MicroRNAs (miRNAs) are non-coding small RNAs, suppressing gene expression or inhibiting translation by binding (sometimes partially) complementary sequences in the 3' UTR of mRNAs. Recent studies indicate that microRNA-AGO complex, when paired with an unusual mRNA, promotes microRNA decay rather than mRNA degradation [3]. MicroRNAs regulate biological processes including cell proliferation, apoptosis, differentiation, metabolism, development, neurogenesis [4], neuronal functions [5, 6] and neoplastic transformation [7, 8]. MicroRNAs have been proposed as diagnostic markers [9] and the therapeutical potentials of microRNA mimics are being tested [10]. MiRNA genes are encoded in intergenic chromosomal regions or within the introns of protein-coding genes. Most recently, miRNAs have been shown to originate from protein-coding gene promoters with unique features [11]. MicroRNA genes are transcribed by RNA polymerase II or III, generating RNA molecules containing the characteristic stem-loop structures (primary RNA). Drosha-DGCR8 microprocessor complex cuts the primary RNA into shorter ~70 nt pre-miRNAs. The pre-miRNAs are exported and further processed by an RNase Dicer to imperfectly paired ~22-bp double-stranded mature miRNA. One of the strands is integrated into RNA-induced silencing complex, and functions during post-transcriptional gene regulation. The strand selection is influenced by, in the case of miR-324, uridylation by TUT4/7 [12]. Mutations among miRNAs have been identified for inherited diseases [13-15]. Genome-encoded transcripts may also regulate the degradation of miRNAs [16]. MicroRNA regulates the de-differentiation and proliferation of cardiomytes, and expression of human microRNA-199a in infarcted pig hearts can stimulate cardiac repair [17] and genes involved in biomechanical strength [18]. McGeary SE et al systematically measured the relative binding affinities of six mammalian miRNAs miR-1, let-7a, miR-7, miR-124, lsy-6 and miR-155 against all ≤12-nucleotide sequences using RNA bind-n-seq and identified noncanonical target sites for each miRNA and the significant impact of dinucleotides flanking each site [19]. J Hordeaux et al integrated a microRNA 183 target sequence in AAV vectors to reduced the expression of transgenes in dorsal root ganglion, and thus reduce the dorsal root ganglion toxicity in AAV-based gene therapies [20].

In miRBase database version 22 [21], released in March 2018, 38589 hairpins and 48885 mature products are listed for 271 species [22]. Some of the entries may be misclassified and are ribosomal RNA-derived small RNAs (rsRNAs) [23]. It is estimated that 90% of human genes are regulated by microRNAs. However, the regulation of target genes is hard to define and elucidate. Bioinformatics software prediction, integrated with gene chips and biological experiments, have been used to identify miRNA functions and their target genes. Gene chip technology profiles the expression of a large number of miRNAs [24]. Comparison of microRNA expression profiles between, for example, normal and tumor tissue samples, can be used for the identification of tumor biomarkers, and even clinical diagnosis of tumors, especially exosome-based miRNAs [25]. Such exosomes can induce tumorigenesis through miRNAs [25]. However, gene chip technology remains semi-quantitative and needs to be corroborated by other experimental methods. Besides, miRNA mimics have been tested as therapeutics, for example, a miR-29 mimic for pulmonary fibrosis [26].

Three methods have been used to detect specific miRNA in tissues or cells: Northern hybridization, in situ hybridization, and stem-loop real-time PCR. Each has its advantages and disadvantages. These three methods can be combined to confirm miRNA expression level. Recently, a novel approach based on barcode DNA with one region for binding specific miRNA and another region for detection and quantitation has been designed, and the method has been shown to detect aM level of miRNAs without enzymatic amplification and to differentiate single-base mismatches [27]. High-throughput sequencing, called small-RNA sequecing (sRNA-seq), has also been explored to detect and quantitate miRNAs [3, 18]. Wang N et al described a detailed protocol for a small RNA library for single-cell RNA-seq in addition to single-cell mRNA-seq [28].

