Sensitive Detection of Motor Neuron Disease Derived Exosomal miRNA Using Electrocatalytic Activity of Gold-Loaded Superparamagnetic Ferric Oxide Nanocubes

Graphical Abstract

Disease detection: Integration of the gold-loaded ferric oxide nanocatalyst (AuNP-Fe₂O₃NC) and methylene blue (MB)/[Fe(CN)₆]³⁻ redox cycling facilitates attomolar level detection of motor neuron derived exosomal miRNA.

Abstract

Dysregulated microRNA associated pathways contribute to the pathology of neurological disorders, hence presenting themselves as a potential candidate for motor neuron disease (MND) diagnosis. Herein, we reported an enzymatic amplification-free approach for the electrochemical detection of exosomal microRNA (miR-338-3p) from preconditioned media of motor neurons obtained from amyotrophic lateral sclerosis (ALS) patients and healthy controls. Our assay utilizes a three-step strategy that involves i) initial isolation and purification of exosomal miR-338-3p from patients and healthy controls using biotinylated complementary capture probe followed by heat-release of the specific target, ii) direct adsorption of target miR-338-3p onto the gold-loaded ferric oxide nanocatalyst (AuNP-Fe₂O₃NC) through affinity interaction between microRNA and exposed gold surfaces within the AuNP-Fe₂O₃NC, and iii) gold nanocatalyst-induced electrocatalytic signal amplification through methylene blue-ferricyanide redox cycling (MB/[Fe(CN)₆]³⁻). The electrocatalytic signal is monitored by using chronocoulometry at the AuNP–Fe₂O₃NC-modified screen-printed carbon electrode (AuNP-Fe₂O₃NC/SPCE). We demonstrated the detection of miR-338-3p as low as 100 aM in spiked buffer samples with a relative standard deviation of (%RSD) <5.0 % (n=5). We also demonstrate the successful detection of miR-338-3p from a small cohort of preconditioned media of motor neurons obtained from ALS patients and healthy controls. The sensor avoids the use of conventional recognition and transduction layers in hybridization-based electrochemical miRNA biosensors, polymerase-based amplifications. It is robust, fast (<2.5 h) and potentially applicable to a wide variety of RNA biomarker detection.

1 Introduction

Motor neuron disease (MND) is a progressive and fatal neurodegenerative disorder, leading to the loss of upper and lower motor neurons and eventually death within 2–3 years of illness onset.1, 2 MND results from the relapse of lower motor neurons in the spinal cord, and/or the brainstem and upper motor neurons in the motor cortex, or both.3, 4 It ultimately leads to progressive loss of the use of their limbs, i. e. it affects the ability to speak, move, breathe and swallow. Among the four clinical phenotypes of MND (amyotrophic lateral sclerosis, isolated bulbar palsy, progressive muscular atrophy and primary lateral sclerosis), amyotrophic lateral sclerosis (ALS) is the most common type representing 70 % of the cases occurring 2–3 per 100.000 people worldwide in a year.5 Recent studies have shown that mutations in the genes encoding the RNA-binding proteins FUS/TLS [ALS6 locus]6 and TARDBP/TDP43 [ALS10 locus]7 recommend the critical roles of regulatory RNA in the pathogenesis of ALS.8 Interestingly, these disease-associated RNA-binding proteins were recognized in neuronal RNA granules and with microRNA (miRNA)-associated complexes.9 To date, several studies have shown that miRNAs are differentially expressed in ALS patients when compared to controls in a variety of biofluids, including CSF, and in the blood-derived components plasma and serum and therefore considered as potential biomarkers for ALS.

