The 3′UTRs of K_ir2 genes were analysed for predicted miRNA target sites by TargetScan [²³], MiRanda [²⁴,²⁵] and PicTar [²⁶].

The pmR-mCherry plasmid (Clontech) was modified by removal of 774 bp downstream of the BamHI site in the multiple cloning site following the mCherry stop codon. First, a fragment incorporating bases 2168–2670 of pmR-mCherry was amplified using Phusion® Hot Start II High-Fidelity DNA Polymerase (Fermentas) with the forward primer, 5′-ATATATGGATCCTATGTATCCGCTCATGAGACAATAACCCTG-3′ and the reverse primer, 5′-CCCAAGCGGCCGGAGAAC-3′, and restricted with BamHI and EagI (Fermentas). This was then ligated into the gel-purified 3468 bp fragment generated by restriction of pmR-mCherry with BamHI and EagI. The resulting plasmid was restricted with XhoI (Fermentas) and BamHI and ligated to the 3.9 kb human K_ir2.1 3′UTR sequence (GenBank® accession number NM_000891) amplified from human genomic DNA (Bioline) by a two-stage nested PCR using Phusion® Hot Start II High-Fidelity DNA Polymerase and the forward primer (outer), 5′-TGCCCTCACAAGCAAAGAGG-3′; the reverse primer (outer), 5′-CTGCACCTGTCTGTTGAGGC-3′; the forward primer (inner), 5′-AGGTCTCTCGAGCCCTGACATAGACCTTCACAAC-3′; and the reverse primer (inner) 5′-GACAGAGGATCCCATTTCTATGGCTCTATGC-3′, restricted with XhoI and BamHI. The final product (pmChKir2.1UTR) has the CMV (cytomegalovirus) promoter driving transcription of mCherry with the human K_ir2.1 3′UTR (Figure 1A). The second stage of the nested PCR was performed with the forward primer, 5′-ATGCCTACTAGTCCCTGACATAGACCTTCACAAC-3′ and the reverse primer, 5′-GACAGAGTCGACCATTTCTATGGCTCTATGC-3′. The product was restricted with SpeI and SalI (Fermentas) and ligated into the NheI (Fermentas) and SalI sites of pmirGLO (Promega), downstream of the firefly luciferase coding sequence, to give pmirGLOKir2.1UTR.

The plasmid pSM30 [²] (encoding EGFP) was used for expression of miRNA and siRNA (short interfering RNA). Complementary pairs of oligonucleotides (25 μM) encoding artificial miRNA and siRNA sequences or a non-targeting negative control sequence (SCR, Invitrogen) [²⁷] were annealed and ligated (T4 DNA ligase, Fermentas) into pSM30 restricted with BsmBI (New England Biolabs). The oligonucleotide sequences used were (mature miRNA/siRNA sequence underlined): hsa-miR-1 (top), 5′-AGCGCTACATACTTCTTTACATTCCATAGTGAAGCCACAGATGTATGGAATGTAAAGAAGTATGTAT-3′; hsa-miR-1 (bottom), 5′-GGCAATACATACTTCTTTACATTCCATACATCTGTGGCTTCACTATGGAATGTAAAGAAGTATGTAG-3′; hsa-miR-212 (top), 5′-AGCGAGCGTGACTGGAGACTGTTATAGTGAAGCCACAGATGTATAACAGTCTCCAGTCACGGCC-3′; hsa-miR-212 (bottom), 5′-GGCAGGCCGTGACTGGAGACTGTTATACATCTGTGGC TTCACTATAACAGTCTCCAGTCACGGCT-3′; SCR (top), 5′-AGCGATCTCCACGCGCAGTACATTTCTAGTGAAGCCA CAGATGTAGAAATGTACTGCGCGTGGAGAC-3′; SCR (bottom), 5′-GGCAGTCTCCACGCGCAGTACATTTCTACATCTGTGGCTTCACTAGAAATGTACTGCGCGTGGAGAT-3′; KCNJ2si (top), 5′-AGCGCAACGAGACTCTTGTCTTCAGTCTAGTGAAGCCACAGATGTAGACTGAAGACAAGAGTCTCGTTT-3′; and KCNJ2si (bottom), 5′-GGCAAAACGAGACTCTTGTCTTCAGTCTACATCTGTGGCTTCACTAGACTGAAGACAAGAGTCTCGTTG-3′.

