Choice of materials
Carbon nanodots (CNDs) have been chosen for the development of efficient DNA ECL biosensors, due to their biocompatibility and their property to act as co-reactants for the anodic ECL of [Ru(bpy)3]2+ .
Synthesis and characterization of carbon nanodots
Photoluminescent CNDs were synthesized by carbonization of tiger nut milk using a microwave reactor synthesizer, as shown in Scheme 1 and explained in detail in the “Experimental” section. The temperature and reaction time were optimized using emissions from synthesized CNDs (see Fig. S1). Best results, considering the balance of speed and performance, were obtained with a reaction time of 30 min at a temperature of 200 °C, maintaining a constant pressure of 170 psi and a power of 150 W.
Hydrothermal treatments usually require carbonization for up to 18 h . However, only 30 min is required with microwave radiation. Increasing the time of exposure to microwave radiation produces an increase in the emission intensity of the CNDs, which is more evident between 5 and 30 min (Fig. S1B). After 30 min, emissions remain almost constant, so this has been selected as the optimal synthesis time in terms of the fluorescence of the resulting product and the speed of the process.
The obtained CNDs were characterized by different techniques to elucidate their morphology, composition, and properties. Elemental analysis revealed that the CNDs contain 47.19% C, 5.75% H, 1.05% N (see Fig. S2), and 46.01% O (calculated). Zeta potential measurements gave an average value in aqueous solution of − 6.7 ± 0.3 mV. This negative value of zeta potential is probably due to the surface‐contained carboxylate groups on the CNDs.
The morphology and dimensions of the CNDs were determined by TEM. As shown in Fig. 1A, CNDs have a quasi-spherical morphology with diameters distributed in a range from 4 to 9 nm and an average size of 6.8 ± 1.5 nm (n = 35) (see histogram of Fig. S3).
The UV–visible molecular absorption spectrum of the CNDs is shown in Fig. 1B. It can be seen that the CNDs have an absorption band at 284 nm that can be assigned to n → π* transitions from C = O groups and π → π* transitions from C = C groups . This feature is characteristic of CNDs synthesized by carbonization of carbon-based materials  and demonstrates the presence of carbonyl groups on the surface of the synthesized CNDs. The fluorescence emission spectrum recorded by exciting CNDs at the maximum excitation peak (380 nm) presents a maximum at 478 nm (Fig. 1B). The emission peak decreases and shifts its position to a higher emission wavelength (from 471 to 550 nm) as λex moves from 300 to 500 nm (∆λex = 200 nm) (see Fig. S4). The redshift of the fluorescence emission can be attributed to differences in particle sizes and in the distribution of emissive sites on the surfaces of the CNDs. Smaller particles would be excited at shorter wavelengths than larger ones . The fluorescence micrograph (Fig. 1C) again demonstrates the fluorescence of the CNDs, which appear as light spots.
The high carbon and oxygen content suggests that the particles obtained are nanometer-sized carbonaceous materials with a large number of carboxylic groups on the surface , a fact that can be confirmed by the FTIR spectrum (Fig. 1D). As can be seen, the spectrum exhibits the characteristic stretching band of OH and NH vibrations at 2900 and 3400 cm−1; a band located at 1650 cm−1 and attributed to C = O stretching vibration confirms that − COOH groups are present, explaining the high-water solubility of the CNDs. The band present at 2900 cm−1 can be assigned to C–H tension vibrations, and those above 1050 cm−1 are assigned to C–O and C–O–C bonds. The band at 1400 cm−1 probably corresponds to C–N or C = C vibrations. Therefore, carbonyl groups, alcohols, and amines are present in the synthesized CNDs.
X-ray photoelectron spectroscopy (XPS) studies were carried out to corroborate the results obtained from FTIR spectroscopy. The results revealed the presence of carbon, oxygen, nitrogen, phosphorus, and sodium, with the following percentage atomic concentrations: 64.4% C, 31.3% O, 3.9% N, 0.2% P, and 0.2% Na (Fig. S5A). Peak fitting of the C1s core level region (Fig. S5B) showed peaks at C = C (284.3 eV), C–C/C–H (284.8 eV), C–O/C–N (286.0 eV), C = O (287.6 eV), O–C = O (288.8 eV), and the characteristic π → π* vibrations of carbon atoms in graphene-like structures (292.4 eV). The O1s line fitting for the CNDs (Fig. S5C) shows two different chemical components centered at 531.5 eV (O = C) and 532.5 eV (O–C). The N1s spectrum (Fig. S5D) shows two peaks at 399.9 eV and 401.5 eV, which were attributed to the presence of C–N and N–H, respectively. The CND surface components determined by X-ray photoelectron spectroscopy are in good agreement with FTIR results.