MicroRNAs (miRNAs) are a large group of small (∼22 nucleotides) non-coding single-strand RNAs that play essential roles in gene expression via post-transcriptional regulation. They generally bind the 3'-untranslated region of mRNAs, leading to the gene silencing through mRNA cleavage, translational repression and adenylation.10, 11 Dysregulation of these highly conserved regulatory RNAs can potentially impact on the progression and prognosis of the disease and have gained immense interest in recent time as diagnostic biomarkers, especially in cancer.8 Non-coding RNAs are also reported as profusely expressed in the central nervous system (CNS), and aberrancies in miRNA expression patterns have been described in several neurodegenerative diseases.12 However, the roles of individual miRNAs are not fully understood in neurodegenerative disease (probably due to the nervous system complexity and technical difficulties), but several reports have provided evidence that miRNAs play significant roles in neurodegenerative disorders, including miR338-3p/miR-106/miR-451 in ALS.13, 14 For instance, miR-338-3p is stated to control several molecular pathways and contribute to ALS pathological processes through apoptosis, neurodegeneration, and/or glutamate clearance.15 Furthermore, miR-338-3p expression was noticeably upregulated in ALS patient leukocytes, serum, CSF and spinal cord in comparison to that of healthy controls, identifying it as an attractive prognostic circulating biomarker of ALS disease onset.16 Nevertheless, circulating miRNA suffers from RNAase cleavage and other environmental damage, therefore encapsulated miRNA such as exosomal (the lipid-bilayer protects them from RNase degradation, providing a stable source of miRNAs) or apoptotic miRNAs could be highly informative for early diagnosis.10, 17 Currently, miRNA detection mostly relies on conventional nucleic acid detection assays such as quantitative reverse transcription PCR (RT-qPCR), microarrays, Northern blot, and RNA-sequencing. Despite being reliable in laboratory settings, these conventional techniques are expensive and not suitable for the resource-poor and decentralized settings.10, 11 Electrochemical assays, on the contrary, have shown more potential for clinical application due to their inherent advantages of being inexpensive, simple, rapid, and miniaturized. Most of the electrochemical sensors for miRNA however still rely on multiple sensor fabrication steps, some sorts of enzymatic amplification (e. g. isothermal transcription mediation amplification (TMA)18 or reverse transcription-PCR (RT-PCR)19 and target RNA modification (e. g., polyadenylation, labelling) which could destabilize RNA and complicate the assay protocol.10 Furthermore, having rigorous target selectivity and faster analysis time, many of these sensors lack additional signal enhancement steps, thereby failing to achieve the sensitivity levels required for the analysis of miRNA in clinical samples.20

With the state-of-the-art advancement in nanotechnology, nanostructured materials with superparamagnetic behavior and biocompatibility are currently transitioning to a new paradigm of applications in the fields of biosensing, where they are used in developing novel methods and devices for diagnosis and monitoring of specific diseases via detecting levels of disease-specific biomolecules.11, 21 They exhibit advantages in molecular diagnostics, particularly in disease diagnosis by breaking down the barrier for structural miniaturization of diagnostic platforms, accelerating the signal transduction and hence boosting the assay sensitivity. Their electrocatalytic properties can be exploited to espouse many novel transduction schemes.21-23 Recently, we have reported a new class of gold-loaded superparamagnetic ferric oxide nanocubes (AuNP-Fe₂O₃NC), which exhibits multiple functionalities, e. g. enhanced catalytic activity toward the common electroactive molecules,21 direct adsorption aptitude for a large number of nucleic acids (DNA, RNA) through gold−nucleic acid affinity interactions,23 magnetic dispersible capture vehicles²⁴ and peroxidase mimetic activity (as nanozymes).25, 26 In our previous reports, we have achieved a limit of detection (LOD) of 10 pM for miRNA.21 This LOD could be further improved by chronocoulometric interrogation of methylene blue (MB), a commonly used nucleic acid redox marker, within the surface-bound miRNA coupled with the [Fe(CN)₆]^3−/4− redox system.

Herein, we report the electrocatalytic activity of AuNP-Fe₂O₃NC21 towards the MB/[Fe(CN)₆]^3−/4− redox cycling for the first time to achieve ultrasensitive detection of neurodegenerative disease (e. g. ALS)-specific exosomal miRNA. The exosomes were first extracted from the preconditioned media of motor neurons (ALS, cell culture and cell culture media collection were carried out using previous reports27-30). Target miRNA were then extracted from the exosomes and magnetically purified. After purification, they were directly adsorbed onto the AuNP-Fe₂O₃NC modified a screen-printed carbon electrode (SPCE) (magnetically bound onto SPCE) via gold-RNA affinity interaction. The level of electrode-bound miRNAs was quantified by chronocoulometric (CC) charge interrogation in the presence of MB, which was electrostatically attached with the guanine bases of target miRNA. The signal was further amplified by using the ferri/ferrocyanide ([Fe(CN)₆]^3−/4−) system. To test the applicability of our assay both in synthetic and ALS motor neuron-derived exosomal miRNA samples, we used miR-338-3p as a model target. This miRNA was reported to have a strong correlation with the progression of MND. The electrocatalytic activity of AuNP-Fe₂O₃NC coupled with the MB/[Fe(CN)₆]^3−/4− redox cycling facilitates our assay to achieve a LOD of 100 aM with good reproducibility (relative standard deviation, %RSD, of <5 % for n=3).