The KCNJ2si sequence was based on the siRNA reported by Rinne et al. [²⁸]. Following confirmation of inserts by DNA sequencing (Genomics Core, Queen's University Belfast, Belfast, U.K.), plasmids were prepared with a Maxiprep kit (Qiagen) following the manufacturer's instructions. Custom oligonucleotides were purchased from Eurogentech.

An adaptation of the SLIM (site-directed ligase-independent mutagenesis) method [²⁹] was used for mutation of the seed region of the putative miR-212 target site in the human K_ir2.1 3′UTR. The pmChKir2.1UTR plasmid (prepared from dam⁺Escherichia coli) was used as the template in two separate PCRs with Platinum Pfx DNA polymerase (Life Technologies) and 0.5× enhancer buffer. The primers in reaction A were: sense, 5′-GCATGTCAGTGGCATTTGAAACCTTTTTCTAGTTATTAGCGTTT-3′ (mutated bases underlined) and antisense, 5′-TGAAATAAATAACAGAAGAACATAGC-3′. The primers in reaction B were: sense, 5′-AAACCTTTTTCTAGTTATTAGCGTTT-3′ and antisense, 5′-CAAATGCCACTGACATGCTGAAATAAATAACAGAAGAACATAGC-3′ (mutated bases underlined). The methylated template was digested with DpnI (New England Biolabs) to leave the unmethylated mutagenesis product. Reaction products A and B were mixed and re-annealed, then used for transformation of XL-2 Blue ultracompetent E. coli (Agilent). The plasmid preparations (pmChKir2.1UTRm212) were sequenced to confirm the presence of the mutated site (Genomics Core, Queen's University Belfast).

RNA was extracted using the miRNeasy kit (Qiagen) according to the manufacturers’ protocols. RNA (1 μg) was polyadenylated using PAP [poly(A) polymerase; Ambion] at 37°C for 1 h in a 25 μl reaction mixture. RNA was reverse transcribed with 200 units of SuperScript III reverse transcriptase (Life Technologies) and 0.5 μg of poly(T) adapter (5′-GCGAGCACAGAATTAATACGACTCACTATAGGTTTTTTTTTTTTVN-3′). The sequence for the miR-212 primer was 5′-ATAACAGTCTCCAGTCACGGCC-3′. The reverse primer was the 3′ adapter primer: 5′-GCGAGCCACAATTAATACGAC-3′ [3′ RACE (rapid amplification of cDNA ends) outer primer in the FirstChoice RLM-RACE kit, Ambion]. qRT–PCR was performed using Maxima SYBR Green qPCR mastermix (Fermentas) in 10 μl reaction mixtures containing 2 μl of 1:15 cDNA dilution. Reactions were performed on a LightCycler 480 (Roche), with the initial pre-incubation at 50°C for 2 min and activation at 95°C for 10 min, followed by 40 cycles at 95°C for 15 s and 60°C for 60 s, with fluorescence acquired after 15 s of the 60°C step. Gene expression data were normalized to 5S RNA. The relative expression was determined as 2^−ΔC_t, where ΔC_t=C_t(miRNA)−C_t(5S).

pmChKir2.1UTR was linearized with PasI (Fermentas) (Figure 1) and transfected into HEK (human embryonic kidney)-293 cells using FuGene HD (Roche Diagnostics) in a T25 flask according to the manufacturers’ protocol. After 2 days the cells were detached with 0.25% trypsin/0.9 mM EDTA (Invitrogen), diluted 10-fold into complete medium [DMEM (Dulbecco's modified Eagle's medium), 10% fetal bovine serum and 1% Primocin (InvivoGen)] containing 0.3 mg/ml of G418 (Sigma–Aldrich) and re-plated in ten separate aliquots. Red fluorescence was monitored from this time onwards. Cells were maintained for 14 days with replacement of medium every 3–6 days. The G418 concentration was then changed in pairs of cultures to 0.3, 0.4, 0.5, 0.6 and 0.7 mg/ml to determine the optimum selection pressure. At 2–5 days later the cultures were selected for re-plating at single-cell density into 24- or 96-well plates. Wells containing single colonies were selected for propagation after a further 12 days. The mChKir2.1UTR clone used in the present study (clone A4) was selected in 0.7 mg/ml of G418 and subsequently maintained in 0.5 mg/ml of G418.

Genomic incorporation of the intact mCherry-K_ir2.1 3′UTR sequence was confirmed by PCR analysis. Genomic DNA was extracted with DirectPCR Lysis reagent (Viagen Biotech) according to the vendor's protocol. The forward primer was in the mCherry coding sequence (5′-ACCTCCCACAACGAGGAC-3′) and the reverse primer was the inner reverse primer used for subcloning as described above (4055 bp product).

