggressive AML phenotype, thereby functionally validating the previously reported prognostic results [2,9,28].
Next we showed that the modulation of KRAS, NRAS and MAPK1 expression by miR-181a was caused by direct binding to the respective 3′-UTRs. We first validated KRAS as a direct miR-181a target. We identified two miR-181a-binding sites in the KRAS 3′-UTR and observed a 28 ± 4% (P = 0.003) and a 25 ± 1% (P = 0.007) downregulation of luciferase activity on site 1 and site 2 after co-transfecting 293T cells with miR-181a compared with scramble expressing controls. Mutations in the seed sequences of the KRAS 3′-UTRs rescued the miR-181ainduced downregulation ( Figure 2B). Next, to demonstrate that NRAS is also a direct miR-181a target, we cloned the predicted miR-181a-binding-site in the NRAS 3′-UTR into a luciferase reporter, and we observed a 26 ± 6% (P < 0.0001) downregulation of luciferase activity. An introduced mutation in the seed sequence rescued the miR-181a-induced downregulation ( Figure 2C). We also identified two putative miR-181a binding sites in the MAPK1 3′-UTR. Because of the short distance between the two binding sites (149 base pairs), we cloned the two binding sites into the same luciferase reporter construct. We observed a 33 ± 2% (P = 0.0002) downregulation of luciferase activity with miR-181a treatment compared to cells with scramble control treatment. When we mutated the two sites separately, we observed a 13 ± 3% (site 1; transduced with lentiviral constructs either overexpressing miR-181a (lenti-181a), expressing a scramble sequence (lenti-sc; = control) or a knock-down construct of miR-181a (lenti-anti-181a). (c) Five million lentiviral transduced cells were engrafted subcutaneously in NOD/SCID mice. At day 11, tumors from lenti-anti-181a and lenti-sc group (n = 3 in each group) were isolated and weighed (no tumor in lenti-181a group). At day 23, tumors from lenti-181a and lenti-sc group (n = 3 in each group) were isolated and weighed. (D) 1.5 million lentiviral transduced MV4-11 cells were engrafted into NSG mice. Survival curves of the mice in the three groups. www.impactjournals.com/oncotarget P = 0.004) and a 15 ± 3% (site 2; P = 0.006) miR-181ainduced downregulation of the luciferase activity. However, mutations on both sites of MAPK1 could completely rescue the miR-181a-induced downregulation ( Figure 2D). Collectively, these results support that KRAS, NRAS and MAPK1 are direct miR-181a targets.
Having shown that the Tf-NP-delivery-system was able to deliver miR-181a to AML blasts, we next tested the impact of Tf-NP-miR-181a on RAS activity. First, we found that the delivered synthetic miR-181a was functional, as it downregulated KRAS, NRAS and MAPK1 proteins (KG1a: 4.3, 4.4 and 5.5-fold; OCI-AML3: 3.2, 3.9 and 2.2-fold; MV4-11: 1.5, 4.4 and 4.6-fold, respectively) compared to Tf-NP-sc treatment ( Figure 3B). Compared to Tf-NP-sc, Tf-NP-miR-181a decreased p-MEK protein by 6.8, 2.2 and 4.5-fold and p-AKT , NRAS (c) and MAPK1 (D) wild-type or mutated 3′-UTRs and miR-181a mimics or scramble mimics (as controls). The firefly luciferase activity was normalized to Renilla luciferase activity. The data are shown as relative luciferase activity of miR-181a mimic transfected cells with respect to the scramble control of nine data points from three independent transfections. Error bars represent the standard deviation (SD). protein by 2.0, 2.5 and 5.7-fold in KG1a, OCI-AML3 and MV4-11 cells, respectively ( Figure 3B). Finally, we assessed the expression of the oncogenic transcription factor MYC, whose protein stability is enhanced by the RAS-MAPK1 phosphorylation pathway [48]. There was a 4.8, 4.3 and 7.8-fold reduction of MYC protein in KG1a, OCI-AML3 and MV4-11 treated with Tf-NP-miR-181a compared to those treated with Tf-NP-sc control ( Figure 3B). Consistent with these results, anti-miR-181a treatment resulted in upregulation of the KRAS, NRAS and MAPK1 proteins in HL60 cells that present with higher levels of endogenous miR-181a ( Supplementary  Figures S2A and S3).
To validate these results, we treated primary AML blasts having activated RAS from three AML patients Figure S2) with Tf-NP-miR-181a and again observed an increase in miR-181a ( Figure 3C). After 24 hours, mature miR-181a levels increased 45 ± 4, 35 ± 0.1 and 125 ± 16-fold, respectively, in the three patient blasts samples treated with Tf-NP-miR-181a compared to the Tf-NP-sc treated controls, whereas levels of miR-181b and miR-140 remained unchanged ( Figure 3C). Increased levels of miR-181a resulted in decreased protein levels of KRAS, NRAS and MAPK1 by 6.3, 6.8 and 5.6-fold in patient 1; 6.4, 1.6 and 19.7-fold in patient 2; and 2.3, 2.4 and 3.4-fold in patient 3, respectively ( Figure 3D). Downregulation of RAS and MAPK1 resulted in RAS-MAPK1 inhibition, decreased MEK and AKT phosphorylation and decreased MYC levels. We observed a 1.4, 3.5 and 2.0-fold decrease of p-MEK, 1.8, 9.3 and 2.0-fold decrease of p-AKT, as well as a 5.3, 7.6 and 2.8fold decrease of MYC normalized in the patient blasts treated with Tf-NP-miR-181a compared to Tf-NP-sc treatment ( Figure 3D).
In summary, we showed the effective delivery of miR-181a via Tf-conjugated nanoparticles and in turn downregulation of KRAS, NRAS and MAPK1 and inhibition of the RAS-MAPK1 and RAS-AKT-kinase signaling cascade.

