We hypothesized that miRNAs could play a critical role in governing molecular mechanisms responsible for the establishment and growth of melanoma metastasis, such as chemotaxis, adhesion, migration, survival, or proliferation in the host microenvironment. First, we performed a loss-of-function in vivo screen for metastasis suppressors using a lentiviral library of miRNA decoy vectors targeting and inhibiting 291 conserved mouse and human miRNAs ( Figure 1A ). The efficacy of this decoy library for miRNA knockdown and loss-of-function studies has been previously demonstrated (25). As a melanoma model, we utilized 113/6-4L cells, which have been previously described to display tropism to the liver after subcutaneous injection (26). In our hands, these cells have limited ability to colonize the brain upon ultrasound-guided intracardiac injection, a well-described model of brain metastasis (27, 28). We transduced the 113/6-4L melanoma cells with the miRNA decoy Library or Empty Vector, which encodes Green Fluorescent Protein (GFP) but no miRNA decoy. Cells were transduced at a low multiplicity of infection [(MOI) 0.3] to achieve a single vector copy per cell, and then GFP+ cells were sorted to obtain a pure population of vector-expressing cells. The cells were injected into the left ventricle of athymic/nude mice, and after 2–3 months, when mice showed signs of discomfort or weight loss, they were euthanized and organs were collected for analysis. The number of brain and liver metastases was scored and GFP-positive metastatic foci were dissected, followed by deep sequencing for the detection of enriched decoys in the metastatic lesions ( Figure 1B ). To determine whether any particular miRNA decoys were enriched in the metastases, we extracted DNA from the tissues and performed deep sequencing to determine the frequency of each decoy in the library. Strikingly, out of the 291 vectors in the library, the miR-124a decoy was the dominant decoy in the metastatic lesions found in 3 out of 4 different mouse brains ( Figure 1C ). These findings suggested that inhibition of miR-124a promotes 113/6-4L melanoma metastasis to the brain.

miR-124a was initially described as a neuron-specific miRNA and is not detected at appreciable levels in cells outside the nervous system (29, 30). It has an important role in neuronal differentiation (31) and was subsequently shown to function as a tumor suppressor gene in the brain and found to be lost in glioblastoma (32). However, a role for miR-124a in melanoma metastasis has not been described, though it was intriguing that its inhibition was associated with brain metastasis. To validate miR-124a’s potential function as a metastasis suppressor, we transduced 113/6-4L melanoma cells with lentiviral vectors carrying mCherry-luciferase and a GFP-tagged single decoy against miR-124a (Dc-124a) or a scrambled decoy control (Dc-Scr). We confirmed that the miR-124a decoy efficiently inhibited miR-124, as miR-124a levels were reduced ( Figure 2A ), an indicator of decoy activity (33), and established targets of miR-124a EZH2 (34), MAPK14 (35), and SPRY2 (36) were significantly upregulated ( Figure 2B ). Nude mice were inoculated with Dc-124a- or Dc-Scr-transduced melanoma cells in the left heart ventricle and monitored by luminescence imaging throughout the experiment ( Figure 2C ). The group inoculated with Dc-124a-expressing cells displayed a significantly accelerated brain and extracranial metastasis burden relative to control (**p = 0.0069 and ***p = 0.001, respectively) ( Figures 2D, E ). Accordingly, miR-124a knockdown significantly shortened mouse survival in this model (**p = 0.0055) ( Figure 2F ). These data validate the findings of our positive selection screen, confirming miR-124a as a metastasis suppressor miRNA.