HEK-293 cells and mChKir2.1UTR cells were plated on to 8-mm diameter cover slips in a 24-well plate or directly into a 24-well plate. The next day the mChKir2.1UTR cells were transfected with pSM30 containing an insert encoding miR-212, miR-1 or the non-targeting negative control sequence (pSM30-miR-212, pSM30-miR-1 and pSM30-SCR respectively), or MB-IRK3PGEMHE (non-fluorescent negative control plasmid) using JetPEI™ transfection reagent (Polyplus-transfection SA) or FuGene HD transfection reagent (Roche Diagnostics). Cells to be transiently transfected were transfected with either pmChKir2.1UTR or pmChKir2.1UTRm212, in addition to one of pSM30-miR-212, pSM30-miR-1, pSM30-SCR or MB-IRK3PGEMHE using FuGene HD. At 2 days later the coverslips were transferred on to microscope slides for fluorescence microscopy at ×20 magnification using a Nikon Eclipse 80i microscope equipped with G-2A (red) and B-2A (green) filter sets. Cells plated directly into 24-well plates were imaged at ×20 magnification using a Nikon Eclipse TE2000-U microscope equipped with a 488 nm filter (red; mCherry fluorescence) and a 550 nm filter (green; EGFP fluorescence). Red and green images were acquired of multiple fields using NIS Elements software (Nikon). Images in.tif format were analysed by Volocity software (Volocity 5.5.1, PerkinElmer).

We developed a Volocity pipeline for identifying cells in green images and measuring the red and green intensities of each identified cell. The Volocity pipeline identified green cells by intensity. The sum of pixel intensities of each identified cell was then measured. The outlines of the cells were transposed on to the red channel image of the same field and the sum red intensity of each cell was measured. Background fluorescence and cross-channel bleed-through were corrected using images of HEK-293 and mChKir2.1UTR cells transfected with MB-IRK3PGEMHE, HEK-293 cells transfected with pmChKir2.1UTR or pmChKir2.1UTRm212 and MB-IRK3PGEMHE, and non-transfected mChKir2.1UTR cells. Data were exported in Microsoft Excel format for further analysis (Prism 4, GraphPad).

HEK-293 cells were seeded in 96-well plates and co-transfected with two plasmids: (i) one of pSM30-SCR or pSM30-miR-212; and (ii) pmirGLOKir2.1UTR, using FuGene HD. At 16–24 h post-transfection, EGFP expression was assessed with a fluorescent light source (Zeiss filter set 09, 450–490 nm excitation, 510 nm beam splitter, 515 nm emission). Dual-Glo® Luciferase assays (Promega) were carried out according to the manufacturer's instructions. Luminescence measurements were made on a plate-reading luminometer (Tristar LB 941, Berthold Technologies).

Total mRNA was extracted from HeLa cells grown in a T25 flask (Sigma–Aldrich) with RNeasy Plus Mini kit (Qiagen) and cDNA was generated with random hexamer primers by reverse transcription (Superscript III, Life Technologies) according to the manufacturer's recommendations. The resulting cDNA was subjected to qualitative PCR, which was conducted in a Veriti thermal cycler (Applied Biosystems), and was performed with DreamTaq™ Green DNA Polymerase (Fermentas). Each PCR amplification reaction (50 μl volume) contained 2 μl of HeLa cDNA or 200 ng of human heart genomic DNA (BioChain Institute), 1.25 units of polymerase, 5 μl of dNTP (2 mM each) 5 μl of DreamTaq™ Green buffer and 5 μM of primers. The primer sequences (and predicted fragment sizes) were: K_ir2.1 forward, 5′-CGACGTCTATCCTTTCCTGC-3′; K_ir2.1 reverse, 5′-TGTGGCATCTTCCCTAAGGT-3′ (278 bp); K_ir2.2 forward, 5′-AACCCCTACAGCATCGTGTC-3′; K_ir2.2 reverse, 5′-GCGATGACCCAGAAGATGAT-3′ (296 bp); 18S (ribosomal RNA) forward, 5′-CCTGCGGCTTAATTTGACTC-3′; and 18S reverse, 5′-ACCAACTAAGAACGGCCATG-3′ (121 bp). The cycling conditions consisted of an initial denaturizing step at 95°C (60 s) followed by 35 cycles of a thermal step protocol consisting of 95°C (45 s), 60°C (20 s) and 72°C (60 s) and a final step of 72°C (60 s). A negative control for cDNA was performed without reverse transcription (−RT control). An aliquot (10 μl) of the RT–PCRs were analysed by electrophoresis on a 1.5% agarose and 0.3 μg/ml ethidium bromide gel in TBE (Tris/borate/EDTA).