systemic delivery of tf-NP-mir-181a in an AML mouse model
Next, we examined the anti-leukemic activity of Tf-NP-miR-181a in vivo. Saline (control), Tf-NP-sc or Tf-NP-miR-181a were administrated (1.5 mg/kg/d miR three times/week) through a tail vein 10 days after the engraftment of MV4-11 cells in NSG mice (each group n = 11). Randomly, three mice from each group (i.e. saline, Tf-NP-sc or Tf-NP-miR-181a treated group) were sacrificed after eight treatment doses. The spleen weights were measured and resulted in 187.3 ± 25.93 mg, 174.3 ± 13.65 mg and 77 ± 50 mg (vs. Tf-NP-sc; P = 0.03) in the saline, Tf-NP-sc and Tf-NP-miR-181a groups, respectively ( Figure 5A). The spleen weight was 58.3 ± 10.5 mg for agematched blank control mice (Supplementary Figure S6). Cytospins of bone marrow cells and histopathology of sternum, spleen and liver sections from MV4-11 cell engrafted mice treated with either saline or Tf-NP-sc showed infiltration of blast cells. In contrast, cytospins of bone marrow cells and histopathology of sternum, spleen and liver from Tf-NP-miR-181a treated leukemic mice were similar to that of the age-matched blank control groups ( Figure 5B). Furthermore, the population of leukemic cells in spleen samples, measured by flow cytometry, was significantly reduced in mice treated with Tf-NP-miR-181a compared to mice treated with Tf-NP-sc or saline ( Figure 5C). We observed a 2.6-fold and a 35-fold increase of miR-181a levels in MV4-11 cells harvested and sorted from bone marrow and spleens, respectively, in the Tf-NP-miR-181a treated mice compared to Tf-NPsc ( Figure 5D and 5E). In these cells, RAS and MAPK1 proteins were downregulated in the Tf-NP-miR-181a treated mice ( Figure 5D and 5F).
The median survival time of the remaining mice was 26, 28.5 and 35 days for the animal groups treated with saline, Tf-NP-sc and Tf-NP-miR-181a, respectively. Tf-NP-miR-181a treatment significantly reduced the disease burden and prolonged survival compared to Tf-NP-sc (P = 0.0002) or saline (P = 0.0001) treatment ( Figure 5G). Interestingly, Tf-NP-sc treatment also had some minor anti-leukemic effects compared to the saline treated control group (P = 0.04). www.impactjournals.com/oncotarget
In other types of cancers miR-181a has been associated with both tumor suppressor and oncogene functions [20][21][22][23][24][25][26][27][28], implying context-specific effects. Whereas in colorectal cancer [20] and lymph node metastasis in oral squamous cell carcinoma [21] a high miR-181 level seems to be associated with worse clinical outcomes, in glioma this miR has tumor suppressor function [22]. In these brain tumors miR-181a was shown to target the anti-apoptotic genes BCL2 and MCL1, and downregulated miR-181a reduced glucose deprivationinduced apoptosis and caused mitochondrial dysfunction in astrocytes [22,49,50]. The miR-181-family has been reported to be an effector in inflammatory response by TNF-α, IL-6, IL-1β, IL-8 and IL-10 [17,18,[51][52][53]. With regard to AML, we previously provided preliminary evidence that miR-181 may target elements of the "inflammasome" that ultimately lead to NF-κB activation and leukemia growth, while Li et al. showed that miR-181 promoted apoptosis, reduced viability and delayed leukemogenesis in MLL-rearranged AML by downregulating the homeobox gene PBX3 [28]. Bai et al. also demonstrated that miR-181a may reduce BCL2 and thus enhance chemosensitivity of AML cells [46]. However, the mechanisms through which miR-181a attenuates disease aggressiveness and the full spectrum of its targets still remain to be fully understood in AML.
