Cell lines and clinical samples
Experimental studies described in this manuscript used the human epidermal melanocyte cell line HEM-l (ScienCell, Catalog # 2200, grown in MelM media containing MelGS growth supplements, 0.5% fetal bovine serum (FBS), penicillin and streptomycin), human epidermal keratinocytes (HEK, ScienCell, Catalog # 2100, grown in Keratinocyte Medium, ScienCell, Catalog # 2101), and the melanoma cell lines WM793B (stage 1, Wistar Institute), WM278 (stage 2, Wistar Institute), WM1552C (stage 3, American Type Culture Collection Number: CRL-2808), and A375 (stage 4, American Type Culture Collection). Melanoma cells were grown in Complete Tu Medium containing a 4
1 mixture of MCDB-153 medium with 1.5 g/L sodium bicarbonate and Leibovitz's L-15 medium with 2 mM L-glutamine, 2% FBS, and 1.68 mM CaCl2
All clinical samples were graciously donated by Dr. James Goydos, Robert Wood Johnson Medical School.
Isolation and bisulfite treatment of genomic DNA
Genomic DNA prepared from 107 of each cell line. Cells were harvested by trypsinization, washed once in phosphate-buffered saline (PBS), and purified using the QiaAmp DNA mini kit (QIAGEN). DNA from 25 mg of patient samples was isolated by overnight incubation with proteinase K at 55°C, with subsequent purification using the QiaAmp DNA mini kit (QIAGEN). All samples were quantified using the ND-1000 spectrophotometer (Nanodrop). DNA (0.5 µg) was treated with sodium bisulfite using the EZ DNA methylation kit (Zymo Research) and eluted in 10 µL elution buffer.
Sequencing of PCR products from bisulfite-converted genomic DNA for detection of CpG island methylation
Bisulfite-treated genomic eluate (2 µL) was used for bisulfite PCR using the following primer combinations: let7i f1 (GGGGGTAGTTTAGAATTAGTTGGTGTTTG) and let7i r1 (CCCCTTCTTTTCCTTTACCTTCCC) to produce a 301-bp product, 124a No-C-For (GGAAAGGGGAGAAGTGTGGG) and 124-3 Rev (CACCGCGTACCTTAATTATATAAAC) to produce a 260-bp product, & miR34b f1 (GAATTTGGGTTTTTATTTTTTAGG) and miR34b r1 (CCAAACCCTAAAACTAACTCTCTC) to produce a 236-bp product. PCR was performed using a 6-min hot start at 95°C, followed by 35 cycles at 94°C for 20 s, 54°C for 25 s, and 72°C for 30 s, ending with a 10-min extension at 72°C using AmpliTaq Gold (Applied Biosystems/Life Technologies). PCR products were gel purified using the QiaQuick gel extraction kit (QIAGEN) and cloned into the pCR4-TOPO vector (Invitrogen/Life Technologies). Nine clones for each miRNA candidate and 6 clones for each patient sample were sequenced using M13 primers and the BigDye terminator kit v1.1 (Applied Biosystems/Life Technologies), analyzed on a 3130×l Genetic Analyzer (Applied Biosystems/Life Technologies), and aligned using VectorNTi AlignX (Invitrogen/Life Technologies).
Treatment of WM1552C cells with 5-aza-2′-deoxycytidine (5-Aza-dC)
WM1552C cells (5×105) were plated into 75-cm2 flasks. Each flask was treated with 1, 2.5, 5, 7.5, or 10 µg/mL 5-Aza-dC or left untreated. Each day for 5 days, the cells were washed once with PBS, fed fresh medium, and treated as above. After day 5, the cells were trypsinized, washed once with PBS, and centrifuged at 1200 rpm for 5 min. Cell pellets were prepared for total RNA using the Trizol protocol (Invitrogen/Life Technologies), and RNA was quantified using the ND-1000 spectrophotometer (NanoDrop). The assay was performed in duplicate.
TaqMan Low Density Arrays
For cDNA synthesis and real-time PCR by Taqman Low Density Array (TLDA), total RNA (800 ng) was subjected to 8 separate reverse transcription reactions (100 ng each) using Multiplex RT for TaqMan® MicroRNA Assays, Human Pool Set kit (Applied Biosystems/Life Technologies). The resulting cDNA (10 µL) was diluted 1
62.5 with nuclease-free water. TLDAs consisting of a panel of 365 human miRNAs and 3 miRNA endogenous controls were run in triplicate for each sample. Diluted cDNA (50 µL) was added to 50 µL of 2× TaqMan Universal Master Mix (No AmpErase UNG; Applied Biosystems/Life Technologies). This 100 µL mixture was applied to the respective array port and the TLDA was then centrifuged twice and sealed. Quantitative real-time PCR was performed using the Applied Biosystems 7900 Real-Time PCR Sequence Detection System with the following thermal cycling parameters: 94.5°C for 30 s followed by 40 cycles of 97°C for 30 s, 59.7°C for 1 min.