Quantitative PCR was carried out using TaqMan® Gene Expression assays (Life Technologies): Hs01003267_m1 [HPRT1 (hypoxanthine–phosphoribosyltransferase 1); Hs99999905_m1 [GAPDH (glyceraldehyde-3-phosphate dehydrogenase)]; and Hs00265315_m1 (KCNJ2), following reverse transcription with a high-capacity kit (catalogue number 4374966, Life Technologies).

Following transfection, the plasma membrane was isolated from HeLa cells for Western blotting. Cells were scraped in PBS and pelleted at 1000 g for 5 min at 4°C. The cell pellet was resuspended in 20 mM Mops, 250 mM sucrose, 1 mM PMSF, 0.1 mM EDTA, 50 mM sodium fluoride and 2 mM sodium vanadate and snap frozen with liquid nitrogen to break the cell membranes. The samples were then homogenized with a Biospec Tissue Tearor for 2×20 s then sonicated for 2×15 s. Samples were centrifuged at 200 g for 5 min at 4°C to pellet unbroken cells and nuclei, and the supernatant was ultracentrifuged at 40000 rev./min for 1 h at 4°C in a Beckman TL 100 rotor to pellet the membrane fraction. The membrane pellet was then resuspended in NuPAGE LDS buffer (Invitrogen), sonicated for 30 s and boiled for 5 min. Membrane samples were resolved by SDS/PAGE (12% gel) and transferred by electrophoresis on to PVDF membranes. The membrane-bound protein was probed with anti-K_ir2.1 antibody (1:500 dilution; Abcam) followed by anti-(mouse horseradish peroxidase-conjugated) secondary antibody (1:1000 dilution; Santa Cruz Biotechnology). Equivalent loading of protein was checked by probing with an antibody against Na/K-ATPase α1 (1:500 dilution; Santa Cruz Biotechnology). Blots were imaged on a UVP Autochemi system.

Inward-rectifier K⁺ current was measured by patch-clamp recording. HeLa cells were seeded on to coverslips 24 h after transfection with pSM30-miR-212, pSM30-SCR or the pSM30-based siRNA plasmid (pSM30-KCNJ2si) and allowed to grow in complete DMEM for a further 18–24 h. Coverslips were then transferred to a 2 ml recording chamber mounted on the stage of an inverted microscope (Axiovert 100, Zeiss) equipped with a fluorescent light source. Cells were superfused with normal Tyrode solution (in mM: 140 NaCl, 4 KCl, 2.5 CaCl₂, 1 MgCl₂, 10 glucose and 10 Hepes, pH 7.4 with NaOH) at 22°C. Whole-cell currents were recorded from fluorescent cells in voltage-clamp mode using an Axopatch 200B amplifier (Molecular Devices) grounded to the recording chamber via a 3 M KCl agar bridge. Electrodes (1–4 MΩ) were pulled from filamented borosilicate glass capillaries (1.5 mm outside diameter and 1.12 mm inside diameter; World Precision Instruments). The internal pipette solution was K⁺-based [in mM: 65 KCl, 40 KF, 5 EGTA, 10 glucose, 0.24 NaVO₃, 10 Hepes, 5 tetrasodium pyrophosphate and 15 KOH (pH 7.4)]. Reversal potentials and normalized current densities were corrected for the pipette junction potential (7.8 mV) offline. Access resistance was typically 2–8 MΩ and cell capacitances were 39±4 pF as assessed by the membrane-test facility of Clampex 10.2 (Molecular Devices). Both access resistance and cell capacitance were compensated using the amplifier (~80%); currents were filtered at 10 KHz and normalized to capacitance for comparison of the current density.

Prior to recording membrane current using an ascending voltage ramp protocol (−100 to +50 mV, 60 mV/s; applied every 5 s after an initial 200 ms step to −100 mV to allow for inactivation of any hyperpolarization-activated or uncompensated capacitative current) the external solution was switched via a rapid-solution device positioned adjacent to the cell (ALA Scientific Instruments) to one containing 120 mM K⁺ [in mM: 115 KCl, 20 NaCl, 2.5 CaCl₂, 1 MgCl₂, 10 glucose, 10 Hepes and 5 KOH (pH 7.4)] resulting in a predicted E_K of −3.9 mV. I_K1 (inward-rectifier K⁺ current) was isolated as the current sensitive to application of 100 μM Ba²⁺ in the external solution.