The stability of the CNDs in aqueous solution was also studied following their fluorescence spectra. The results show that emission intensity remains constant for at least 30 days and then begins to decrease significantly (Fig. S6). We have also checked the reproducibility of the CND synthesis method. For this, the synthesis procedure has been repeated several times and no significant differences have been observed in the corresponding batches obtained, other than small differences in the concentration of CNDs. As we describe in detail in the “Electronic supplementary.material (ESM)” section, from the results obtained by elemental analysis, TEM, and UV–visible spectroscopy, we estimated the concentration of the synthesized CNDs. It was found to be of 400 µM, using the calculated molar extinction coefficient of 1.82 × 106 M−1 cm−1 (see ESM and Fig. S7).
CNDs have been previously used as co-reactants in the anodic ECL of [Ru(bpy)3]2+ [27, 53]. To evaluate whether these synthesized CNDs could act as co-reactants in the anodic ECL of [Ru(bpy)3]2+, we recorded cyclic voltammograms of the CNDs, [Ru(bpy)3]2+, and [Ru(bpy)3]2+/CNDs in 0.2 M PB, pH 8.0, at an AuSPE. As can be seen in Fig. 2A, [Ru(bpy)3]2+ shows the characteristic redox process ascribed to the [Ru(bpy)3]3+/[Ru(bpy)3]2+ system, and the CNDs show no voltammperometric peaks within the potential window of the experiment. However, in presence of both CNDs and [Ru(bpy)3]2+, there is an electrocatalytic process, since the intensity of the [Ru(bpy)3]2+ oxidation peak increases concomitantly as the cathodic peak disappears; this was more evident on increasing the CND concentration from 40 to 70 µM (40, 60, and 70 µM) when a fixed concentration of 7.0 mM of [Ru(bpy)3]2+ was used.
Figure 2B shows that the ECL response of the [Ru(bpy)3]2+/CNDs mixture produced a strong ECL signal, which is consistent with that observed by cyclic voltammetry, whereas [Ru(bpy)3]2+ or CNDs alone exhibited weak ECL signals. Moreover, as observed by cyclic voltammetry, the ECL [Ru(bpy)3]2+ signal increased on increasing the concentration of CNDs.
We believe that CNDs act as co-reactants in the [Ru(bpy)3]2+/CNDs ECL system, being converted into reductive intermediates by the chemical oxidation of oxygen-containing units present on their surfaces by electrogenerated [Ru(bpy)3]3+ , similar to the role played by TPrA in the anodic ECL of the [Ru(bpy)3]2+/TPrA system . Therefore, we propose the following typical “oxidative-reductive” co-reactant pathway mechanism for the [Ru(bpy)3]2+/CNDs ECL system:
[Ru(bpy)3]2+ and the CNDs are oxidized at the electrode surface to [Ru(bpy)3]3+ and the radical [CNDs], respectively. Then, [CNDs] reduces [Ru(bpy)3]3+ to [Ru(bpy)3]2+*, which is unstable and decays to the ground state, emitting a red light.
To investigate why CNDs assist in the anodic ECL of [Ru(bpy)3]2+, we obtained the UV–visible absorption and photoluminescent emission spectra of solutions of CNDs, [Ru(bpy)3]2+ and CNDs + [Ru(bpy)3]2+. The CNDs + [Ru(bpy)3]2+ mixture has the same absorption spectrum as [Ru(bpy)3]2+, with a negligible contribution from the CNDs (Fig. S8A). Something similar occurs in the case of the photoluminescent spectra (Fig. S8B), where [Ru(bpy)3]2+ and the CNDs maintain their original photoluminescence in the mixture. These results suggest that no reaction takes place between excited-state or ground-state CNDs and [Ru(bpy)3]2+ . Hence, the light emission must be due to a reaction between the electrogenerated [Ru(bpy)3]3+ and the CNDs. [Ru(bpy)3]2+ is identified as the emitter in the mixture; therefore, the CNDs must be the co-reactant. Consequently, the ECL signal of [Ru(bpy)3]2+ increases on increasing the concentration of CNDs from 40 to 70 µM and then level off (Fig. 2B) from the inner filter effect of CNDs . Based on these results, 70 µM of CNDs and 7.0 mM [Ru(bpy)3]3+ were chosen as optimal. Moreover, different pH values were also studied; the ECL signal of [Ru(bpy)3]3+ in 0.2 M PB increases with the pH up to 9.0. However, since biological samples will be used, pH 8.0 was chosen as more adequate. Thus, 70 µM of CNDs and 7.0 mM [Ru(bpy)3]3+ in 0.2 M PB, pH 8.0 were chosen as optimal conditions for further experiments.