Northern blot is one of the widely used methods for detecting RNA levels. MicroRNAs are small molecules, and some of the microRNA expression levels are very low. The RT-PCR method tends to result in poor reproducibility, and the steps involved are complicated. Most researchers currently use Northern blot, which has good reproducibility, high sensitivity, a direct approach, and can be used to detect the change in microRNA expression level. Among various ways to detect the hybridized probe, isotope labeling of the probe has some limitations due to the concern of radiation contamination. Locked-nucleic acid (LNA) probes, with high stability, specificity, and no radioactive contamination are widely used [3].

Figure 1. Northern blot indicates both pre- and mature miR-499 bands from human and mouse. From [1].

Basic principles: RNA molecules are denatured and separated by urea polyacrylamide gel electrophoresis, transferred to nylon membrane, fixed, and then hybridized with DNA or RNA probes labeled with an isotope, digoxin [23], or other markers. Target miRNAs with sequences complementary to a probe are detected, and the relative size and intensity can be measured.

1. prepare 15% of Urea-PAGE gel without APS and TEMED. 50ml of Pre-Gels: Urea: 24 g; 40% acrylamide: 18.75 mL; 5x TBE Buffer: 10 mL; ddH2O: 0 mL.
2. clean containers and other devices: the containers for Western blot device can be used.
   1. wash with water
   2. soak electrophoresis tank, glass, and comb in H₂O₂ for 10 minutes
   3. wash with 0.1% DEPC-treated water
3. make gel: add 33.3- 45 ul APS and 10-20ul TEMED in 10ml pre gel and mix gently, then add it into the glass and insert the comb into the gel.
4. prepare RNA samples: add 2ul 10xRNA loading buffer in the RNA sample dissolved in the 18ul DEPC H₂O. 95°C for 5min, cooled on ice. Samples should be collected by centrifugation.
5. run electrophoresis: in 1x TBE. 180V.
   Ambion: 300V pre-electrophoresis 10min (to prevent leakage while activating adhesive glue) before adding the sample, then rinse the holes with electrophoresis solution 1xTBE and carefully add the sample.
   Roche: provides LNA-labeled DNA as control, synthetic RNA Oligo as positive controls (1 pmol total).
   Bromophenol Blue indicates 13nt, Xylene Cyanole FF 40nt.
6. transfer to membrane: in 0.5x TBE
   1. soak the gel in 0.5x TBE about 10 min.
   2. stain the gel with EtBr (0.5ug/ml) for 10 min. observe the RNA by UV. wash the gel with 0.5x TBE for 5 min.
   3. soak the Nylon membrane (GE Corporation) and two pieces of thick filter paper in 0.5x TBE for 10 min.
   4. the filter paper, plastic, film, filter paper form a sandwich structure, the gel should be close to the cathode side. No air bubbles between each layer.
   5. assemble the membrane device
   6. transfer on ice 1 h 300mA
   7. the membrane (RNA side up) on the UV cross-linking apparatus, for 4000J UV cross-linking
   8. the film clip in the middle of thick filter paper, 80°C, bake 0.5 - 2 h.
   9. methylene blue staining 3 ~ 5min.
7. hybridize
   1. pre-hybridize 1 h in hybridization solution (7% SDS, 0.2M Na₂HPO₄) in a hybridization oven at 37 - 42°C. Add the membrane into the hybridization bag, and then add the Dig-labeled probe, hybridize overnight in the oven.
   2. 37 - 42°C membrane washing solution (2× SSC, 0.1% SDS), wash membrane three times, each time 15 min.
   3. room temperature MABT (0.1M maleic acid, 0.15M NaCl, 0.3% Tween-20, PH 7.5), wash membrane three times, each time 5 min.
   4. block with the blocking solution for 1 h.
   5. add the anti-Dig antibody and incubate at RT for 40 min. The antibody solution should be centrifuged at 12000g for 5 min prior to use.
   6. wash membrane with MABT buffer twice, each time 15 min.
   7. balance the membrane in the detection buffer for 5 min.
   8. add 1ml CSPD buffer in the bag, and incubate for 5 min.
   9. squeeze out the CSPD buffer, and incubate for 3-15 min at 37°C
   10. expose for 2 min ~ 1 h.