Statistical analysis was performed with Prism 4.0 software (GraphPad). Student's t test was used for comparing two groups. ANOVA was used for comparing more than two groups.

MiR-212 was predicted to target the 3′UTR of human K_ir2.1 by TargetScan [²³] and MiRanda [²⁴]. Predicted base-pairing between miR-212 and the target sequence is shown in Figure 1(B).

The ability of the pSM30 system to direct expression of a mature miRNA was confirmed by qRT–PCR as shown in Figure 2. Transfection with pmiR-212 increased expression of mature miR-212 in HEK-293 (Figure 2A) and in HeLa (Figure 2B) cells.

If a microRNA co-expressed with EGFP targets the 3′UTR cloned downstream of the red fluorescent protein coding sequence (mCherry) we would expect to observe a decrease in red fluorescence and a concomitant reduction in the ratio of red/green fluorescence. HEK-293 cells were therefore co-transfected with pmChKir2.1UTR and a pSM30-based plasmid expressing either miR-1 or a non-targeting miRNA control (SCR). The total green and red intensities of multiple cells were measured using fluorescent microscopy and the values for each cell plotted against each other (Figure 3A). Comparison of the lines of best fit clearly illustrates the expected reduction in red fluorescence in those cells overexpressing miR-1 compared with the non-targeting miRNA. Similarly, the relative frequency histograms of log red/green fluorescence ratios for cells transfected with pmChKir2.1UTR/pSM30-miR-1 are shifted to the left-hand side relative to those for cells transfected with pmChKir2.1UTR/pSM30-SCR (Figure 3B). A plot of red compared with green intensity for cells transfected with pmChKir2.1UTR/pSM30-miR-212 and pmChKir2.1UTR/pSM30-SCR is shown in Figure 3(C). Red/green fluorescence was significantly lower for pmChKir2.1UTR/pSM30-miR-1-transfected and pmChKir2.1UTR/pSM30-miR-212-transfected compared with the pmChKir2.1UTR/pSM30-SCR-transfected cells (Figure 3D).

To confirm that the reduction in red fluorescence observed following overexpression of miR-212 is mediated through the predicted target site, we tested whether this effect could be prevented by mutating the sequence of the site to disrupt binding to the miR-212 seed region. The percentage of red/green fluorescence ratio in pSM30-miR-212- compared with pSM30-SCR-transfected cells was significantly lower in cells carrying the wild-type (pmChKir2.1UTR) compared with the mutated (pmChKir2.1UTRm212) construct (Figure 3E), demonstrating that the mutation attenuated the sensitivity to miR-212.

The use of a reporter plasmid in addition to the miRNA expression vector has some potential drawbacks. For example variations in transfection efficiency may introduce noise that multiplies with the number of co-transfected constructs. A strategy that employs single transfections may abrogate this problem. We therefore generated a cell line stably transfected with the mCherry reporter linked with the K_ir2.1 3′UTR (mChKir2.1UTR cells). This was confirmed by PCR analysis of genomic DNA. Figure 4(A) shows an agarose ethidium bromide electrophoresis gel of PCR products from mChKir2.1UTR genomic DNA (lane 2), untransfected HEK-293 genomic DNA (lane 3), no-template control (lane 4) and pmChKir2.1UTR plasmid (lane 5). The forward primer site is in the mCherry coding sequence and the reverse primer site is at the end of the K_ir2.1 3′UTR sequence (Figure 1A). The bands in lanes 2 and 5 are consistent with the predicted product size of 4055 bp. These data confirm the presence of the K_ir2.1 3′UTR downstream of the mCherry coding sequence in the genomic DNA of mChKir2.1UTR cells.

Down-regulation of mCherry expression by miR-212 in mChKir2.1UTR cells is shown in Figure 4(B). Dual-fluorescence analysis was performed as described above. Red/green fluorescence was significantly lower for pSM30-miR-212-transfected compared with the pSM30-SCR-transfected cells. These data again indicate targeting of the K_ir2.1 3′UTR by miR-212.

The novel miR-212 result from the dual-fluorescence assay suggested that a similar result may be obtained with a more traditional luciferase-based assay. The effect of miR-212 on the expression of firefly luciferase with the K_ir2.1 3′UTR is shown in Figure 5. Firefly luciferase activity (normalized to Renilla luciferase) was less for pSM30-miR-212- than pSM30-SCR-transfected cells.

