Bioelectrochemistry of nucleic acids for early cancer diagnostics – analysis of DNA methylation and detection of microRNAs

Current methods of DNA methylation analysis

Vast majority of techniques for analysis of DNA methylation are based either on sodium bisulfite reaction or on application of methylation-sensitive restriction enzymes. Sodium bisulfite treatment of DNA converts cytosine to uracil but not the methylcytosine (mC), which results in two different DNA sequences depending on their original methylation status (Figure 2A). This is important, as both C and mC pair with guanine, making conventional hybridization techniques ineffective if some kind of pretreatment is not used. A pioneering work by Herman et al. (1996) introduced methylation-specific PCR (MSP), amplifying bisulfite-treated DNA using primers specially designed either for unmethylated DNA (the primer contained more adenines as template after bisulfite conversion had more uracils) or for methylated DNA (primer with more guanines) (Figure 2B). The method is simple, fast, and still used today; the main drawback is that it is not quantitative enough. Quantification, on the other hand, is provided by MethyLight assays, combining MSP with fluorescent TaqMan probes (Eads et al. 2000; Olkhov-Mitsel et al. 2014). These probes are sequence-specific oligonucleotides with a fluorophore attached at the 5′-end and a quencher moiety located internally or at the 3′-end. During the PCR, the probe is cleaved by the exonuclease activity of DNA polymerase, separating the fluorophore from the quencher, increasing the fluorescence intensity. Usually, two TaqMan probes are used, one, designed for unmethylated site and the other, for methylated site, making the method highly specific.

Figure 2:

(A) Sodium bisulfite converts cytosine into uracil (in series of steps), but methylcytosine is not converted. (B) An example of DNA sequences that become modified after bisulfite treatment and subsequent PCR amplification, depending on whether they were originally unmethylated (top) or methylated (bottom).

Bisulfite pyrosequencing provides information on methylation status of individual cytosines (Tost and Gut 2007), although only in shorter DNA sequences up to 250 bp in length. DNA is first treated with sodium bisulfite and then amplified using primers, of which one is biotin-labeled. This generates double-stranded amplicon with biotin at one end. DNA is then denatured, and biotin-labeled strand is attached to streptavidin magnetic beads, ready for sequencing. During this process, DNA polymerase incorporates deoxynucleotide triphosphates, one type at a time (i.e. either dATP, dCTP, dGTP, or dTTP), into the growing strand of DNA template. The incorporation of dNTPs results in the release of pyrophosphate, which is converted into ATP by sulphurylase, driving subsequent conversion of luciferin to oxyluciferin (catalyzed by luciferase). The amount of light released in this process, proportional to the number of incorporated nucleotides, is detected with camera. Any unconsumed dNTPs are degraded with apyrase so that other dNTP may enter the reaction. Pyrosequencing is especially suitable for identifying methylation status of multiple candidate gene promoters, as demonstrated for genes associated e.g. with acute lymphoblastic leukemia (Kuang et al. 2010) or bladder cancer (Marsit et al. 2010). More recently, a popularity of next-generation sequencing (NGS) platforms helped them to enter also DNA methylation studies, especially regarding genome-wide methylation profiling (Harris et al. 2010).