Quantitative Real-time PCR
Total RNA was isolated by the Trizol method (Invitrogen/Life Technologies) with subsequent quantification and integrity analysis performed using an Agilent 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA, USA). Total RNA (100 ng) was reverse transcribed using a High Capacity cDNA kit (Applied Biosystems/Life Technologies), and quantitative reverse-transcription PCR was carried out using TaqMan miRNA or mRNA Assays or SYBR Green mRNA Assays and a 7500 Real-Time PCR System (Applied Biosytems/Life Technologies) in accordance with the manufacturer's protocols. SYBR Green primers include CDC42 (CDC42 qPCR For – ctgcacctacccacatgcactcgt and CDC42 qPCR Rev - ttaactagtactgggagggggaaggg), FN1 (FN1 qPCR For – ggctgacagagaagattcccgagag and FN1 qPCR Rev - ccagtttagatggatcttggcagagagac), and THBS2 (THBS2 qPCR For – ttaccgcttcgtgcgctttgac and THBS2 qPCR Rev - aacagcgtgcccctggacttg). SDS1.2.3 software (Applied Biosystems/Life Technologies) was used for comparative Ct analysis, with RNU48, GAPDH, or β-actin serving as the endogenous controls.
Northern blot analysis
Total RNA concentrated from each sample (20 ng from cell lines or 5-Aza-dC-treated melanoma cells as above), was analyzed by northern blot. Samples were separated in 15% TBE-urea polyacrylamide gels by electrophoresis and the RNA was electroblotted onto nylon membranes, cross-linked by ultraviolet light, prehybridized in Ultrahyb-Oligo (Ambion) for 30 min at 42°C, and hybridized with 5′-biotinylated anti-miRNA DNA oligonucleotides (100 nM each) at 42°C overnight. The blots were then washed, and the signal was detected by chemiluminiscence (Brightstar Detection kit, Ambion). Anti-U6 probes (10 pM) were used as a reference control.
Construction of a melanoma cell lines stably expressing miR-34b
Oligonucleotides complimentary to the hsa-miR-34b genomic sequences were constructed (miR-34b pre For – gtgctcggtttgtaggcagtg and miR-34b pre Rev – gtgccttgttttgatggcagtg), containing HindIII and BamHI sites on their respective 5′ and 3′ ends, then amplified from melanocyte genomic DNA (Amplitaq Gold, Applied Biosystems/Life Technologies). The product was then TOPO cloned into pCR4-TOPO (Invitrogen/Life Technologies). The vector construct was sequenced and the pre-hsa-miR-34b fragment was sub-cloned into pcDNA4/myc-HisA (Invitrogen/Life Technologies) using the HindIII and BamHI sites to create pcDNA4/miR-34b. WM1552C and A375 melanoma cells (2.5×105) were seeded into single wells of a 6-well plate and transfected with 5 µg pcDNA4/myc-HisA (Vector Control) or pcDNA4/miR-34b using Fugene 6 (Roche). The following morning, cells were selected with 600 µg/mL Zeocin for the following 15 days. The remaining stable cells were then expanded and named WM1552C/34b and WM1552C/VO (vector only) or A375/34b and A375/VO (vector only).
Next-generation DNA sequencing and RNA-seq
Total RNA was isolated from samples using Trizol (Invitrogen/Life Technologies) and fragmented through RNase III digestion. SOLiD sequence protocols (Applied Biosytems/Life Technologies) require reverse transcription of these RNAs, priming with a ligated primer, and the resulting cDNA amplified and size-selected in a 6% urea gel with the help of SYBR Gold dye for sequencing. The sequence libraries (150–200 bp size fragments) were further amplified using a bead-based emulsion PCR optimized to physically isolate a single bead/cDNA molecule. This enables massively parallel amplification of monoclonal DNA species. For RNA-seq experiments, we deposited approximately 90 million beads per sample onto a glass slide and analyzed them using a ligation-based sequencing technology. Data were mapped to the human genome using both BioScope and BFAST software packages and analyzed at the Burnham Institute Bioinformatics core facility.
Methylated DNA enrichment and library construction
Genomic DNA (melanocytes, WM1552C and WM1552C +5-Aza-dC) was fragmented to 50–400 bp (mean ~250 bp) using a Covaris™ S2 System (Woburn, MA), and 10 µg was subjected to MBD-protein capture with the MethylMiner™ Methylated DNA Enrichment Kit (Life Technologies) following the recommended protocol. The methylated DNA was resuspended in 40 µL GibcoR UltraPure™ DNase/RNase-Free Distilled Water (Life Technologies) and quantified by UV absorbance spectroscopy. For single-fraction elution, buffer containing 2 M NaCl was used to elute methylated DNA captured using the MethylMiner™ kit. The methylated DNA fragments recovered were ethanol precipitated and resuspended. This DNA, and a sample of DNA that did not undergo enrichment with the MethylMiner™ kit (whole-genome, unenriched), were used for SOLiD™ System fragment library construction, which includes a gel-based size selection step to obtain a mean insert length of ~150 bp.