Electrochemiluminescent miRNA-21biosensor development
Based on the results described above, using the [Ru(bpy)3]2+/CNDs system, we developed an ECL DNA biosensor for the sensitive detection of miRNA-21 (Scheme 1). The [Ru(bpy)3]2+/CNDs ECL system was used to carry out the transduction that allows the biomarker miRNA-21 to be detected and quantified. Scheme 1 shows the different steps taken to design the biosensor.
As explained in the “Experimental” section, the first step is the immobilization of the miRNA-21-SH probe through the chemisorption of thiols on the surface of the AuSPE. In order to obtain a compact and standed-up thiolated probe monolayer, we have performed the immobilization either by direct adsorption of the capture probe alone for long time or employing mercaptohexanol (MCH) in addition to the thiolated probe. As was the case in previous works reported by us , best results were obtained when the thiolated probe was directly deposited on the electrode surface for long time (24 h). With this method, the hybridization effectiveness is higher since we observe the maximum difference between the ECL signal before and after hybridization of the probe with the target miRNA. These results agree well with those previously reported by us and confirm that immobilization time plays an important role in the self-assemble of the capture probe monolayer.
Our method allows also to determine the amount of thiolated oligonucleotide immobilized on the gold surfaces from the coulometry charge associated with the desorptive reduction of the miRNA-21-SH monolayer, as previously described for alkane thiols . The surface coverage was calculated as 2.8 ± 0.2 × 10−8 mol oligonucleotide/cm2.
Next, hybridization of the probe with the complementary analyte sequence (10.0 pM miRNA-21C solution) was performed on the electrode surface. Experimental parameters used, such as buffer, pH, and ionic strength for hybridization, are explained in detail in the “Experimental” section.
Finally, the hybridization event was detected via changes in the signal from the [Ru(bpy)3]2+/CNDs ECL system, after applying a cyclic potential sweep from 0.00 to + 1.30 V at 10 mVs−1. The resulting ECL signals were plotted as a function of time and compared to the bare electrode ECL signal (a). As can be seen in Fig. 3A, the biosensor response before hybridization (b) is less intense than that observed after hybridization with the fully complementary sequence (c). Specifically, hybridization of the probe produces an increase in the ECL signal of approximately 205 a.u. This can be explained by the formation of a stable helical conformation after hybridization, which enhances the electron-transfer reaction compared to the unstructured single-stranded unit. Therefore, these results demonstrate that the [Ru(bpy)3]2+/CNDs system used in the ECL technique can detect hybridization between oligonucleotides.
As a comparison, the same experiments were carried out using the [Ru(bpy)3]2+ complex alone (in the absence of CNDs). In this case, the ECL spectrum (Fig. 3B) shows that, after modifying the AuSPE with DNA, the signal increases slightly compared to that of the bare electrode. However, the difference in signals from the probe before and after hybridization with the complementary sequence is so small that it is difficult to discriminate between them.
From the above results, we concluded that the synthesized CNDs act as co-reactants for the ECL signal of [Ru(bpy)3]2+, improving the sensitivity of the method. Furthermore, in the presence of CNDs, there is a significant difference between the signals obtained before and after hybridization, which is essential for the development of a sensitive DNA biosensor.
After verifying the ability of the developed biosensor to detect the specific sequence of miRNA-21C, we studied its response to the biomarker at concentrations from 2.34 fM to 100.0 pM. The ECL signal increases linearly on increasing the target concentration, as can be seen in Fig. 4A. The plot of biosensor response versus log [miRNA-21C] fits the linear equation ECL = 29.1 log [miRNA-21C] + 722.3 (R = 0.9996). The method has a sensitivity of 29.1 a.u. log pM−1, obtained from the slope of the plot (Fig. 4B). The detection and quantification limits were found to be 0.721 (S/N = 3) and 2.34 (S/N = 10) fM, respectively.