The enzyme-based approach utilizes a class of restriction endonucleases, which specifically recognize and cut DNA only if the cytosine at restriction site is unmethylated; methylation of the cytosine prevents the restriction. The commonly used enzymes include HpaII (cutting at CCGG site), BstUI (CGCG), or HhaI (GCGC). There are also endonucleases that cleave methylated but do not act upon unmethylated DNA e.g. McrBC recognizing two half sites (G/A) mC separated by up to 3 kb (Badal et al. 2003; Hublarova et al. 2009). In both cases, the analyzed DNA sequence must contain the restriction site, which makes this approach dependent on the availability of methylation-sensitive endonucleases. The original Southern blotting analysis of cleaved fragments (Baldwin et al. 2000) is now being replaced with real-time PCR (Bruce et al. 2008), which quantifies amplicons generated only from originally methylated DNA (which was not cleaved). Another technique, Methylation-Sensitive Multiplex Ligation-dependent Probe Amplification (MS-MLPA), combines ligation, restriction, and amplification (Homig-Holzel and Savola 2012; von Kaenel and Huber 2013). In this technique, two oligonucleotide probes designed to adjacently bind to the target sequence are ligated to form a template, which is then PCR amplified using fluorescently labeled primers. PCR products are then separated and quantified with capillary electrophoresis. MS-MLPA utilizes endonucleases to distinguish methylated from unmethylated strands, as the endonuclease will cut only hybrids of unmethylated target sequence strands and the corresponding probes, blocking the PCR amplification of the ligated probes. MLPA probes have unique lengths, allowing detection of different targets in a multiplexed fashion (up to 50 loci in a single reaction is possible). Main drawbacks include semi-quantitative nature and increased time of reaction. Capillary electrophoresis has also been shown to be useful in quantification of mC amount in genomic DNA after its enzymatic hydrolysis into single nucleotides, as cytosine monophosphates were found to migrate faster than methylcytosine monophosphates (Stach et al. 2003; Berdasco et al. 2009).

COBRA (combined bisulfite restriction analysis), as the name implies, combines bisulfite conversion of DNA with restriction digestion (Xiong and Laird 1997). In this method, bisulfite-treated DNA is PCR amplified, resulting in amplicons containing cytosine residues at originally methylated positions or thymine residues at originally unmethylated positions. Subsequent introduction of suitable restriction endonuclease (e.g. BstUI) leads to a cleavage of originally methylated amplicons (having CGCG site) but not unmethylated (TGTG). Separation on gel electrophoresis enables calculation of methylation percentage, determined from a ratio of undigested and digested fragments. Again, the method has been applied also into cancer research e.g. for analysis of hypermethylation of MGMT gene promoter in glioblastomas (Goedecke et al. 2015) or ECRG4 gene promoter in colorectal carcinoma (Goetze et al. 2009).

Bioelectrochemistry in DNA methylation studies

Novel techniques and technologies have recently emerged, becoming potentially useful tools in DNA methylation studies. Electrochemistry is a good example of such emerging technology, providing benefits in terms of simplicity, cost, or analysis time. It also enables miniaturization of the system and parallel measurement of multiple samples at electrode arrays and chips. These attractive features have enabled electrochemistry to enter various fields of biomedical research e.g. detection and analysis of (i) DNA sequences (including bacterial and viral DNAs or mutated genes) (Campuzano et al. 2011; Palecek and Bartosik 2012; Rosario and Mutharasan 2014; Bartosik et al. 2016), (ii) microRNAs (see below), (iii) protein biomarkers (Bertok et al. 2013; Rusling 2013; Palecek et al. 2015), (iv) DNA-protein interactions (Lee et al. 2009; Balintova et al. 2015), (v) protein conformational changes (Palecek et al. 2015), (vi) DNA damage (Hocek and Fojta 2011; Sontz et al. 2012), or (vii) uptake of potential anticancer drugs (Bartosik et al. 2015).

Different EC strategies for analysis of DNA methylation have also been proposed. Perhaps the simplest strategy is direct discrimination of C and mC at the electrode surface, based on their different oxidation potentials (Goto et al. 2010; Wang et al. 2015; Brotons et al. 2016). For instance, Wang et al. (2015) showed that at polypyrrole-functionalized graphene nanowall interface, the potential difference between C and mC peaks within short oligonucleotides is 184 mV, sufficient for their discrimination. Although this approach does not involve restriction enzymes or bisulfite conversion, it requires rather high (micromolar) DNA concentrations. Moreover, it is questionable whether the method can specifically distinguish which cytosine is methylated and which is not.