RNA-Seq gene analysis
Mapping of the SOLiD sequence data was performed using the standard pipeline of Blat-like Fast Accurate Search Tool (BFAST) 
against the human genome (hg19) 
. The BFAST algorithm first creates indices of the human genome using different masks (in this work, all 10 masks suggested in the BFAST manual were used). It then hashes the reads to a number of genomic locations based on the indices and performs detailed alignments between the reads and the genomic sequence at the hashed locations. The final output of BFAST contains the mapped genomic locations of each read and the qualities (represented via alignment scores) of the mappings.
The expression level of a gene can be calculated based on the mapping results. The BFAST algorithm successfully mapped ~63% of the reads to the reference genome; among the mapped reads, ~89% were uniquely mapped. Only uniquely mapped reads were considered to eliminate possible noise. For each gene (from the RefSeq annotation for hg19), a score was calculated as reads per kilobase of sequence per million reads (RPKM) 
The predefined mapping protocol for SOLiD sequencing on BioScope 1.0 was also used. BioScope was able to map ~63% of the reads to the reference genome, ~83% uniquely. The output was then fed into the Partek Genomics Suite for analysis. Again, an RPKM score was computed for each gene.
The results of the two analysis pipelines were compared to arrive at a consensus. Fold-change was computed for each gene, this was a ratio of the WM1552C/34b RPKM over the WM1552C/VO RPKM. For this calculation, a pseudo-count of 1.0 was used to overcome divide by zero errors and to normalize the data at low RPKM values. A list of genes was produced such that both programs agreed on the regulation direction (i.e. both upregulated or downregulated). The order of the genes was then directed by the corroborated fold change. For the open reading frames, a cutoff was established with a minimum fold change of +/−1.5 (1.5 for miRNAs, −1.2 for target genes). The transcript delta was also obtained; this was used to select genes for further testing, this cutoff was set at 10.0 for open reading frames (1.3 for miRNAs, −2.0 for target genes). The complete gene list is available in the supplemental data.
Of particular interest are the target genes of miR-34b. This target gene list was established from TargetScan 5.1 
utilizing the top 500 miR-34b targets, irrespective of conservation. The data acquired from deep sequencing was then filtered through the target list to arrive at a final putative target list.
Cell adhesion assays
Trypsinized cells were counted, and 250,000 cells were seeded per well into 12-well plates. At 5, 15, 30, 45, and 60 minutes after seeding, floating cells were aspirated by rinsing the wells with PBS. The remaining cells in the wells were trypsinized, resuspended in cell media, and individually counted using the Countess® Cell Counter (Invitrogen/Life technologies). Each sample was assayed in triplicate at each time point, and each experiment was repeated twice.
BD BioCoat™ growth factor reduced insert plates (Matrigel™ Invasion Chamber 12 well plates) were prepared by rehydrating the BD Matrigel™ matrix coating in the inserts with 0.5 ml of culture medium for 2 hrs at 37°C. The rehydration solution was carefully removed from the inserts, 0.75 mL Complete Tu medium containing chemoattractant (1% FBS) was added to the lower wells of the plate, and 0.5 mL of cell suspension (1×104 cells, in serum-free medium) was added to each insert well. Invasion assay plates were incubated for 48 hrs at 37°C. Following incubation, the non-invading cells were removed by scrubbing the upper surface of the insert. The cells on the lower surface of the insert were stained with crystal violet and each trans-well membrane mounted on a microscopic slide for visualization and analysis. The slides were scanned using the Scanscope digital slide scanner, and the number of cells migrating was counted using Aperio software. Data are expressed as the percent invasion through the membrane, relative to the migration through the control membrane:
Mean number of cells invading through the Matrigel insert membrane/Mean number of WM1552C (wild-type) cells migrating through membrane.
In vitro wound healing assay
WM1552C/34b and WM1552C/VO cells or A375/VO and A375/34b cells were seeded on Mat Tek 1.5 mm tissue culture dishes until 90–95% confluent. Cell monolayers were then gently scratched with a pipette tip across the entire diameter of the dish and extensively rinsed with medium to remove all cellular debris. The surface area of the denuded surface was quantified immediately after wounding and again every 20 minutes for 24 hrs on the Nikon Bio Station IM. The extent of wound closure was determined by calculating the ratio of the surface area between the remaining wound edges for each time point to the surface area of the initial wound. These data were then expressed as the percentage of wound closure relative to the control conditions for each experiment. The surface area was calculated using NIS Elements software and performed in triplicate.