Here, we first demonstrated that miR-181a targets the RAS-MAPK1 and RAS-AKT pathways, which have been found to be activated and support AML leukemogenesis [54][55][56][57][58]. Despite extensive efforts, the direct therapeutic targeting of these pathways with small molecule inhibitors remains challenging [59]. Our results show that KRAS, NRAS and MAPK1 proteins may be effectively reduced by utilizing RNA compounds mimicking miR-181a. The efficient delivery of miR-181a mimics by Tf-NPs decreased the targets and their downstream effectors (AKT, MEK, MYC). Altogether, our results support miR-181a replacement as a potential anti-leukemic, RAS targeting strategy in AML. The therapeutic advantage of using miR mimics is in the simultaneous targeting of cross-talking signal transduction pathways (STPs) [38]. Although a use of synthetic mimics may be of relatively difficult in therapeutic application especially compared to the use of anti-miR oligonucleotide, it has been postulated for several types of cancers and is currently being tested in clinical trials (e.g. for miR-34 in NCT01829971). One of the limitations of miR-based therapies is in the optimal delivery of these oligonucleotides as they are subject to rapid hepatic uptake and metabolism and are easily degraded by endonucleases in biological matrices. Nevertheless, we recently reported a novel anionic lipopolyplex nanocarrier system that was designed for the purpose of allowing for efficient miR delivery to AML cells [47]. Here we show that this system could be adapted to the delivery of miR-181a mimics and exert an efficient inhibitory effect on the RAS-MAPK1 and RAS-AKT kinase pathways, thereby resulting in a significant anti-leukemic activity. Interestingly, a very mild anti-leukemic effect and a slight downregulating effect of Tf-NP-scramble treatment on NRAS, KRAS, and MAPK1, as well as on MEK phosphorylation and MYC expression in OCI-AML3 cells was observed. This effect was likely mediated by one of the components of our nanoparticle system, for example, linoleic acid. It has been reported that some fatty acids have anti-tumor activity [60][61][62].
Other strategies to increase miR-181a have also been tested by our group with significant results. In a previous study, we demonstrated that lenalidomide increases endogenous miR-181a [45], by enhancing the expression of C/EBPα isoforms, which bind to the miR-181a promoter and induce the transcription of miR-181a. However, lenalidomide has several unwanted side-effects at the doses necessary to achieve plasma concentrations at which miR-181a was increased. Thus, the targeting NPs that we reported here may present the advantage to be more specifically directed to AML blasts, thereby sparing normal tissues and perhaps reducing unwanted toxicity. Our preclinical studies showed encouraging results with no toxicity in NP-treated mice at doses inducing antileukemic effects [47]. It should also be underscored that we and others have reported that increased levels of miR-181 lead to enhancement of sensitivity to chemotherapy in AML models [45,46,63]. Furthermore, patients with higher levels of miR-181a have a better complete remission rate and longer survival compared with those with lower levels, further supporting a role of this miR as a modifier of the response to chemotherapy [9]. Thus, we envision that potential clinical benefit of miR-181a replacement will be more likely if applied in combination with chemotherapy.
In summary, we unveil here a previously unreported activity of miR-181a that directly downregulates NRAS, KRAS and MAPK1 and RAS-dependent downstream signals supporting leukemogenesis. We showed that a nanoparticle-based delivery system could be used to efficiently increase otherwise low levels of miR-181a and achieve anti-leukemic activity in AML models with no evident toxicity. On the basis of our results, miR-181a-NP may warrant further evaluation for potential clinical applications in AML and other RAS-dependent malignancies.