2 Result and Discussion

2.1 Electrocatalytic Activity of Nanocubes towards Methylene Blue (MB)

The synthesis and detailed characterization of AuNP-Fe₂O₃NC nanocubes (SEM, wide-angle XRD patterns, elemental mapping images for O, Fe, and Au and EDX spectrum) have been reported in our earlier paper.21 The AuNP-Fe₂O₃NC was synthesized by deposition of AuNPs on to the porous Fe₂O₃ nanocubes. The porous iron oxide nanocubes (Fe₂O₃NC) was prepared from Prussian blue (PB) nanocubes via calcination of PB at 250 °C. After deposition of AuNPs, uniformly sized AuNPs (∼3–6 nm) are dispersed on the surface of Fe₂O₃NC. The loading amount of AuNPs was around 2 wt % in the product. The nanocube was found to be superparamagnetic from the complete reversibility of the M–H curve recorded at room temperature (300 K). The presence of gold within the porous structure is highly desirable as it enables easy and simple functionalization of the AuNP−Fe₂O₃NC with miRNA via gold-miRNA affinity interactions.10, 11

To achieve an ultrasensitive detection, nanostructured materials should have superior conductivity, high catalytic activity, biocompatibility and a synergetic effect among them to accelerate the signal transduction towards the electrocatalytic reduction of target-bound or intercalated redox molecules.11 To assess the electrocatalytic activity of AuNP-Fe₂O₃NC, AuNP-Fe₂O₃NC modified GCE is directly used as a working electrode for electrochemical detection of electroactive redox marker MB in 0.01 M PBS (pH 7.0). It has been shown from Figure 1(a), AuNPs-Fe₂O₃NC modified-GCE (GCE/AuNP-Fe₂O₃NC) displayed a pair of well-defined redox peaks in 50 μM MB solution at −241 mV and −272 mV (vs. Ag/AgCl), indicating a two-electron redox process of MB (ΔE=31 mV, Figure 1a). The GCE/AuNP-Fe₂O₃NC offered significantly enhanced cathodic (i_pc), and anodic (i_pa) peak currents in comparison with the GCE/bare electrode (Figure 1a). Notably, i_pc increased approximately 2.8-times (7.50 vs. 18.5 μA cm⁻²) with E_pc shifted by −19 mV, whereas i_pa increased approximately 2.8-times (7.5 vs. 21 μA cm⁻²) with E_pa moved by around −27 mV. The very similar current response was also obtained from differential pulse voltammetry (DPV) measurements (23.2 vs. 71.5 μA cm⁻²; Figure 1b). These data indicate that AuNP-Fe₂O₃NC catalysed both the oxidation and reduction process of MB.

Figure 1

Electrocatalytic activity of Au-Fe₂O₃NC in MB redox system. Comparison of the CVs (a) and DPV (b) obtained at an unmodified GCE and AuNP-Fe₂O₃NC-modified GCE in 50 μM MB (scan rate, 50 mVs⁻¹). CV obtained at GCE/AuNP-Fe₂O₃NC-electrodes at different scan rate (50 μM MB, 0.01 M PBS, pH 7.0) (c); Corresponding curves for i_pc and i_pa (current density) as a function of ν^1/2 (d).

Regarding the electrochemical mechanism of nanocube catalysis (diffusion or adsorption controlled) that take place on the GCE electrode surface, we performed CV measurements of both bare and GCE/AuNP-Fe₂O₃NC modified GCE (Figure S1a in Electronic supplementary Information (Supporting Information) as a function of different scan rate (ν). Both i_pc and i_pa are proportional at the scan rates values (10–1500 mVs⁻¹) and showed a linear relationship with ν^1/2 for both the bare and AuNP-Fe₂O₃NC modified GCE, which indicates the electrode process is diffusion-controlled (Figure. 1c and d)31 and the equations can be expressed as, i_pc (μA)=0.9404 ν^1/2 (mVs⁻¹)−2.7240, r²=0.97 for bare and i_pa (μA)=-0.6443 ν^1/2 (mVs⁻¹)+3.5644, r²=0.9714 for AuNP-Fe₂O₃NC modified GCE. Moreover, the plot of i_pa and i_pc versus ν^1/2 for AuNP-Fe₂O₃NC modified GCE (Figure S1e and f in Supporting Information) showed a steeper slope than that of the bare GCE (Figure S1b and c in Supporting Information), verifying the catalytic performance of AuNP-Fe₂O₃NC towards the redox reaction of MB. However, the slope for i_pc versus ν^1/2 was exhibited steeper than that of i_pa versus ν^1/2, suggesting the AuNP-Fe₂O₃NC has higher catalytic activity towards reduction of MB than that of oxidation. To further examine the performance of AuNP-Fe₂O₃NC, we conducted the CV measurement at different concentration of MB (0 to 200 μM) (Figure S2 in Supporting Information). With increasing strength of the MB, the peak currents increase gradually, reaching saturation at 150 μM MB (Figure S2), revealing that the porous structure of AuNP-Fe₂O₃NC (Figure S3) provides an enormous surface to accelerate the redox reaction.

