Design of SPCE/ZrGA/mDNA-J bioplatform

Figure 1 shows the assembly of the Zr-MOF-rGO-Au (ZrGA) (Fig. 1a), the isolation and extraction of human blood sEV-miRNAs and the detection mechanism of sEV-miRNAs using the proposed portable SPCE/ZrGA/mDNA-J bioplatform (Fig. 1b). Combined with the finger-sized U-disk WEA (plug and play), reliable 2.4-G data transmission and up to 20-m transmission distance can be achieved (Figure S1). We prepared GO tightly packed Zr-MOF (Zr-MOF-rGO) via a one-step method, and the nanocomplex was modified on the SPCE surface. The synthesized graphene contacts all the faces of the MOF octahedron, which effectively accelerates the space charge separation and inhibits the recombination of electron–hole (e−–h+) pairs, affording accelerated interfacial electron transfer. After the electrodeposition of Au nanoparticles (NPs) to form SPCE/ZrGA, the DNA tetrahedron with three “robotic arms” (mDNA-J) was attached to SPCE/ZrGA by Au–S bonds.

Fig. 1

(a) Assembly of Zr-MOF-rGO-Au (ZrGA) and (b) schematic of small extracellular vesicle–derived microRNA (sEV-miRNAs) analysis for lung cancer diagnosis

As shown in the left enlarged part of Fig. 1b, mDNA-J comprises three pendulum-arm chains and a locus chain. The lower end of the pendulum-arm chains can closely bond with the three faces of the DNA tetrahedron and firmly attach to SPCE/ZrGA, and the upper end is the DNAzyme activated by Na+ and the LOCK chain domain. In the absence of target miRNA (miR-21, model), the substrate-cleaving ability of Na+-specific DNAzyme is locked owing to the hybridization between and their locking strands. However, in the presence of target miRNA, the locking strands sense and hybridize with the target miRNAs, release enzyme strands to open the hairpin structure, and then cleave their corresponding substrates, leaving a sticky end, which triggers HCR. The activated arm cuts the “rA” site on the lateral side of the mDNA-J (trace chain) skip by skip. The cut residues can be used as a “toehold” for HCR to form a stable dsDNA polymer in situ until the supply of H1 or H2 hairpins modified with “Fc” signal tags is exhausted. The redox signals from Fc are ultrasensitively measured to quantify and qualify the miRNA electrochemically owing to the excellent redox properties of Fc and the redox-signal-enhancing effect of SPCE/ZrGA/mDNA-J (detailed description, Figure S2).

Characterization of SPCE/ZrGA bioplatform

Figure 2a shows the assembly of the ZrGA. The particle size of Zr-MOF was observed using field-emission scanning electron microscopy (FE-SEM), which revealed that Zr-MOF has a uniform particle size (400 ± 50 nm) and a typical octahedral shape (Fig. 2b). Its morphology was further observed using transmission electron microscopy (TEM), and the obtained images are shown in Fig. 2c and d. Figure 2e and f show the morphologies of Zr-MOF-rGO with a graphene layer wrapping the Zr-MOF octahedron. High-resolution TEM (HR-TEM) revealed the edges of Zr-MOF and rGO (Fig. 2f), showing the lattice structure of rGO with a crystal plane spacing of 0.34 nm. Figure 2g and h show that Au NPs were successfully deposited on the Zr-MOF-rGO surface. The size of the Au NPs was 15 ± 5 nm, and the interplanar spacing was 0.238 nm, as revealed by HR-TEM (Fig. 2h). Further, energy-dispersive spectroscopy (EDS) showed that C, N, O, Zr, and Au were uniformly distributed on the ZrGA surface (Fig. 2i and j).

Fig. 2

Characterization of the prepared electrodes: (a) assembly of Zr-MOF-rGO-Au (ZrGA). Transmission electron microscopy (TEM) images of (b–d) Zr-MOF, (e-f) Zr-MOF-rGO, and (g–h) ZrGA nanoparticles (NPs). (i) Energy-dispersive spectroscopy (EDS) elemental mapping and (j) spectrum of the elemental distribution of Zr-MOF-rGO-Au. (k) X-ray diffraction (XRD) and (l) X-ray photoelectron spectroscopy (XPS) images of Zr-MOF, Zr-MOF-rGO, and ZrGA

The XRD patterns of Zr-MOF (black), Zr-MOF-rGO (red) and ZrGA (blue) showed sharp diffraction peaks (Fig. 2k), which are consistent with the simulated data for the single crystal, indicating high purity and crystallinity [28, 42]. Figure 2l and S3a show full X-ray photoelectron spectroscopy (XPS) images of Zr-MOF, Zr-MOF-rGO, and ZrGA. Zr MOF-rGO-Au showed six characteristic peaks of O 1s, N 1s, C 1s, Zr 3p, Zr 3d, and Au 4 f. High-resolution XPS of C 1s for ZrGA showed a strong peak at the binding energies of 284.99 eV (Figure S3b), which is attributed to the sp2-hybridized C–C/C–H bond, and the peaks at 286.40 and 289.20 eV are attributed to the C–N and O–C = O bonds, respectively. The sample also showed two peaks ascribed to Zr 3d3/2 and Zr 3d5/2 (Figure S3c), and the peaks at the binding energies of 84.54 and 88.20 eV (Figure S3d) are attributed to the Au 4f7/2 and Au 4f5/2 chemical binding states of Au 4f, respectively, indicating that Au NPs were successfully synthesized on Zr-MOF-rGO.