The results of the dual-fluorescence assay predicted that miR-212 would suppress the expression of endogenous K_ir2.1. We first confirmed the expression of K_ir2.1 mRNA in HeLa cells, as shown by the results of RT–PCR experiments shown in Figure 6. Primers were designed to amplify K_ir2.1, K_ir2.2 and 18S ribosomal RNA. Human genomic DNA was used as a positive control for these primers (lanes 2–4). PCR products from HeLa cDNA are shown in lanes 5–7. There was clearly a product for K_ir2.1 (lane 5) and a very faint product for K_ir2.2 (lane 6) despite a strong signal with the K_ir2.2 primers from the genomic DNA template (lane 3).

MicroRNAs have been reported to promote degradation of the target mRNA in some cases. We performed qPCR to determine whether miR-212 caused a reduction of K_ir2.1 mRNA in HeLa cells. As shown in Figure 7, miR-212 expression did not significantly change K_ir2.1 mRNA content relative to two different endogenous control genes, GAPDH (Figure 7A) and HPRT1 (Figure 7B). Despite the lack of effect at the mRNA level, miR-212 did down-regulate endogenous K_ir2.1 protein in HeLa cells. Figure 8(A) shows an example of a Western blot of membrane extracts from HeLa cells transfected with pSM30-miR-212 or pSM30-SCR, probed with antibodies against K_ir2.1 and Na⁺/K⁺-ATPase α1 subunit (loading control). Figure 8(B) summarizes the data from three such experiments and shows that miR-212 reduced K_ir2.1 protein.

We sought further evidence for targeting of K_ir2.1 by miR-212 by investigating the effect on inward-rectifier K⁺ channel activity itself in HeLa cells. Green fluorescent HeLa cells were selected 48 h post-transfection with pSM30-miR-212 or pSM30-SCR for measurement of inwardly-rectifying K⁺ current using the patch-clamp technique. Voltage ramps (−100 mV to +50 mV at 60 mV/s; Figure 9A) applied from a holding potential of −40 mV resulted in a current which was inward at negative voltages and reversed close to 0 mV (see Figure 9Bi; ctrl). Application of 100 μM Ba²⁺ (a relatively selective inhibitor of inward rectifier K⁺ channels [³⁰–³³]) inhibited portions of both the inward and outward whole-cell current with preferential inhibition of the inward portion (Figure 9Bi; Ba²⁺). On average the current at −108 mV (corrected for pipette junction potential) recorded in pSM30-SCR transfected cells was reduced from −219.02±47.61 pA/pF to −178.71±47.79 pA/pF by the application of 100 μM Ba²⁺ (Figure 9Ci; n=8; P<0.01 determined by paired Student's t test). As plotted in Figure 9(D) the Ba²⁺-sensitive current exhibited strong inward rectification and reversed at −3.55±2.61 mV. In cells transfected with pSM30-miR-212 the whole-cell inward current was smaller in amplitude to that in cells transfected with pSM30-SCR (Figure 9Bii). On average the current at −108 mV (corrected for pipette junction potential) recorded in pSM30-miR-212-transfected cells was reduced from −104.98±41.07 pA/pF to −100.91±41.10 pA/pF by the application of 100 μM Ba²⁺ (Figure 9Cii; n=13; P<0.01 determined by paired Student's t test). The Ba²⁺-sensitive current in cells transfected with pSM30-miR-212 was significantly smaller than the equivalent current in pSM30-SCR-transfected cells (Figure 9D; P<0.01 determined by ANOVA for the entire IV relationship). Transfection with pSM30-KCNJ2si also reduced the Ba²⁺-sensitive current [Figure 9E; P<0.01 determined by ANOVA for the entire IV (current–voltage) relationship].

Luciferase-based assays have been widely used for investigating miRNA–target interactions. In our hands, and anecdotally in the hands of others, they tend to be highly variable, necessitating time-consuming optimization of conditions. They also require cell lysis, leading to loss of information. Large sample sizes are impractical because each transfected well yields only one data point and the required reagents tend to be quite expensive. The dual-fluorescence assay yields a data point for each cell, of which there are typically several hundred in an experiment. A large sample size enables the detection of subtle effects of the miRNA of interest and comparisons of the relative efficacy of different miRNAs. It also reduces the reliance on assumptions when testing for statistical significance. Luciferase data are typically expressed as the ratio reporter/normalizer (e.g. luciferase/β-galactosidase or firefly/Renilla) and tested for significance using a parametric test such as Student's t test. Such tests rely on the assumption of a Gaussian distribution. The larger sample sizes achievable with our dual-fluorescence assay enable the construction of a frequency histogram (Figure 3B), allowing the direct assessment of the degree of approximation to a Gaussian distribution. Results of parametric tests can then be viewed with greater confidence.