Several papers reported on EC monitoring of DNMT activity and inhibition by using restriction endonucleases and properly designed DNA sequences (Wu et al. 2012; Muren and Barton 2013; Deng et al. 2014b; Furst et al. 2014; Furst and Barton 2015; Li et al. 2015; Zhang et al. 2015). Although the papers differ in choice of endonucleases, DNA sequences, electroactive labels, or even solid substrate, they all share basic idea behind this strategy. Electrode surface (or magnetic beads) was modified with unmethylated DNA duplex (with one strand usually labeled), containing restriction site for selected endonuclease. In the DNMT absence, incubation of the DNA duplex with the endonuclease caused DNA cleavage, removal of the label, and thus decrease of the signal. On the other hand, if the DNMT was present in the sample, DNA duplex has become methylated, resistant to the cleavage, which enabled the labeled duplex to generate EC signal. Although the papers were primarily aimed at monitoring DNMT activity, information whether certain DNA sequence was or was not methylated could be easily drawn. For instance, Barton’s group developed a platform for the detection of human DNMT1 activity from crude lysates of colorectal tumor biopsies, using gold electrode for duplex immobilization and EC measurement, a methylation-specific restriction enzyme BssHII (recognizing GCGCGC sequence), methylene blue as the electrocatalyst, and ferricyanide for amplification of the signal (Figure 3) (Furst et al. 2014; Furst and Barton 2015). They observed increased DNMT1 activity in colorectal tumors over healthy tissue, making it a possible indicator of cancerous transformation, but it should be noted that Western blotting and quantitative reverse transcription PCR (qRT-PCR) did not confirm this finding. Yet this report, one of the few successfully applied into human tumor tissue, represents an important step in development of new EC biosensing technologies, worth further attention. Using a similar approach, Li et al. (2012) employed graphene oxide to enhance assay sensitivity and tested inhibition effects of two anticancer drugs, 5-azacytidine and 5-aza-deoxycytidine, showing a decrease in DNMT activity with increasing inhibitor concentrations.

Figure 3:

Electrochemical scheme (left) and platform (right) for the detection of human methyltransferase activity from crude cell lysates. (Left) Overview of electrochemical detection scheme at the electrode. Crude cell lysate is added to the electrode containing the patterned DNA. If methyltransferase (green) is present (blue arrows), the hemimethylated DNA on the electrode is methylated (green dot) by the methyltransferase to a fully methylated duplex; if methyltransferase is not present (red arrows), the hemimethylated DNA is not further methylated. A methylation-specific restriction enzyme, BssHII (purple), is then added, cleaving only hemimethylated duplex, leading to a diminished signal. If the DNA is fully methylated, DNA is not cleaved and the electrochemical signal associated with methylene blue (MB⁺) binding to DNA remains protected. (Right) The electrochemical detection platform contains two electrode arrays, each with 15 electrodes in a 5×3 array. Reprinted from Furst et al. (2014) with permission from Proceedings of the National Academy of Sciences of the United States of America.

miRNAs role in cancer

Alteration of miRNA expression is a very frequent phenomenon in cancer. MicroRNA genes are often located in chromosomal regions susceptible to amplifications or deletions/mutations, causing miRNAs upregulation or downregulation, respectively. Upregulation of miRNAs targeting tumor suppressor genes leads to a decreased expression of tumor suppressor proteins (favoring tumor formation). Conversely, downregulation of miRNAs targeting proto-oncogenes causes their increased expression (again, favoring tumor formation by increased synthesis of oncoproteins, enhancing thus cell proliferation). These two processes are depicted in Figure 4.

Figure 4:

Two alternative routes to carcinogenesis involving microRNA deregulation. (Left) Overexpression of miRNA regulating tumor suppressor gene leads to its decreased expression, resulting in fewer tumor suppressor proteins to protect the cell. (Right) Downregulated expression of miRNA regulating oncogene (i.e. gene coding an oncoprotein, which supports cell proliferation and division) is not present in sufficient amounts, and thus increased number of oncoproteins is synthesized.