MAtErIALs AND MEtHODs cell lines and patient samples
KG1a, MV4-11, HL60, HEK 293T and HEK 293TN cells were obtained from ATCC (Manassas, VA); OCI-AML3 cells were obtained from DSMZ (Braunschweig, Germany). Primary, unselected AML blasts from apheresis samples collected from nine patients were obtained from The Ohio State University (OSU) Leukemia Tissue Bank. Patients signed an informed consent to store and use their tissue for discovery studies according to OSU institutional guidelines.

Luciferase assays
Luciferase assays were carried out as previously described [7]. 293T cells were co-transfected with luciferase vector (pGL4.24), Renilla control vector and miR-181a mimic or scramble control. Luciferase activity was normalized to Renilla activity. See supplementary material for more detailed information.

Quantitative rt-Pcr (qrt-Pcr)
Total RNA was extracted with TRIzol reagent (Invitrogen). cDNA was synthesized using Superscript III (Invitrogen) or the Taqman miR Reverse Transcription kit (Applied Biosystems, Foster City, CA) for miR-181a, miR-181b, miR-140 and U44. qRT-PCR was performed with Taqman gene expression assays (Applied Biosystems) following the manufacturer's protocols. miR-181a, miR-181b and miR-140 expression were normalized to U44. KRAS, NRAS and MAPK1 expression were normalized to GAPDH. The comparative cycle threshold (C T ) method as previously described was used for relative quantification of gene expression [47].

colony assays
Methylcellulose colony formation assays were carried out as previously described [66] and counted after 15 days.

In vivo studies
Animal studies were performed according to the Ohio State University institutional guidelines. A total of 5 million lentiviral transduced MV4-11 cells were injected subcutaneously into eight-week female NOD/SCID gamma mice (NSG; The Jackson Laboratory, Bar Harbor, ME). At day 11, 3 mice from each lenti-anti-181a and lenti-sc group were sacrificed, and tumors were weighed. At day 23, 3 mice from each lenti-sc and lenti-181a group were sacrificed, and tumors were weighed.
For the therapeutic study, six-week-old NSG mice were injected with 0.3 million MV4-11 cells intravenously through a tail vein. The treatment started 10 days after the engraftment. Mice were treated with saline, Tf-NPsc or Tf-NP-miR-181a (1.5 mg/kg/d three times/week).
Randomly, 3 mice of each group were sacrificed after 8 doses of treatment for pathology analysis. Age-matched NGS mice without MV4-11 cell engraftment were used as blank control. The treatment was continued for the remaining mice. Eight mice from each group were monitored for survival. The experiment was repeated for biomarker analysis. Bone marrow and spleen cells were isolated from scarified mice and sorted for human CD45positive cells for further analysis.

statistical analysis
Data are presented as mean ± SD of at least 3 independent experiments and analyzed by the two-tailed Student's t-test. The mean and SD were calculated and displayed in bar graphs as the height and the corresponding error bar, respectively. Mouse survival was calculated using the Kaplan-Meier method, and survival curves were compared by the log-rank test.