CV and DPV responses of AuNP-Fe₂O₃NC modified SPCE was also studied using 2 mM [Fe(CN)₆]³⁻ system. As shown in Figure S4a (Supporting Information), the nanocube modified SPCE generated a CV with two redox peaks with the j_pa and j_pc of 178 μC cm⁻² and 181 μC cm⁻² respectively. The electrode also generated a DPV redox peak current of 178 μC cm⁻² (Figure S4b in Supporting Information).

2.2 Electrocatalytic Detection of miRNA

The principle of our miRNA assay is schematically depicted in Figure 2. To demonstrate the working principle of the developed sensor, we initially extracted the total small RNA from the exosomes obtained from preconditioned media of motor neurons (healthy controls and ALS patients). To isolate the specific miRNA target from this bulk RNA pool, we designed a biotinylated capture probe complementary to the target miR-338-3p sequence and used a magnetic bead-based capture and isolation protocol outlined in our previous report.21 After magnetic isolation, the target miR-338-3p was released from the surface of the magnetic beads via heating. The released miR-338-3p was directly adsorbed on to the AuNP-Fe₂O₃NC-modified SPCE via RNA–gold affinity interaction, which follows conventional physisorption and chemisorption mechanism through the direct interaction of nitrogen atoms of nucleobase ring's with gold and the partial contribution from the exocyclic amino group and charge transfer between the aromatic ring and gold surface.11, 21 The profound understanding of the mechanism underlying adsorption of single-stranded nucleic acid (miRNA) was provided by the landmark studies from Mirkin, Tarlov and Rothberg group.32-35 The AuNP-Fe₂O₃NC is magnetically attached to the SPCE surface through the magnetic interaction between a static magnetic disc (placed under the SPCE) and the superparamagnetic nanocubes. This allows us to reuse the SPCE by removing the static magnet which ultimately removes nanocubes from the electrode surface. The adsorbed miRNA was detected by CC interrogation in the presence of MB. It has been shown that MB can bind with DNA sequences in at least three different manners; i) specific binding between MB and guanine bases,36 ii) intercalation of MB in the DNA double helix,37 and iii) electrostatic interaction between anionic DNA and cationic MB.38 Herein, MB cations act as a signaling molecule that stoichiometrically binds to the guanine bases of miRNA and quantitatively indicates the amount of miRNA localized at the electrode surface. In this method, the nanocube (AuNP-Fe₂O₃NC) offers ultra-sensitive detection through the catalytic properties of mesoporous Fe₂O₃NC and greater adsorption of miRNA onto AuNPs of the nanocube, which can provide pico-molar level detection of miRNA.21

Figure 2

Schematic representation of the assay. Exosomal RNA was isolated from exosome that derived from preconditioned media of motor neurons (ALS). Target miR-338-3p was hybridized with a biotinylated capture probe followed by the addition of streptavidin-magentic beads for magnetic isolation and purification. After magnetic purification, target miRNA were released from magnetic beads and allowed to adsorb onto the gold surface of AuNP-Fe₂O₃NC/SPCE electrode for electrochemical detection. The presence of adsorbed miRNA interact with positively charged MB⁺ redox molecules, and the amount of charge intercalation is measured by CC readout in the presence of 4 mM [Fe(CN)₆]³⁻ in 40 mM Tris-HCl buffer (pH 7.4). In this electrocatalytic redox cycling process, electrons flow from the electrode surface to intercalated MB⁺ in miRNA, resulting in the formation of leucomethylene blue (LB⁺, the reduced from of MB⁺). LB⁺ then reduces [Fe(CN)₆]³⁻ in solution and thereby regenerating MB⁺ catalytically (redox-cycling).