notes:

1. urea solution can not be heated; 15% of the urea-PAGE gel without APS and TEMED can be stored at room temperature or 4°C refrigerator for a month or so.
2. all devices should be maintained as RNAase free.
3. LNA probes should be aliquoted before use, and then kept frozen in a -70°C refrigerator.
4. hybridization's time and temperature need to be optimized with different miRNAs.
5. nylon membrane after UV cross-linking and baking can be stored at 4°C for more than six months.

Many protocols and variations for in situ hybridization with different types of tissues (whole-mount, paraffin-embed, or frozen tissues), and with different detection paradigms are available from various sources, especially online. Commercial kits are available. For example, Gabisonia K et al deteced the expression of microRNA-199a using LNA probes for miR-199a-3p and U6 snRNA with a Qiagen miRNA ISH kit for formalin-fixed paraffin-embedded heart tissues [17]. SE Sillivan et al examined the expression of mir-135b-5p on frozen brain sections with LNA dual 5’- and 3’- DIG-labeled probes and tyramide signal amplification [29].

Figure 2. In situ hybridization of miRNA. Both negative and positive controls are important in in situ hybridization. From [2].

One of the widely cited protocols is by Dr. Pena in 2009 [30]. The protocol is for mammalian tissues fixed with formaldehyde and 1–ethyl–3–(3–dimethylaminopropyl) carbodiimide (EDC). The supplement in that article lists detailed steps of the experimental procedure. Since miRNAs are very small, one of the crucial steps is to prevent the loss of miRNAs during the hybridization procedure. Fixation with EDC can irreversibly immobilize miRNAs at its 5' phosphate to the protein matrix. Fixation with formaldehyde or EDC alone can not prevent the loss of miRNA from tissue sections during in situ hybridization. Besides, the melting temperature between the probe and the intended miRNA differs from usual calculation due to the interference of formamide during the hybridization step. This alteration of melting temperature should be taken into consideration [30].

Regular real-time quantitative polymerase chain reaction (PCR) can only be used to detect miRNA precursors, while stem-loop real-time quantitative RT PCR technology can be used to detect mature microRNA [31]. Stem-loop real-time quantitative RT polymerase chain reaction (PCR) is highly sensitive and highly specific for the detection of miRNA expression. It includes two steps: design of reverse transcriptase primers with stem-loop structure and RT-PCR with fluorescently labeled miRNA probes. This technology has the following advantages: highly specific; able to distinguish homologous miRNA sequences; broad linear detection range of miRNA concentrations; highly sensitive; small sample consumption, only 1 ~ 10 ng of total RNA; can be used with total RNA samples, purified RNA samples, or cell lysates. For instance, Gabisonia K et al quantified miR-199a-3p in pig heart tissue by isolating total RNA with Qiagen miRNeasy Mini kit, reverse-transcribing with Exiqon miRCURY LNA PCR synthesis kit and performing RT–PCR with Exiqon pre-designed miRCURY LNA PCR primer sets and miRCURY LNA SYBR Green master mix [17]. Hyun Y et al extracted total RNA from plant shoot apex using the NucleoSpin RNA set for NucleoZOL from MACHEREY-NAGEL and quantified miR156 level against small nucleolar RNA 101 (snoR101) in qPCR [32]. Specific cDNA synthesis and PCR kits such as TaqMan Advanced miRNA cDNA Synthesis Kit (A28007) and TaqMan Advanced miRNA Assays (Thermo Fisher) using the TaqMan Fast Advanced Master Mix (4444557) from Thermo Fisher can be used [9]. H Qian et al added 10 μg/ml glycogen during Trizol extraction of total RNA to enhance precipitation of small RNAs [4].