Measurement of fluorescence is minimally invasive for the cell so the dual-fluorescence assay could be readily adapted for time-course studies. We anticipate that it could also be scaled up and adapted for flow cytometry, which would give a higher sensitivity associated with larger sample sizes. Further modifications such as more efficient methods for producing stably transfected cell lines [e.g. ViraPower™, Flp-In™ (Life Technologies)] and robotic sample handling may make the assay suitable for high-throughput screening of miRNA–target interactions.

The method described in the present paper requires an investment of time and effort in the construction of the miRNA expression plasmid, but once cloned the construct can be used indefinitely in future for assays against any sequences of interest. Part of our planned future work is to develop a catalogue of miRNA expression plasmids for this purpose. In its present form the method also involves an investment of labour in the acquisition of images. Throughput could be increased by the use of a high-content analysis system. The available software is able to analyse multiple images automatically so the image analysis is not onerous. The results of the present study were analysed with Volocity software. The method also works well with CellProfiler software [³⁴], which is available via free download.

We used the luciferase-based Dual-Glo® assay to validate the miR-212 result (Figure 5). We also used the Dual-Light® system (Applied Biosystems) with similar results (results not shown). These data were obtained with considerable time and resources expended on optimizing the conditions, which was in part the motivation for investigating an alternative assay. We anticipated that the variability of luciferase-based assays may be due to the requirement for transfection with multiple constructs simultaneously. However, we found the pmChKir2.1UTR/pSM30 transient co-transfection version of the dual-fluorescence assay to be surprisingly reliable. Nevertheless we developed the mChK_ir2.1UTR stably transfected cell line with a view to reducing any inter-experimental variability associated with double transfections. We again found down-regulation of mCherry-K_ir2.1 3′UTR by miR-212 using this single transfection approach (Figure 4B).

In this study we used the dual-fluorescence assay to confirm the known miR-1 target in the K_ir2.1 3′ UTR [³⁵] (Figures 3A, 3B and 3D) and to identify a novel target of miR-212 (Figures 3C and 3D). The inhibitory effect of miR-212 on mCherry-K_ir2.1 3′ UTR expression was significantly attenuated by mutation of the seed region of the predicted miR-212 target site (Figure 3E), supporting the notion that this is indeed a functional site.

pSM30 is a miRNA expression plasmid designed by Dr G. Du (University of Texas Health Science Center at Houston, Houston, TX, U.S.A.; [²]) into which an artificial miRNA sequence of choice can be inserted within an artificial intron of the EGFP gene. The advantages of this method are that miRNAs are efficiently transcribed and processed by the cell's own machinery, and expression can be monitored by the simultaneous expression of EGFP which (unlike conventional expression reporter vectors) is expressed only if the miRNA insert is correctly spliced. We have confirmed expression of the mature miRNA in pSM30-miR-212-transfected cells by quantitative PCR (Figure 2). The dual-fluorescence method takes advantage of the co-expression of EGFP and mature miRNA in transfected cells.

Having demonstrated an inhibitory effect of miR-212 on the K_ir2.1 3′UTR in reporter systems (Figures 3–5), we investigated the effect of miR-212 on the expression of endogenous K_ir2.1. The results of quantitative PCR experiments (Figure 7) indicated that miR-212 did not suppress K_ir2.1 mRNA, although a decrease in K_ir2.1 protein was seen (Figure 8). The effect of miR-212 on K_ir2.1 expression therefore appears to be mediated through inhibition of protein translation rather than mRNA degradation.

We have also shown that miR-212 expression inhibited inward-rectifier K⁺ channel function. Inward-rectifier K⁺ channels were previously identified in HeLa cells [³⁶] and later characterized as K_ir2.1 [³⁷]. We confirmed the presence of K_ir2.1 mRNA in HeLa cells by PCR (Figure 6). This stable endogenous expression in an easily transfected cell line made HeLa cells ideal candidates for investigation of the functional regulation of K_ir2.1 by miR-212. Transfection of HeLa cells with pSM30-miR-212 did indeed reduce I_K1 density, as demonstrated by whole-cell recording (Figure 9). This was most likely to be due to reduced expression of K_ir2.1 (Figure 8), although we cannot rule out a minor contribution from K_ir2.2 which was detected faintly by PCR of HeLa cDNA (Figure 6) and is predicted to be targeted by miR-212 (by TargetScan, MiRanda and PicTar). This would also be consistent with the apparently smaller effect of KCNJ2si on HeLa I_K1 density compared with miR-212 (Figures 9D and 9E). The possibility of miR-212 regulating both K_ir2.1 and K_ir2.2 is an important subject for future study because there is evidence that K_ir2.2 contributes significantly to the cardiac I_K1 [¹⁷,³⁸–⁴⁰].