Croce’s group was among the first who linked miRNA deregulation to carcinogenesis. They found that a DNA region frequently deleted in chronic lymphocytic leukemia does not code for a protein but rather for two miRNA sequences, miR-15a and miR-16-1, which became downregulated (Calin et al. 2002). This stimulated further research focused on miRNA roles in cancer, with thousands of papers published each year. miRNAs might play several roles in connection with cancer diagnostics and treatment. First, they represent potentially useful biomarkers for early cancer diagnostics. As an example, a panel of miRNAs was reported for lung cancer already 1–2 years before its development, which could help to select individuals for further surveillance (Boeri et al. 2011). Interestingly, miRNAs were also shown to help in identifying the origin of poorly differentiated tumors, which already metastasized to different sites, indicating that primary tumors maintain a unique miRNA signatures (Lu et al. 2005). Moreover, monitoring selected miRNA levels as predictors of response to therapies might be of great value. For instance, inhibition of overexpressed miR-21 and miR-200b increased sensitivity towards gemcitabine treatment of cholangiocarcinoma (Meng et al. 2006) and inhibition of miR-221 and/or miR-222 sensitized breast cancer cells to tamoxifen-induced cell growth arrest and apoptosis (Zhao et al. 2008).

Perhaps the most visionary but also the most challenging is to use miRNAs as therapeutic tools, usually involving either construction of oligonucleotides (so-called antagomiRs) inhibiting oncogenic miRNAs or reintroduction of downregulated tumor suppressor miRNAs, miRNA mimics, into the cancer cells and tissues (known as miRNA replacement therapy). Yan et al. (2011) tested anti-miR-21 and showed inhibited growth and migration of MCF-7 and MDA-MB-231 cells in vitro and tumor growth in nude mice, providing potential therapeutic applications in breast cancer treatment. However, the questions regarding the stability of miRNAs in vivo, the effectivity of the delivery to target organs, or potentially severe side effects remain to be answered (Bertoli et al. 2015).

Table 1 shows just a few of many microRNA sequences that were found to be either overexpressed or underexpressed in various tumors.

Nonelectrochemical techniques

Due to so many important biological roles of miRNAs, there is no surprise that a plethora of techniques for miRNA detection has been developed or modified from already existing techniques to better meet the specific needs of miRNA detection. One such requirement is specificity, which is critical as members of miRNA family often exhibit highly similar sequences, differing in only one or two nucleotides. In addition, miRNAs represent only a minor fraction (~0.01%) of total RNA extracted from biological samples, and thus highly sensitive assays are required (Hunt et al. 2015).

Conventional techniques used in miRNA detection comprise Northern blotting (NB), microarrays, or qRT-PCR. In NB, which has been used since the very beginning of miRNA research, electrophoretic separation of RNA molecules with different lengths is followed by their transfer to a positively charged membrane and hybridization of miRNAs with DNA probes labeled with either ³²P radioisotopes or with haptens (e.g. digoxigenin for subsequent chemiluminescence detection). In terms of sensitivity, NB does not particularly excel in comparison with other methods and requires relatively high input amounts of the RNA. Certain modifications of NB settings have led to an improved sensitivity e.g. application of locked nucleic acids (LNA) probes instead of DNA probes (improving also mismatch recognition) (Válóczi et al. 2004) or cross-linking of the RNA to the membrane via 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide chemistry (avoiding UV irradiation of miRNAs), coupled to a ligation step that joins together miRNA and labeled DNA probe (Wu et al. 2014).