It is noteworthy to mention that some clinical samples may contain a trace amount of target miRNA (probably less than pico-molar), which impulses the need for an ultrasensitive assay. To obtain such an assay, we used electrocatalytic redox cycling based on MB and [Fe(CN)₆]³⁻ redox system (shown in equation 1 and 2). In this process, electrons flow from the electrode surface to intercalated MB⁺ in miRNA resulted in the formation of LB⁺ (reduced form of MB⁺) (eqn 1). LB⁺ then reduces ferricyanide in solution, thereby regenerating MB⁺ catalytically (eqn 2), and the catalytic cycle continues.39, 40 This catalytic redox cycling allows multiple turnovers of MB⁺ resulting in a large increase in the electrochemical signal.41 Previously Boon et al. have shown that the negatively charged ferricyanide gave no CV response at the DNA-modified electrode. However, in the presence of a micro-molar concentration of MB⁺, the electrode showed a significant peak current similar to that expected for the reduction of ferricyanide at the diffusion-controlled limit.42 Therefore, the amount of intercalated/attached MB within the surface-bound miRNA appears to be a key element in our redox-cycling amplification process and the CC charge generated by MB and [Fe(CN)₆]³⁻ system should have a clear correlation with the concentration of miRNA.

(1)

(2)

2.3 Assay Functionality and Specificity

To check the assay functionality and specificity, the target miR-338-3p and a series of control miRNA sequences were analyzed. As can be seen from Figure 3, the bare SPCE generated a negligible amount of charge. This can be explained by the fact that there was no electrocatalytic AuNP-Fe₂O₃NC or miRNA on the electrode surface and hence negligible amount of MB⁺ present on the surface. The control samples (PBS were used instead of miRNA) generated 5.2 μC cm⁻², whereas the Q_dl (absence of MB⁺) generated 4.6 μC cm⁻² of charge. Over two-fold charge response for the control sample in comparison to that of the bare SPCE was due to the electrocatalytic effect AuNP-Fe₂O₃NC. The slight reduction of charge generation of Q_dl than that of the control sample could be due to the blockage of some catalytic sites by the adsorbed miRNA. Notably, no additional charge from the adsorption of target miRNA was observed. This is because in the absence of MB⁺, no intercalation occurred and hence no charge was associated from adsorbed miRNA. The assay was also tested with miR-21 (named as wrong target-WT) which sequences mismatched with the target miR-338-3p and non-complementary to the capture probe used for miR-338-3p isolation and purification. Like control, a very similar response (4.4 μC cm⁻²) was attributed for WT sequence of miR-21 (a 22 bp miRNA). This is because no miR-338-3p were isolated and hence no adsorption of the target onto the surface of AuNP-Fe₂O₃NC has occurred. However, a minuscule increase of charge was due to adsorption of the small amount of MB⁺ onto the AuNP-Fe₂O₃NC surface. A substantial amount of charge (4.6 vs. 16.1 μC cm⁻²) have been observed when the same amount (10 pM) of miR-338-3p was interrogated onto the sensor surface, indicating the functionality and specificity of the sensor. This is because a significant amount of target miR-338-3p were isolated via the specific-capture probe and adsorbed onto the AuNP-Fe₂O₃NC/SPCE surface. The intercalated MB⁺ within the surface-bound miR-338-3p resulted in a high level of CC charge. As expected, a further enhancement in CC charges resulted in the presence [Fe(CN)₆]³⁻ system. With the same amount of starting miRNA (10 pM), the redox cycling system offered almost 3-times of higher CC charges (45.16 vs. 16.1 μC cm⁻²), which is due to the multiple turnovers of MB⁺, demonstrating the high specificity of our assay. We envisage that along with high loading capacity and electrocatalytic activity of AuNP-Fe₂O₃NC (towards MB⁺, and/or MB⁺/[Fe(CN)₆]³⁻) a few others features attributed this high specificity of our assay; (i) magnetic bead-based purifications deliver purified targets miRNA to the sensor surface, (ii) the direct adsorption of miRNA on to the sensor surface revoke the tedious surface chemistry for nucleic acid immobilization and detection, (iii) the clinical (plasma or serum) or spiked samples are spatially separated from the sensor surface, thereby avoids the adsorption of unwanted-species (e. g. proteins, non-targeted nucleic acids).

Figure 3

Assay specificity. CC (inset) charge density for the assays obtained for different target sequences (SPCE/miR-338-3p, SPCE/NC/miR-338-3p (control), SPCE/NC, SPCE/NC/miR-21 (wrong target), SPCE/NC/miR-338-3p/MB in 40 mM Tris-HCl buffer (pH 7.4) and SPCE/NC/miR-338-3p/MB in 4 mM [Fe(CN)₆]³⁻ in 40 mM Tris-HCl buffer (pH 7.4) for redox cycle. Concentration of miR-21 and miR-338-3p were 10 pM (NC: AuNP-Fe₂O₃NC).