Basic principle: microRNA RT-PCR generally uses stem-loop structure primers. Such stem-loop microRNA probes are designed to detect target miRNAs, and thus possess specific sequences. Small RNA U6 is usually used as an internal control, and random primers are used for the reverse transcription of small RNA U6. The reversely transcribed cDNA from RNA samples or cell lysates serve as the template for RT-PCR. Specifically designed forward and reverse primers against the target miRNA are used to detect the specific miRNA, while forward and reverse primers against small RNA U6 are used as an internal control. For serum samples, miR-16 has been used as an internal control [33].

Procedure

1. total RNA extraction with Trizol. same as a regular RNA Trizol extraction. except during isopropanol processing, the sample is chilled at 4°C. store the extracted RNA in -80°C.
2. reverse transcription of the RNA sample. use 12.5 ul system. RNase free water ; Total RNA: 0.625ug; Impron buffer: 2.5ul; dNTPs: 2.5ul; RNase inhibitor: 0.625ul; Impron MgCl₂: 1.5ul; Primer: 0.5ul; Reverse transcriptase: 0.625ul. detailed steps: calculate the amounts of template and RNase free water; add that amounts to PCR tubes; store the left template at -80°C; prepare the reaction mix as above; aliquot the reaction mix to each PCR tubes; and run the reverse transcription reaction. reverse transcription process: 25°C 5 min, 42°C 60 min, 70°C 15 min. and store at 4°C.
3. run the RT-PCR reaction in triplicates. use the following amount in each RT-PCR reaction. SYBR: 5.0ul; Primer mix: 0.2ul; cDNA template: 1.0ul; water: 3.8ul.
4. add 3.4 ul cDNA to the sample tubes.
5. calculate the amount of SYBR, primer mix and water; prepare the mix; add the mix to the sample tubes;
6. aspirate 10 ul from sample tubes, and add it to the 96 well plate. in triplicates.
7. cover the plate, and seal it.
8. load the sample to a machine. Specific PCR program is as follows:
   1. 50°C 2 min ;
   2. 95°C 5 min ;
   3. 95°C 30 s ;
   4. 60°C 40 s ;
   5. 72°C 30 s ;
   6. 95°C 15 s ; 60°C 30s ;95°C 15 s

note:

1. due to the high sensitivity of RT-PCR, always run experiments in triplicates.
2. the aliquot and sampling amount must be accurate.
3. mix the samples well before aliquoting them to the 96-well plate. The sample tubes can be vortexed (with slow speed to prevent spilling).

Bioinformatic and experimental methods have been used to identify the target miRNA genes. In combination, the accuracy can be improved significantly.

Bioinformatic methods use specific algorithms to screen and rank target genes. There are many algorithms. Although various prediction methods employ different calculations, they all center on some common features between microRNAs and their target genes.

1. complementarity between miRNAs and their target sites
2. conservation of homologous miRNAs between species
3. thermal stability between miRNA-mRNA double-strand
4. lack of complex secondary structure at the binding site of microRNAs
5. miRNA 5' end can bind to the target genes stronger than the 3' end

Very often, different prediction methods are used to generate lists of potential target genes, and the common genes among the lists become the focus of further investigation.

Although bioinformatic approach can predict target genes for microRNAs, the predictions need to be validated through experimental methods. Target genes can be examined on both protein and RNA levels. On the RNA level, the expression levels of many genes can be evaluated through microarray analysis after exogenous expression of specific miRNA in cells.

A straightforward method for target gene validation is through the transfection or knockdown of a miRNA and subsequent detection of RNA and protein levels of any target genes by quantitative PCR and Western blot. Another approach is to remove the seed sequence from a targeted gene and compare the effects of the miRNA on the target gene expression with or without the seed sequence. For example, Pandolfini L et al deleted the miRNA responsive element (MRE) from the 3' UTR of Hmga2 to study the involvement of let-7 on Hmga2 gene expression [34]. Moro A et al mutated MRE or used an miScript Target Protector from Qiagen to study the role of microRNA-regulated genes in endothelial mechanotransduction, and also used a sensor-seq approach to detect the activities of 98 MREs within 51 genes [18].