Microarrays rely on miRNA hybridization with complementary DNA probes attached to a solid substrate; it is worth noting that it is usually miRNA that is enzymatically labeled. A major advantage is the ability of parallel analysis of hundreds of miRNAs due to distinct localization of individual probes to specific spots within the solid substrate, making the method truly high-throughput. On the other hand, microarrays suffer from being a semi-quantitative technique, requiring another type of validation to quantify expression. Moreover, as different miRNA sequences exhibit different melting temperatures (T_m), hybridization of many miRNAs in parallel would inherently require individual T_m settings for each miRNA/probe duplex in order to obtain high specificity, a task hardly achievable on microarray plates. Again, many modifications to address the above-mentioned challenges were suggested, based e.g. on (i) the application of the Klenow fragment of DNA polymerase I for extension of unmodified miRNAs (hybridized to DNA probes) with biotinylated adenosines and detection via fluorophore-conjugated streptavidin (in so-called RNA-primed, array-based Klenow enzyme, or RAKE assay) (Nelson et al. 2004), (ii) biotin-labeled structure-specific RNA binding protein recognizing array-captured miRNAs (Lee et al. 2010), or (iii) short fluorophore-linked Universal Tag oligonucleotides selectively captured by the target-bound probes via base-stacking effects (SHUT assay) (Duan et al. 2011).

One of the most popular techniques for validating and accurately quantifying miRNAs is qRT-PCR. The first step is a reverse transcription of RNA into cDNA. Due to a short miRNA length, comparable to that of a primer, ingenious strategies must have been developed to make RT more effective e.g. application of stem-loop primers (Chen et al. 2005) or incorporation of poly(A) tail (Shi et al. 2012). Once the cDNA is generated, qPCR may proceed using miR-specific and universal qPCR primer set and TaqMan probes for detection of amplicons. The method is highly sensitive and specific but remains laborious and is especially demanding on quality of the input material (as degradations or impurities result in errors that are amplified during PCR) (Hunt et al. 2015). It has been recently used for analysis of circulating miRNA in serum and plasma (Kroh et al. 2010; Zhu et al. 2014a). However, careful optimization and standardization are still needed to obtain reliable results.

Bioelectrochemistry of miRNAs

EC analysis of RNA has started long before the first papers on microRNA appeared. These early papers reported e.g. polarographic discrimination of dsRNA and ssRNA (Palecek and Doskocil 1974), determination of picomole levels of RNA in a mixture of DNA using differential pulse voltammetry (Palecek and Fojta 1994), or trace measurements of RNA using potentiometric stripping analysis at carbon paste electrodes (Wang et al. 1995). Nowadays, majority of EC assays on RNA focus on miRNA detection, often referred to as biosensors, which are a fruitful area of study with many different strategies available. These vary e.g. in a choice of the solid support on which hybridization happens (electrode, magnetic beads, etc.), a type of label or reporter attached to one of the DNA probes, or employed EC technique (amperometry, voltammetry, impedance spectroscopy, etc.). There are many reviews on EC detection of miRNAs available in the literature, providing more detailed view on research strategies, detection limits, EC techniques, etc. (Palecek and Bartosik 2012; Hamidi-Asl et al. 2013; Campuzano et al. 2014a; Lautner and Gyurcsanyi 2014; Hunt et al. 2015; Keshavarz et al. 2015; Labib and Berezovski 2015). This is not surprising given the fact that almost 200 papers appeared in the last few years on this topic. Here, we wish to briefly describe major challenges in miRNA detection and give some examples how they can be addressed.

During biogenesis, a mature ~21–23-nt miRNA is formed from its longer intermediates, pri-miRNA (1–3 kilobases) and pre-miRNA (60–100 bases), possessing the same sequence motif as mature miRNA. When working with cell lysates or other real samples, these intermediates present in the sample may also become captured by complementary DNA, leading to false negatives. An interesting strategy to avoid this is to use viral protein p19, which specifically binds and sequesters only short RNA duplexes. Campuzano et al. developed a magnetosensor for detection of endogenous miR-21 (playing a role in variety of cancers) in total RNA extracted from cancer cells and human breast-tumor specimens without PCR amplification and sample preprocessing (Figure 5A) (Campuzano et al. 2014b; Torrente-Rodriguez et al. 2014) and also reported dual amperometric sensor for simultaneous detection of miR-21 and miR-205 (Torrente-Rodriguez et al. 2015). Similarly, Labib et al. (2013) described a so-called HPD sensor based on hybridization (H), p19 protein binding (P), and protein displacement (D) and applied the sensor for detection of five miRNAs i.e. miR-21, miR-32, miR-122, miR-141, and miR-200.