To further validate these findings, we conducted a gain-of-function approach. We transduced 113/6-4L cells with a lentivirus expressing mCherry-luciferase, together with a lentivirus overexpressing miR-124a coupled to GFP (miRH-124a) or its scrambled control (miRH-Scr). MiR-124a upregulation ( Figure 3A ) effectively reduced the expression of its established targets EZH2, MAPK14, and SPRY2 ( Figure 3B ). miRH-124a- or miRH-Scr-transduced cells were inoculated in the left ventricles of nude mice using ultrasound-guided intracardiac injection ( Figure 3C ). Mice were euthanized 5 weeks post-injection when some started to display signs of discomfort. In agreement with its loss-of-function effects, miR-124a overexpression significantly suppressed brain and extracranial metastases to lungs and ovaries (two-sided chi-square test, ****p < 0.0001; Figure 3D ), as measured by bioluminescence at experiment termination (brain metastasis, **p = 0.0043; extracranial metastasis, **p = 0.0087; Figure 3E ). Remarkably, while all mice injected with miRH-Scr-transduced cells developed brain metastases, none of the mice injected with miRH-124a-overexpressing cells developed any brain metastasis ( Figures 3D–F ). We conclude that miR-124a overexpression abolishes brain metastasis formation and significantly inhibits melanoma metastasis to other organs.

To determine if miR-124a was suppressing growth, rather than metastasis specially, we examined the effect of either constitutive or inducible miR-124a overexpression in relevant xenograft models. We subcloned the miR-124a precursor and its Scrambled control (Scr) into the TRIPZ™ inducible lentiviral vector (Dharmacon). This vector is configured for the expression of shRNAs and miRNAs of interest, together with the turboRFP (tRFP) reporter, in the presence of doxycycline (Tet-On configuration). Importantly, in this construct, miRNA expression and tRFP expression are coupled, as both arise from the same promoter, in a doxycycline-dependent manner.

Melanoma cells were transduced with a GFP-luciferase–expressing construct, followed by TRIPZ-Scr/miR-124a lentiviral particles, and were injected subcutaneously in the flanks of immune-compromised NOD/Scid/IL2γRnull (NSG) mice. Cells were treated with doxycycline prior to inoculation, and mice were fed doxycycline-containing food pellets immediately for constitutive miR124a overexpression ( Figure 4A ). Once palpable, tumor growth was regularly measured by caliper. In another set of experiments, mice were fed doxycycline-containing food pellets starting 14 days after inoculation, when tumors became palpable ( Figure 4D ). Comparison of tumor growth found that neither miR-124a constitutive ( Figures 4B, C ) nor inducible overexpression after tumors’ engraftment ( Figures 4E, F ) affected tumor growth. This indicates that miR-124a specifically suppresses melanoma metastatic progression without affecting primary tumor growth.

We postulated that miR-124a overexpression in already seeded metastatic cells might impair further growth. To test this hypothesis, we aimed to conditionally induce miR-124a in melanoma cells after reaching distal organs. Melanoma cells were transduced with GFP-Luciferase followed by TRIPZ-Scr or TRIPZ-miR-124a lentiviral infections. Efficient miRNA-124a overexpression after doxycycline treatment in vitro was assessed by qRT-PCR ( Figure 5A ). Subsequently, puromycin-selected cells cultured in tetracycline-free serum were inoculated in the left ventricle of NSG mice. We have found in previous studies that most melanoma cells have already extravasated from the vasculature and colonized the brain microenvironment by day 7 after intracardiac injection in mice (Kleffman et al., Biorxiv). We thus started feeding mice with doxycycline-containing chow at day 8 after cell inoculation, aiming for efficient miR-124a induction 2–3 days later. The experiment was terminated 36 days after inoculation, when some mice showed signs of discomfort ( Figure 5B ). Harvested organs (brain, liver, lung, and kidneys) were imaged using a dissecting scope to detect both GFP+ (which marks all melanoma cells) and RFP+ (which marks miR-Scr- or miR-124a-expressing cells) ( Figure 5B ). In the group injected with miR-Scr-transduced cells, most mice developed brain metastases, and the GFP and RFP signals vastly overlapped ( Figure 5C , upper brain panels). In the miR-124a-transduced group, however, RFP+ lesions were found absent from all harvested brains, with the exception of a small lesion in one mouse ( Figure 5C ; brain panels, eighth counting from left). In the liver and kidney, we found a similar suppression of RFP+ lesions, further supporting a negative selection of cells carrying miR-124a overexpression. These results strongly suggest that miR-124a expression is deleterious for metastases, and only those cells that escape miR-124a induction, either by insufficient doxycycline levels or by compensatory adaptations, are able to successfully colonize the brain. To prove this, we tested if RFP fluorescence was a faithful reporter of miR-124a overexpression. In contrast to the brain, several liver metastatic lesions were GFP+/RFP+ in the mice injected with miR-124a-transduced cells. We carefully dissected RFP-/GFP+ and RFP+/GFP+ liver metastatic foci from 2 different mice of the miR-124a-overexpressing group and assessed miR-124a expression by qRT-PCR. We found that RFP+/GFP+ lesions have significantly higher miR-124a expression than RFP-/GFP+ lesions from the same mice ( Figure 5D ). While miR-124a-overexpressing lesions were still present in liver and kidneys ( Figure 5C ), metastasis burden and organ size were reduced in mice injected with miR-124a-overexpressing cells than those instilled with miR-Scr-transduced cells ( Figure 5E ). Taken together, our data indicate that miR-124a acts as a potent barrier against metastatic growth, and that cells expressing miR-124a are negatively selected in general and even more so in the brain parenchyma, supporting its value as an anti-metastatic miRNA.

