3.1 Design Principle of the Biosensing

The NPG-mediated Y-shaped ratio biosensor is as illustrated in Scheme 1. The biosensor hybridizes by opening the hairpin structure of the CP in the presence of both AP and TPa. CP-AP-TPa hybridization product with a Y-shaped structure was constructed. The current signals of the Fc and Mb will change through the conformational change of CP and the introduction of AP, since CP and AP are connected to the redox reporter Fc and Mb, respectively. Then the TPa concentration is calculated based on the change of IMb/IFc ratio caused by the change in the Mb and Fc current signals before and after hybridization, which reflects the hybridization efficiency. It is worth emphasizing that the increase in the spatial distance between probes has little effect on the specific hybridization efficiency as the TPa concentration decreases within a certain range [22].

Scheme 1

Schematic illustration of the NPG-mediated Y-shaped ratio electrochemical biosensor

3.2 Characterization of NPG

Scanning electron microscopy (SEM, TESCAN MIRA LMS, Czech Republic) was used to study the nano- and micrometer-scale morphological features by photographing the samples at 50 k × and 100 k × magnifications (Fig. 1a, b). The pore morphology of the NPG is the most important feature that affects the electrochemical performance. As shown in Fig. 1a, b, the prepared NPG can be seen in the entire field and the three-dimensional nanoporous structure distributed uniformly with a pore size about 10 to 100 nm, which can provide a large number of binding sites for the CP probe and amplify the signal of the sensor.

Fig. 1

SEM images of nanoporous gold in NPG at a: 50 k × and b: 100 k × magnifications. c EIS and d CV curves of different modified electrodes. 1-Bare Au-E, 2-Au/NPG, 3-Au/NPG-CP, 4-Au/NPG-CP-MCH, 5-Au/NPG-CP-MCH-AP-TPa

3.3 Characterization of the Biosensor Fabrication

The stepwise fabrication of the biosensor was demonstrated by the results of EIS and CV (Fig. 1c, d). The bare gold electrode (bare Au-E) displayed a very small semicircle of EIS signal, where the semicircle diameter equals to the interfacial electron transfer resistance (Ret), as well as a few reversible [Fe (CN) 6]3−/4− redox peaks (curve 1). Compared to the bare Au-E, the Ret and redox intensity of the NPG-modified electrode (Au/NPG) were significantly reduced (curve 2), which was attributed to the large surface area and high conductivity of NPG. When the capture probe CP was assembled onto the electrode surface (Au/NPG-CP), electron transfer was greatly hindered by the negative charge of the DNA [23], increasing the Ret and redox intensity (curve 3). Subsequently, the non-conducting MCH coated on the electrode (Au/NPG-CP-MCH) blocked the active sites on the electrode surface, effectively covering the uncovered space and causing Ret to continue increasing (curve 4). As AP and TPa further hybridized on the electrode (Au/NPG-CP-MCH-AP-TPa), the electrode surface became more negatively charged, leading to a further increase in Ret and redox intensity (curve 5). The results indicated the successful construction of the Y-shaped biosensor.

3.4 Feasibility Analysis

PAGE was used to analyze the feasibility of probe hybridization (Fig. 2a). Lanes 1–4 show the molecular weights and structures of CP, AP, TPa, and TPb. There is no hybridization between CP and AP, TPa, or TPb (lanes 5–7), and there are hybridizations between AP and TPa or TPb (lanes 8 and 9). The result of the three-probe hybridization system indicated that CP-AP-TPa successfully hybridized in lane 10, but CP and AP-TPb did not exhibit hybridization (lane 11). The results of PAGE indicated that the CP-AP-TPa hybridization had high specificity. However, we can judge from the brightness of the electrophoretic band in lane 10 that the hybridization efficiency of CP, AP, and TPa at a concentration of 1 μmol/L was insufficient.

Fig. 2

Feasibility analysis by a PAGE (lanes 1–4: CP, AP, TPa, TPb; lanes 5–9: CP + AP, CP + TPa, CP + TPb, AP + TPa, AP + TPb; lanes 10–11: CP + AP + TPa, CP + AP + TPb) and b SWV detection

This hybridization property was achieved by selection of complementary pairing sequences between AP, CP, and TPa. AP and CP were designed to have a 9-base complementary pairing region (-GGATACCGG-), with a melting temperature of approximately 30 °C. The hybridization temperature was set below 25 °C. TPa and CP were paired with a 9-base complementary pairing region (–CTATCTGCA–) at a hybridization temperature of about 21 °C. The CP hairpin structure with a free energy of −2.5 kcal/mol according to John’s calculations [24] was stable. A temperature about 50 °C was required for opening this hairpin structure, which was significantly higher than that of the hybridization between AP or TPa and CP. Therefore, at room temperature, neither AP nor TPa alone was able to open the hairpin structure of CP for hybridization. The experimental results, especially the PAGE analysis of the probe hybridization, aligned with the design expectations. This indicated that the AP and CP sequences can meet assay requirements of specificity, sensitivity, and false-positive prevention.

The SWV detection was used to verify the feasibility of this Y-shaped biosensor (Fig. 2b). In similar conditions, this Y-shaped biosensor with Au/NPG was able to detect larger Fc and Mb signals than the bare Au-E at room temperature. In the presence of only CP, a strong Fc signal but no Mb signal was detected due to the hairpin structure that brought Fc close to the electrode surface. However, in the absence of TPa, the CP hairpin structure was difficult to be opened by AP, resulting in a high Fc signal and a weak Mb signal from non-specific adsorption of probe AP. The CP hairpin structure was easily transformed into the Y-shaped structure by hybridization when both AP and TPa existed. The alteration in probe shape caused the Fc to move away from the electrode surface while Mb to move closer, resulting in a weaker Fc signal and a higher Mb signal. When TPb and AP were both present, CP cannot fully hybridize with TPb, making it difficult to open the hairpin structure of CP by AP-TPb. As a result, the changes in the Fc and Mb signals were relatively weak. The tests using only 100 pmol/L TPa and mixing with equal concentration of TPb showed nearly identical results, proving the biosensor’s exceptional specificity. The above results validated the feasibility of the present Y-shaped electrochemical biosensor for target detection.