Compared to other reported methods for the determination of miRNA-21, the ECL biosensor developed in this work exhibited a lower detection limit (Table 2). It also compares well with previously reported ECL biosensors and has much a wider linear concentration range. Furthermore, the method is simple, without the need for complex approaches or labeling steps.
The selectivity of the ECL biosensor was tested using a non-complementary sequence (miRNA-21NC) and a single mismatched sequence (miRNA-21SM). In addition, in order to consider the high homology between miRNA families, the biosensor response to a miRNA144 and miRNA-155 sequence was tested (see Fig. S9). For this purpose, ECL biosensor responses were recorded before and after hybridization with a 100.0 pM solution of a non-complementary (miRNA-21NC), a single-mismatched (miRNA-21SM), interferent sequences (miRNA144 and miRNA-155), or a totally complementary sequence (miRNA-21C) used as control. When the probe is hybridized with the complementary sequence, an increase in ECL signal of around 205 ± 8 a.u. is observed. However, when hybridization takes place with the mismatched sequence, the increase in the ECL signal is lower (around 125 ± 7 a.u.). The hybridization with the single-mismatched target will give a distorted double-helix, causing a decrease in the ECL signal. In the case of the non-complementary and interferent sequences (miRNA-144 and miRNA-155), a response very similar to that of the probe is observed. This result suggests that no hybridization process is taking place; the small increase observed is probably due to a nonspecific adsorption. The reproducibility of the method was estimated from five devices prepared using the same protocol. In all cases, the relative standard deviation (RSD) of the ECL signal was found to be less than 7%. Moreover, the biosensor can detect the target miRNA-21C over a period of 1 month.
miRNA-21 determination in spiked human serum samples
Finally, we evaluated the applicability of the biosensor by determining miRNA-21 in human serum samples using an external calibration method, as described in detail the “Experimental” section. ECL measurements gave an average value of 723.8 a.u. From the calibration plot, the miRNA-21 concentration in the spiked serum sample (final miRNA-21 concentration of 1.00 pM) was found to be 1.13 pM with a recovery of 113% and a RSD of 4%. This result demonstrates that the proposed biosensor can be used for practical applications and has great potential as an alternative to the classical methods for detecting miRNA-21 in human serum samples.
miRNA-21 determination in serum samples from heart failure patients
Based on the great interest of having simple methodologies for rapid biomarker detection and considering the good results obtained with the developed biosensor in spiked human serum samples, we take a step forward and applied our methodology to detect miRNA-21 directly in serum samples from heart failure patients, provided and qRT-PCR checked by IRICYS (Instituto Ramón y Cajal de Investigación Sanitaria), as we described in detail in the “Experimental” section. Two different types of serums were analyzed (serum 1 and serum 2, latest used as control). It is worth to note that in the case of qRT-PCR method, it is necessary to include a RNA extraction step before qRT-PCR analysis and the whole process requires long time of analysis. The developed ECL biosensors only require a denaturation step (see the “Experimental” section) before the analysis. The ECL biosensor response to the serum 1 (Fig. S10) shows a signal of 724 ± 10 a.u. compared to the signal of the probe (575 ± 6 a.u.). The concentration was estimated using the calibration plot (ECL = 29.1 log [miRNA-21C] + 722.3). A concentration of 1.14 pM was obtained. As we describe in the “Experimental” section, serum 2 was used as control. In this case, the biosensor response to serum 2 shows a signal of 600 ± 8 a.u., suggesting that the miRNA-21 concentration is extremely low and no miRNA-21 overexpression is observed. These results agree well with those obtained by qRT-PCR which renders Ct = 23 in serum sample 1 and Ct = 31 in serum sample 2. Normalizer and controls (miR103, UniSp2, and cel 39) did not show differences in Ct between both samples.
Based on the above results, we can conclude that the biosensor developed can detect a miRNA-21 sequence directly in clinical human serum samples from heart failure patients without any previous amplification process. It is a simple, sensitive, fast, and scalable tool for detecting this biomarker, allowing early detection of the disease and increasing the probability of patient survival.