miRNA profiling of 92 human primary melanoma specimens (stages I and II, GSE62372) (20) revealed higher expression of miR-124a in non-recurrent (n = 44) vs. recurrent (n = 48) primary melanoma samples (p = 0.047; Figure 6A ). Moreover, primary tumors that eventually metastasized to the brain (n = 26) displayed lower miR-124a levels compared to those that did not (n = 66) (p = 0.02; Figure 6B ). Finally, low miR-124a levels associate with decreased brain metastasis-free survival (p = 0.05; Figure 6C ). These data further support the clinical relevance of miR-124a as a negative regulator of metastasis in general and of brain metastasis in particular.

Cell lines were grown at 37°C in an atmosphere of 5% CO₂. HEK293T cells purchased from American Type Culture Collection (ATCC) were maintained in Dulbecco's Modified Eagle's Medium (DMEM) medium (Invitrogen) containing 10% (v/v) FBS and 1% (v/v) penicillin/streptomycin. The 113/6-4L cells (gift of Dr. Robert Kerbel, University of Toronto) were cultured in DMEM (Invitrogen) containing 10% (v/v) FBS (Corning) and 1% (v/v) penicillin/streptomycin (Invitrogen). The 113/6-4L is a metastatic variant of the WM239A human melanoma cell line (26). Cells were maintained in culture for no more than 20 passages and were routinely tested for mycoplasma contamination (Universal Mycoplasma Detection kit, ATCC).

In this study, 3 × 10⁶ HEK293T cells were seeded per 10-cm tissue culture dish and co-transfected with lentiviral expression constructs (12 μg), viral packaging plasmid (pSPAX2, 8 μg), and viral envelope plasmid (pMD2.G, 4 μg) using Lipofectamine2000 (Invitrogen), following manufacturer’s recommendations. Viral supernatant was collected 48 h post-transfection, filtered through 0.45-μm filters, and stored at -80°C for long-term storage.

Target cells were seeded, incubated overnight prior to infection, and transduced at 30% of cell confluence. Medium was replaced with 1:4 diluted viral supernatant and 4 μg/ml Polybrene (EMD Millipore) and incubated for 6 h, followed by replacement with growth medium. Cells were checked for fluorescent protein expression, or drug selection agents were added on subsequent days to ensure pure populations of transduced cells. For viral transduction of melanoma cells with the Decoy library, cells were transduced at an MOI of 0.3 to achieve a single vector copy per cell, resulting in transduction of 30% of cells plated in 150-mm² dishes. The percentage of GFP-positive transduced cells and their viability were assessed by flow cytometry 3 days after transduction (BD LSR II, BD Biosciences).