2.4 Optimization of the Experimental Parameters and Stability of the Sensor

To maintain the physiological pH, we carried out all experiments using the buffer solutions of pH of 7.4. In order to optimize the amount of AuNP-Fe₂O₃NC nanocubes and incubation time required for microRNA adsorption in sensor design, we have measured the sensor response for 10 pM of miR-338-3p. As can be seen from Figure S5 (a), the sensor-generated negligible CC charges (Q_miRNA signal) when it was prepared using 1 mg of nanocubes. The CC charge increased with increasing the amount of nanocube, where a plateau obtained for >5 mg of nanocubes. To minimize the use of nanocube amount, we have chosen 5 mg of nanocube as optimal amount for designing the sensors for all subsequent experiments. For optimizing the miRNA incubation time, the 10 min incubation provided negligible Q_miRNA response. The response increased with increasing the incubation time up to the 30 min (Figure S5b) and reached a plateau region. At a longer time, more microRNA can be adsorbed onto the sensor surface, resulting in enhanced CC interrogation and thus higher responses. Although >30 min incubation generated higher responses, we have used 30 min incubation time as optimal for all subsequent experiments (to minimize the overall assay time). In order to check the stability of the sensor response, we have taken multiple CC readout of the SPCE/AuNP-Fe₂O₃NC/miRNA (10 pM) electrode using the MB/[Fe(CN)₆]^3−/4− redox cycling system. It has been shown that the sensor generates almost the same charge for over 10-individual runs (Figure S6).

2.5 Assay Sensitivity

To evaluate the assay sensitivity, different concentration of miR-338-3p (synthetic probe spiked in 0.01 M PBS) ranging from 1 μM to 10 pM were initially detected without [Fe(CN)₆]³⁻ redox cycling. It has been observed that the charge generated by MB intercalation was augmented with increasing concentration of miRNA. This was attributed to the higher amount of miRNA was isolated and thus adsorbed on to the AuNP-Fe₂O₃NC-attached SPCE surface. An increased amount of adsorbed miRNA contains a relatively higher amount of negatively charged phosphate backbone, thus binding with a greater number of positively charged MB⁺ intercalator and thereby, generating higher charge (Figure 4a and c). The level of engendered redox charge indicates that our inexpensive and reusable AuNP-Fe₂O₃NC-attached SPCE can detect 10 fM miRNA with higher selectivity and specificity.

Figure 4

Assay sensitivity. CC curves showing amount of charge generated for the different concentration of starting miR-338-3p targets before (a; a-control; b to g-10 fM to 1 nM) and after (b; a-control; b to i −100 aM to 1 nM) coupling with [Fe(CN)₆]³⁻ (redox cycle). The figure, c and d show the analogous bar diagrams (inset; linear calibration plots) before (c) and after coupling with [Fe(CN)₆]³⁻ (d).

To achieve the ultra-low level of detection, we have coupled our assay with [Fe(CN)₆]³⁻ redox system. The MB⁺/[Fe(CN)₆]³⁻ redox cycling resulted in boosted charge amplification, which consequences in the detection of 100 aM level of miRNA (Figure 4 b and d). This redox cycling ensuing in the sharp upsurge in the electron flux produces an enhanced CC readout. The response for 100 aM miR-338-3p was clearly distinguishable from the controls. By using the signal-to-noise ratio (S/N) of 3.0, we have estimated the limit of detection (LOD) of our assay is 100 aM with a dynamic range from 100 aM to 1 nM. It is noteworthy to state that before redox cycle (only MB⁺), the assay can detect up to 10 fM, and indicating 100-fold more sensitive detection of miR-338-3p via redox cycling (Figure 4). The analytical performance of our assay is highly comparable with that of the recently reported nanotechnology (nanowire, carbon nanotube) based strategies for mRNA detection.11 For instance, the LOD of our assay is about 10-times more sensitive than recently reported core-satellite structure of magnetic nanobeads-quantum bead-based fluorescent assay.43 Our assay also showed higher sensitivity compared to that of the recently reported nanostructure-based strategies44-53 (details comparison are shown in Table S2, in Supporting Information). Most of these assays relied on cumbersome enzymatic amplification processes, displacement reactions, multi-step cascade electrocatalysis or different redox systems. On the contrary, our assay design is relatively simple (non-enzymatic, external magnet-based AuNP-Fe₂O₃NC attachment, direct adsorption of target miRNA, electrocatalytic redox cycling signal amplification), inexpensive (reusable AuNP-Fe₂O₃NC-modified SPCE, easy synthesis of AuNP-Fe₂O₃NC), and enzymatic amplification- or label-free. We believe that the LoD of our assay (100 aM) is suitable to quantify the level of miRNA biomarker present in samples collected from patients with different stages of various diseases including MND.