The effect of miR-212 on I_K1 density was greater than the effect of miR-212 on red/green fluorescence (Figures 3 and 4) and endogenous K_ir2.1 protein (Figure 8). This may be due to the selection of highly expressing cells for patch-clamp, the over-expression of target protein in the dual-fluorescence assay and the background level of K_ir2.1 protein from untransfected cells. The difference in rectification of the control Ba²⁺-sensitive current shown in Figures 9(D) and 9(E) is probably due to a subtle difference in recording conditions. In Figure 9(D) Ba²⁺ was applied after run-up of outward current which was probably a result of washout of intracellular polyamines [⁴¹–⁴³]. In Figure 9(E), in the interests of time, Ba²⁺ was applied earlier in the recording.

MiR-212 is one of the most up-regulated miRNAs (8.1-fold) in human heart failure [²¹]. Its targeting of K_ir2.1 is of interest because K_ir2.1 is considered to be the major component of cardiac I_K1 [¹⁷,¹⁸]. Down-regulation of I_K1 contributes to the risk of cardiac arrhythmia during the process of myocardial remodelling en route to heart failure [⁵–⁷,⁴⁴]. MiR-1 has been identified as important in down-regulating I_K1 in acute myocardial infarction [³⁵], but miR-1 is not consistently up-regulated in chronic human cardiomyopathy, being found to be actually decreased in four [⁴⁵–⁴⁸] and unchanged in one out of seven studies [⁴⁹].

Inward-rectifier K⁺ channels have an important role in regulating vascular tone in small cerebral arteries such as the basilar artery [¹⁹,⁵⁰]. Chronic alcohol consumption is associated with dysfunction of the cerebral vasculature [⁵¹–⁵³], which may contribute to cerebrovascular disorders seen in alcoholism [⁵⁴]. K_ir2.1 may be involved, as inward-rectifier K⁺ channel activity and K_ir2.1 protein were reduced in basal arteries of alcohol-fed rats [⁹]. Interestingly, miR-212 was up-regulated 20-fold in colon biopsy samples from alcoholics [²²]. The effect of alcohol on miR-212 expression in basilar artery is unknown, but it would be intriguing to investigate the possible link between miR-212 and K_ir2.1 in alcohol-induced basilar artery dysfunction.

The miR-212 target site in the K_ir2.1 3′UTR was predicted by TargetScan and MiRanda, but not by PicTar. The different prediction algorithms do not agree with each other in many instances, illustrating the need for an efficient functional assay for identifying miRNA target sites unequivocally. The miR-212 site is classified as poorly conserved by TargetScan, although it is conserved across sixteen of the listed mammalian species. We decided to focus on miR-212 over other miRNAs because of its dramatic up-regulation in conditions that are characterized by a decrease in inward-rectifier K⁺ channel function, as discussed above. Thus we approached the issue from a physiological perspective and used bioinformatics as a tool, rather than using the bioinformatics analysis as a starting point.

The results of the present study identify miR-212 as a candidate regulator of K_ir2.1 expression. Future studies may show a functional link between miR-212 and K_ir2.1 in diseases such as heart failure and the cerebrovascular complications of alcoholism. MiR-212 would then become a potential therapeutic target. Until very recently the use of miRNA-based therapy seemed a distant prospect; however, Montgomery et al. [⁵⁵] have used an ‘antagomir’ to improve cardiac function and survival in Dahl hypertensive rats. Others have employed a viral vector (AAV9)-dependent approach to successfully and relatively selectively target short hairpin RNA molecules to the heart in rodents [⁵⁶]. In terms of therapeutic value for humans the first human clinical trials of miRNA inhibition are underway (Santaris Pharma; www.clinicaltrials.gov).

We have presented a novel assay for identifying functional interactions between miRNAs and a 3′UTR. Further refinements could make the assay suitable for high-throughput screening. We have shown that a strategy that combines analysis of genes likely to be responsible for a pathogenic phenotype with those miRNAs altered in the same disease is an effective way to identify functional interactions. Specifically, the down-regulation of K_ir2.1 in heart failure and alcoholic cerebrovascular dysfunction may be functionally linked to the up-regulation of miR-212.