RNA was extracted using the RNeasy mini kit (Qiagen) following manufacturer’s recommendations. Eluted RNA was quantified by Nanodrop 2000 or Qubit (Thermo Fisher) following manufacturer’s recommendations and stored at -80°C. When indicated, metastatic foci with RFP and/or GFP signal were dissected using a dissecting fluorescent scope (Leica). Samples were immediately frozen in TRIzol buffer prior to RNA extraction.

Reverse transcription (RT) was carried out using TaqMan MicroRNA Reverse Transcription Kit in the presence of RNase inhibitor (Applied Biosystems). Briefly, 25 ng of input RNA was reverse transcribed by stem-loop method following manufacturer’s recommendations. To determine the levels of mature miRNA-124a, quantitative Real time PCR (qPCR) was performed following cDNA generation. Briefly, 1 µl of cDNA was used as template in a 10-µl qPCR reaction by adding TaqMan Universal Master Mix II, no UNG (Applied Biosystems), and predesigned TaqMan MicroRNA Assays PCR primers and probes (FAM dye-labeled) for target miRNA-124a (ID: 001182). RNU44 small RNA was used as an endogenous reference gene (ID: 001094). All reactions were performed in triplicate using Biorad CFX 384 or ABI StepOne Plus real-time cyclers following manufacturer’s recommendations.

In this study, 1,000 ng of RNA was reverse transcribed using Applied Biosciences Taqman RT kit (Applied Biosystems, Thermo Fisher) following manufacturer’s recommendations. cDNA was diluted with nuclease-free H₂O prior to use in qPCR reactions. Glyceraldehyde 3-phosphate dehydrogenase (GAPDH) was used as endogenous reference gene in RT-qPCR experiments. Biorad CFX 384 or CFX96 real-time cyclers were used with the following 2-step cycling parameters: 10 min at 95°C, 40 cycles of 95°C for 15 s, followed by 60°C for 30 s, followed by melt curve analysis. Technical triplicates of PCR reactions were performed. Primers were purchased from IDT DNA with the following sequences: EZH2_Fw (5’-TGCTTCCTACATCGTAAGTGCAA-3’); EZH2_Rv (5’-GGTGAGAGCAGCAGCAAACT-3’); MAPK14_Fw (5’-GTGGCCACTAGGTGGTACAG-3’); MAPK14_Rv (5’-GGGGTGTTCCTGTCAGACG-3’); SPRY2_Fw (5’-CCGCGATCACGGAGTTCA-3’), and SPRY2_Rv (5’-GACATGTACCTGCTGGGTGAG-3’).

CMV-Luciferase-EF1α-copGFP (GFP-luc) lentiviral plasmid was purchased from BD Biosciences (BLIV511PA-1), and the lentiviral plasmid mCherry-Luc was a kind gift from Dr. Christian Badr (Massachusetts General Hospital). The miRH-Scr and miRH-124a encoding lentiviral vectors were purchased from SBI Biosystems. All plasmid constructs were propagated in Stbl3 (Thermo Fisher Scientific) or XL-1 Blue Ultracompetent bacteria (Agilent Technologies) on LB plates or in LB media with appropriate antibiotics. Plasmids were extracted by mini- or maxi-prep (Qiagen) following manufacturer’s recommendations. All cloned constructs were verified by Sanger sequencing prior to use.

Primary human melanoma samples were collected at the time of surgery from 92 patients. Informed consent was obtained from all patients, and approval was acquired from the institutional review board (IACUC) of NYU School of Medicine (protocol #10362). miRNA expression profiling of FFPE-extracted RNA from primary melanomas was performed by Exiqon, Inc., using dual-color miRCURY™ LNA arrays. The quantified signals were background corrected and normalized using the global Lowess regression algorithm. Data are deposited in GEO (accession: GSE62372) (20).

Statistical analyses were performed with GraphPad Prism (GraphPad Software, Inc.) Data are presented as the mean ± SD. Significance was determined using unpaired Student’s t test, chi square test, Mantel–Cox test, or Breslow–Wilcoxon test, where appropriate. The statistical analyses were performed, and p values were indicated in each figure legend. p values are represented as *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, and *****p < 0.00001.