2.6 Analysis of miR-338-3 p in Exosomes from ALS Patient and Healthy Control Motor Neurons

To examine the applicability of this assay for clinical (patient) samples of ALS, we interrogated our method in exosomal miRNA sample extracted from preconditioned media of motor neurons (ALS). Exosomes are nanovesicles consisting of different functional proteins and genetic materials, e. g. mRNA and miRNA, which are secreted from many types of cells.54 Exosomes work as an intercellular communication vesicle, and they guide axonal development, modulate synaptic activity in the nervous system and help to regenerate peripheral nerve tissues.29 Xin et al. testified that exosomes derived from mesenchymal stem cell (MSC) promote neurogenesis, neurite remodeling, and functional recovery after stroke. Furthermore, they demonstrated that the transfer of miR-133b from MSCs to neurons and astrocytes via MSC-derived exosomes promotes neurological recovery after stroke.30 miRNA is also known to affect cell growth and direct differentiation of stem cells into many types of cells, including neurons.55

In this proof of concept assay, we have chosen mir-338-3p, which was extracted from exosomes obtained from preconditioned media of motor neurons. It has been reported that miR-338-3p is over-expressed in blood, CFS, serum and spinal cord from sporadic ALS patients and has been considered as a potential biomarker for ALS diagnosis^.[15,54] Herein, we have analyzed two healthy patient samples (named control; C1 and C2) and three ALS samples (P1, P2 and P3) in 2 and 4 weeks of motor neuron differentiation. As can be seen in Figure 5 (a), our assay enabled detection of miR-338-3p from both healthy and patient-derived motor neuron exosomes. The data showed that the miR-338-3p levels were expressed differently in different samples (Figure 5a). With the same amount of starting cell culture media, our method detected the lower expression of miR-338-3p in one of the ALS samples (P1) compared to those of two controls and two ALS samples. From the standard curve (shown in Figure 4d), the concentration of miR-338-3p estimated to be as; P1 (for W2 and W4) - less than 100 aM (out of assay range); P2–5.5 pM (W2) and 0.87 (W4) pM, and P3–1.1 pM (W2) and 0.87 (W4) pM. Figure 5b represents the adjusted p values obtained using two-way ANOVA comparing the control (healthy) samples and ALS patient samples. We validated our proof-of-assay performance with a standard RT-qPCR method (Figure 5), which shows consistent results with the electrochemical measurements. However, control C1 and ALS sample P3 shows no significant difference in expression of miR-338-3p. In future studies, we will address these differences and evaluate the effects of time during differentiation with miR-338-3p as well as the combination of multiple miRNAs on ALS diagnosis.

Figure 5

Assay performance on clinical samples. a) The electrochemical signal obtained from sample C1, C2, and P1 to P3 in week 2 and week 4 intervals, where C1 and C2 represent the motor neuron samples from healthy patients and P1, P2 and P3 represent motor neuron samples from ALS patients. b) RT-qPCR validation of exosomal miR-338-3p expression levels in the two healthy control samples and three ALS samples over the two and four weeks of motor neuron differentiation. Below: Corresponding p values obtained using two-way ANOVA comparing the control (healthy) samples and ALS patient's samples.