Figure 5:

Examples of EC-based assays for miRNA detection. (A) Strategy based on p19 protein, which selectively captures RNA/RNA hybrids, in this case, a miRNA of interest hybridized with a properly designed RNA probe labeled with biotin. Subsequent addition of streptavidin/peroxidase generate EC signal, measured at screen-printed carbon electrodes (shown right). (B) Osmium-based approach. Biotinylated capture probes (black) immobilized at the streptavidin magnetic beads (SMB) hybridize with six-valent osmium-labeled miRNA complementary to the probe (green), followed by stringent washing to remove interfering molecules and by a release of the captured miRNA into solution. Labeled miRNA is then adsorbed at the electrode surface and transferred into the blank electrolyte for EC measurement. Reprinted from Bartosik et al. (2014a) with permission from Wiley. (C) Reaction of Os(VI)L with ribose at the 3′-end of the RNA molecule. The Os(VI)L complexes do not react with nucleic acid bases or deoxyribose. L is nitrogenous ligand e.g. 2,2′-bipyridine, temed, etc.

Another challenge in miRNA detection is its short length, making a traditional way of using two DNA oligo probes rather ineffective, as the oligos would have to be very short. One way out of it is to use a ligase. Here, two short DNA strands adjacently hybridize to the miRNA of interest, forming a duplex with a gap between the DNA strands. Then T4 DNA ligase is usually applied, which acts upon the duplex by covalently joining 3′-hydroxyl group of one DNA strand with the adjacent 5′-phosphate group of the second DNA strand (Cheng et al. 2014; Zhu et al. 2014b). This creates one longer DNA oligo, with one end originally attached to the solid substrate and the other end labeled for generating signal. The strategy was used e.g. for simultaneous detection of miR-155 and miR-27b (overexpressed in breast cancer) in a single-tube experiment (Zhu et al. 2014b). Magnetic beads were modified with capture DNA probes for both miRNAs, while PbS and CdS quantum dots served as labels for miR-155 and miR-27b, respectively, reaching fM detection limits. Similar work was reported for let-7b detection, using biotinylated DNA probe and a conjugate of streptavidin and alkaline phosphatase (Cheng et al. 2014). Authors applied the system to the cell lysates and obtained good correlation with qRT-PCR when monitoring content of let-7b extracted from HeLa, MCF-7, and A549 cancer cells.