3.5 Optimization of the Biosensing Conditions

In Y-shaped biosensors, factors like CP concentration and hybridization time can affect the experimental results. We used 100 pmol/L of TPa and 2 μmol/L of AP to detect the signals of CP with different concentrations, and the response of IMb/IFc changed with different CP concentration. Since a weak Mb signal was still detected in the presence of AP only, we optimized the CP concentration based on the signal-to-noise ratio (IMb/IFc)(AP-TPa)/(IMb/IFc)AP. The results in Fig. 3a indicated that the signal-to-noise ratio of the system increased rapidly at the CP concentrations of up to 2 μmol/L, but decreased beyond that, so the CP concentration of 2 μmol/L was chosen for subsequent experiments. As shown in Fig. 3b, at CP concentration of 2 μmol/L, the value of IMB/IFc significantly rises with longer incubation time of up to 80 min.

Fig. 3

Optimizations of experimental parameters: a concentration of CP probes and b incubation time of surface hybridization. c The linear range, d selectivity, e repeatability, and f stability of the biosensor. (Error bars represent the standard deviations measured from three different tests. Scan range from −0.5 to 0.6 V, amplitude 0.05 V, frequency 100 Hz in 10 mmol/L PBS buffer.)

3.6 Sensitivity, Selectivity, Repeatability, and Stability Analysis of Y-Shaped Biosensor

Under the optimal conditions, different concentrations of TPa were detected by SWV. As the concentration of TPa increased, the current IMb became larger while the IFc became smaller. The linear regression equation is IMb/IFc = 0.06213lgc(TPa) + 0.05629, with the concentration of TPa ranging from 10 fmol/L to 100 pmol/L (Fig. 3c). The correlation coefficient was 0.9982, and the LOD was 0.211 fmol/L (S/N = 3).

To ascertain the selectivity of the Y-shaped electrochemical biosensor, the concentrations of TPa and TPb were maintained at 1 nmol/L, a level above the detection limit. The achieved data indicated that the difference of IMb/IFc between TPa and TPb was more than five times (Fig. 3d). The results demonstrated strong selectivity in separating well-matched targets from mismatched target sequences, which aligns with the outcomes seen in the PAGE analysis (Fig. 2a). This is because the Y-shaped structure reduces the binding rates of mismatched targets, while the CP hairpin structure can only be opened by the precisely matched TPa. Several studies [25, 26] on single nucleotide polymorphism (SNP) have proposed that hairpin structures unite binary DNA probe molecular beacons (also known as binary probes) with the shorter probe binding to the mismatched base. Both probes hybridize only when they bind to the target, significantly enhancing the sensitivity and specificity of SNP detection. This study demonstrated that the Y-shaped biosensor has excellent selectivity for recognizing sequences with single base mismatches. In the case of mixing with same concentration of miR-18b, the signal intensity was almost the same as that of detecting only miR-18a. This anti-interference ability is crucial for clinical applications, as serum samples contain numerous interferences.

In addition, we analyzed the results of 5 different batches of biosensors for detecting 100 pmol/L of TPa. The relative standard deviation (RSD) was 4.6%, proving that the electrodes have good repeatability between batches (Fig. 3e).

Furthermore, the biosensor’s stability was evaluated. As shown in Fig. 3f, Au/NPG-CP/MCH electrodes were stored at 4 °C and measured miR-18a every 5 d. After 15 d, the biosensor response still kept up to 95.59% of the initial. This indicated that the biosensor was stable, which can increase its lifespan. These features are beneficial in clinical assays, as the biosensor is likely to perform with adequate power.

3.7 Recovery Test of miR-18a in Diluted Serum Samples

The Y-shaped biosensor successfully detected miR-18a levels of about 24.165 fmol/L and 121.997 fmol/L in tenfold and twofold of healthy human serum, but not in FBS. miR-18a levels in healthy human serum samples were calculated to be about 242.823 fmol/L, which is almost identical to the data of Hirajima et al. [27], which were between 0.19 and 1.27 amol/μL, with a mean value of 0.73 amol/μL. Using qRT-PCR combined with a standard curve was able to measure the levels of miR-18a in the serum of patients with esophageal squamous carcinoma and healthy individuals. In this study, the process of detecting miR-18a in serum samples only requires dilution of serum samples and heating at 95 °C for 15 min [28, 29], without separation, extraction, and amplification. miR-18a in the blood is present in exosome vesicles, and high-temperature cleavage of the vesicles releases the miR-18a so that it can be detected by the biosensor. Moreover, the protein in serum can be denatured after high temperature and the interference of protein components can be removed completely by centrifugation, which makes the detection of miRNA in serum more easy.

To further evaluate the efficiency of the biosensor response to miR-18a in real samples, the tenfold serum samples with miR-18a concentrations of 10 pmol/L, 1 pmol/L, and 100 fmol/L were prepared and measured (Table 1). Each concentration was examined three times. The amount of miR-18a in serum was subtracted from the detected result and divided by the added concentration to obtain the recovery. The results showed the recovery ranged from 95.41% to 104.81% with the RSD of 1.81%–4.01%. This indicated the proposed Y-shaped biosensor can accurately detect miR-18a in serum samples.

Table 1 Actual and measured concentrations of miR-18a in human serum samples (n = 3)