3 Conclusions

A simple and inexpensive method has been reported for detecting the ultra-low level of exosomal miRNA obtained from preconditioned media of motor neurons of ALS patients and healthy control samples. The target miRNA was directly adsorbed onto the AuNP-Fe₂O₃NC/SPCE surface via gold-miRNA affinity interaction, and the amount of adsorbed miRNA was quantified via chronocoulometric (CC) charge interrogation of the target-bound MB. The signal was amplified by the synergic effect of the following three: (i) AuNP-Fe₂O₃NC-modified SPCE electrode provides electrocatalytic signal amplification (through the catalytic redox reaction of MB) (ii) specific binding capacity of redox intercalator MB with surface-attached target miRNA, provides precise signals, and (iii) the electrocatalytic redox cycling (MB/[Fe(CN)₆]³⁻) boosted the catalytic signal amplification. The method is relatively fast (<2.5 h), specific, sensitive, and potentially applicable to a wide variety of miRNA detection. This method offers some additional advantages- (i) direct adsorption of target miRNA on an AuNP-Fe₂O₃NC/SPCE surface substantially simplifies the overall sensing approach by avoiding the use of complicated fabrication processes involved in each step of the recognition and transduction layers of conventional nucleic acid biosensors, (ii) use of magnetic bead-based isolation and purification steps reduce the matrix effect as non-targets species can be removed via magnetic washing and thus could enhance the isolation purity and efficiency, (iii) disposable screen-printed electrode for detection at a relatively low cost (AUD$3 per electrode), and (iv) electrochemical detection that can complement with the portable and inexpensive detection platform, and thus decidedly potential to translate the method into a simple and affordable screening of miRNA in biological samples. We foretell that the assay could also be radially extended to other miRNA marker detection through the further design of nanostructures and adaption of capture probe sequences.

Experimental Section

Cell Culture and Cell Culture Media Collection

All experimental protocols were approved by the University of Wollongong Human Research Ethics Committees (the human ethics approval number is HREC13/272). The methods were carried out by the guidelines as set out in the National Statement on Ethical Conduct in Research Involving Humans, and informed consent was obtained from all donors. Skin samples were collected from two healthy individuals and three ALS patients, and human feeder-free iPSCs were generated as previously described.27, 28 The iPSCs were confirmed pluripotent by Pluritest and were karyotyped to verify the lack of chromosomal changes during reprogramming. The iPSCs were cultured on Matrigel (Corning) coated tissue culture plates in TeSR−E8 (Stem Cell Technologies) at 37°C, 5 % CO₂ in a humidified incubator. Differentiation of iPSCs to motor neurons was performed as previously described.29, 30 Using cell scrapers, iPSC colonies were detached from the plate and cultured in a non-tissue culture plate for 4 days in neural induction media to form embryoid bodies. (Please see details in Supporting Information).

Isolation of Exosome and Exosomal (Total) Small RNA

Total exosome was obtained from motor neuron preconditioned media using total exosome isolation reagent (Invitrogen) following manufacturers guidelines. The RNA was then extracted from exosomes using the total exosome RNA and protein isolation kit (Life Technologies, Australia) as per the manufacturer's instructions. Briefly, the exosome was diluted with 10 mM PBS to a total volume of 200 μL followed by the addition of the same amount of 2× denaturing solution. To lyse the exosomes, acid-phenol: chloroform was added and vortexed vigorously for a minute. The mixture was then centrifuged at 12000 g for 5 min at the room-temperature to split up the mixture into aqueous (upper) and organic (lower) phases. The upper phase was then subsequently treated with ethanol to dissolve and remove larger RNA and DNA present in the sample. The loop of small miRNA was eluted from filter cartridge and stored −20 °Ϲ for further analysis.

Electrocatalytic Detection of Adsorbed microRNA

The nanocube adsorbed miRNA was quantified by chronocoulometric measurements in absence and presence of 2.0 μM MB with a potential step of 5 mV, a pulse width of 250 ms, and a sample interval of 2 min. Using CC, the amount of miRNA adsorbed onto the AuNP-Fe₂O₃NC/SPCE surface was calculated from the number of cationic redox molecules (MB) electrostatically associated with the surface-attached anionic phosphate backbone of miRNA (Please see details in Supporting Information).

Acknowledgements

This work supported by Australia-Korea Foundation (AKF) Grant (AKF2018043) to M.J.A.S. and Y.B.S., Australian Research Council (ARC) Discovery Project (DP190102944) to Y.Y, M.J.A.S and M.S.H. This work was partly supported by the ARC Future Fellow (FT150100479) to Y.Y, a National Health and Medical Research Council (NHMRC) Boosting Dementia Research Leadership Fellowship (APP1135720) to L.O. and Australian Government Research Training Program (RTP) Scholarship (the University of Queensland) to M.K.M. M.K.M. would like to thank AINSE Limited for providing financial assistance (AINSE PGRA Award 2018). This work was performed in part at the Queensland node of the Australian National Fabrication Facility (ANFF−Q), a company established under the National Collaborative Research Infrastructure Strategy to provide nano- and micro-fabrication facilities for Australia's researchers.

Conflict of interest

The authors declare no conflict of interest.