Above, we mentioned that good specificity (i.e. ability to capture only target sequences and not mismatches) is a very important feature of any miRNA assay. One option to achieve high specificity is to use structural analogs of DNA-locked nucleic acids (LNA) and peptide nucleic acids (PNA) as capture probes. LNA contains methylene bridge between 2′-O and 4′-C on ribose, locking it in 3′-endo conformation. This causes increased rigidity of LNA and thus elevated melting temperatures of resulting duplex (Lautner and Gyurcsanyi 2014), leading to an increased hybridization affinity to DNA and RNA. Moreover, LNA exhibits improved mismatch discrimination. PNA also possesses improved hybridization affinity towards DNA or RNA, but for a different reason. It contains neutral peptide backbone instead of a negatively charged phosphate backbone, thus no repulsion between hybridizing strands occurs. Both analogs were applied also in miRNA assays, such as in LNA-based strategy for miR-21 detection, which successfully discriminated perfect duplex from single- and triple-base mismatches (Yin et al. 2012), or when PNA probes were used to distinguish let-7b from let-7a and let-7c miRNAs (Deng et al. 2014a). Wang et al. proposed an interesting strategy based on the so-called Y-junction, in which two capture probes were designed to partially hybridize with each other, resulting in a lower T_m of the duplex. Flanking (unhybridized) ends of these probes (having sequences complementary to the target miRNA) formed in the miRNA presence a longer duplex made of three strands having a “Y” shape, with substantially higher T_m. As hybridization occurred at the thermally controlled electrode, hybridization temperature could be set to retain perfect Y-shaped duplexes but to melt mismatches (Wang et al. 2013). Our approach to address specificity issues was also based on applying higher temperature, namely, on a precise determination of T_m of perfect or mismatched duplexes (with T_m decreasing in the order perfect duplex>single mismatch>double mismatch) and on choosing elevated hybridization temperature in such a way that perfect duplexes were retained while mismatches were renatured (Bartosik et al. 2014a; Bartosik et al. 2014b). In this strategy, we used complexes of six-valent osmium and bipyridine [Os(VI)bipy] to create adducts with miRNA, which were then electrochemically detected at mercury-based electrodes (including solid amalgam electrodes). Due to a catalytic nature of the signal, we reached subnanomolar detection limits (Bartosik et al. 2013). Using this quick, simple procedure, labeled miRNAs were hybridized with biotinylated capture probe attached to the streptavidin magnetic beads, and target miRNA was detected in total RNA samples isolated from human cancer cells (Figure 5B) (Bartosik et al. 2014b). Later, by employing so-called adsorptive transfer stripping technique, in which working electrode was dipped into a microliter-sized drop for direct adsorption of miRNA, we achieved increased sensitivity and decreased sample consumption, enabling detection of nanogram quantities of miRNA present in 500-fold excess of total RNA (Bartosik et al. 2014a).

Perhaps the highest emphasis is currently put on sensitivity. Investigators developed a myriad of assays that focus on pushing detection limits as low as possible. These strategies were many times reviewed, so we will only briefly describe the most recent advances made in the last 2 years. Ma et al. (2016) reported so-called isothermal strand-displacement polymerase reaction, in which miRNA served as a displaceable strand during the primer extension. In short, magnetic beads were first modified with a molecular beacon consisting of ssDNA probe with a sequence complementary to the miRNA. After miRNA hybridized to the probe, a primer has bound to the flanking end of the probe, triggering the amplification reaction catalyzed by phi29 DNA polymerase with a strong displacement activity. This released miRNA into the solution, while new DNA strand containing biotin-labeled cytosines has been synthesized, ready for interaction with streptavidin-alkaline phosphatase conjugate required for subsequent EC measurement. Moreover, released miRNA could bind to another beacon, triggering new amplification reaction, and thus few miRNA molecules ultimately translated into an increased amount of biotin-labeled DNA amplicons, providing a limit of detection of 9 fM. A relatively simple impedance-based protocol has been proposed by Zhang et al. (2016), employing capture-probe modified magnetic beads and duplex-specific nuclease, which preferably cleaves DNA in DNA/RNA hybrids but not ssDNA itself. In the absence of miRNA, no cleavage occurred and the probes, which remained at the magnetic beads, formed a compact negatively charged layer with an increased charge-transfer resistance. On the other hand, miRNA presence caused a digestion of the DNA probes, removing the negatively charged layer, which led to appreciable smaller charge-transfer resistance. Limit of detection was 60 aM, and the method was employed to detect miR-21 in human serum samples of breast cancer patients. Another interesting approach comprised 3D tetrahedron DNA structure made of four DNA probes immobilized at the surface of a gold electrode, coupled with a hybridization chain reaction (Ge et al. 2014). After the tetrahedron was formed, target miRNA hybridized with one of the probes protruding into the solution, followed by an addition of two separate biotin-labeled DNA hairpins. miRNA served as an initiator, opening the first hairpin structure (hairpin 1) to create a duplex. Opened (linearized) hairpin 1 then exposed a new single-stranded region that opened hairpin 2, which in turn opened another hairpin 1, etc., resulting in a chain reaction until the hairpin supply became exhausted (Evanko 2004). As both hairpins were labeled with biotin, the growing chain could accommodate multiple streptavidin-peroxidase conjugates, leading to an increased sensitivity.