Characterization of the mDNA-J assembly

Native polyacrylamide gel electrophoresis (PAGE) and atomic force microscopy (AFM) were employed to evaluate the construction and reaction mechanism of the mDNA-J. As shown in lanes 1–13 (Fig. 3a, Table S1), with the addition of new strands, the migration distance decreased owing to the increase in molecular mass and the more complex spatial structure, and the Sw of a single chain showed a smaller migration distance owing to its long sequence (Fig. 3a, lane 4). mDNA-J migrated more slowly than other assemblies constructed by sequences of fewer than nine strands, and the clear bright band on the gel confirms the successful assembly and high yield of mDNA-Js (Fig. 3a, lane 13). AFM confirmed that the prepared mDNA-J has a tetrahedral structure with a diameter of ~ 5.8 nm (Fig. 3b).

Fig. 3

Characterization of the multiarmed DNA tetrahedral jumper (mDNA-J) assembly. (a) Polyacrylamide gel electrophoresis (PAGE) image of mDNA-Js. Lane M: 20 bp ladder; lane 1: S4; lane 2: S1-a-SH; lane 3: S1-r-b; lane 4: SW; lane 5: Lock; lane 6: S4 + S1-a-SH; lane 7: S4 + S1-a-SH + S2-a-SH; lane 8: S4 + S1-a-SH + S2-a-SH + S3-a-SH; lane 9: S4 + S1-a-SH + S2-a-SH + S3-a-SH + S1-r-b; lane 10: S4 + S1-a-SH + S2-a-SH + S3-a-SH + S1-r-b + S2-r-b; lane 11: S4 + S1-a-SH + S2-a-SH + S3-a-SH + S1-r-b + S2-r-b + S3-r-b; lane 12: S4 + S1-a-SH + S2-a-SH + S3-a-SH + S1-r-b + S2-r-b + S3-r-b + SW; lane 13: S4 + S1-a-SH + S2-a-SH + S3-a-SH + S1-r-b + S2-r-b + S3-r-b + SW + Lock (the detailed sequence is shown in Table S1 of the Supporting Information). (b) Atomic force microscopy (AFM) image of mDNA-Js and the corresponding 3D model. Scale bars, 5 nm. (c) Functional domains of the mDNA-Js

As shown in Fig. 3c, the catalytic active core of the Na+-specific DNAzyme was separated into two halves by a locking strand (Lock), inhibiting its catalytic activity. The susceptibility of ribonucleotide to hydrolytic cleavage was 100,000-fold higher than that of its deoxyribonucleotide, a DNA-RNA chimeric sequence comprising an adenosine ribonucleotide (rA) flanked by two DNA domains of the hairpin substrate strand, which can hybridize to two arms of the enzyme strand. To avoid the steric interference of the DNA tetrahedron with the DNA hybridization and increase the accessibility of the hairpin substrate strand to the walking enzyme strand, we incorporated a poly-T spacer between the DNA tetrahedron and the enzyme strand. The DNA tetrahedrons could be anchored on the SPCE/ZrGA electrodes with a highly desirable orientation via Au–S chemical conjugation, and the unique pyramidal structure with three hairpin substrate strands shows that all immobilized hairpin substrate strands were distributed at fixed distances to the DNA tetrahedron, thereby maintaining spatial orientation for the effective assembly of the Na+-specific DNAzyme. The DNA tetrahedron was adopted as the foundation because a DNA tetrahedron of this size can be defined as a nanostructure, effectively decreasing the hindrance effect and maintaining spatial orientation for improved miRNA recognition.

Electrochemical properties of SPCE/ZrGA/mDNA-J bioplatform

The properties of the raw materials for the bioplatform are vital. Thus, cyclic voltammetry (CV) was employed to evaluate the electrochemical performance of the SPCE/ZrGA bioplatform at different scan rates (10–295 mV s− 1) in 5 mM [Fe(CN)6]3−/4− containing 0.1 M KCl (Fig. 4a). Figure 4b shows the variation of peak currents with the square of the scan rate. Both the anode and cathode peak currents showed linear relationships with the equations Ip.a.(µA) = 293.34v1/2(V s− 1)1/2 +9.97 (R2 = 0.997) and Ipc(µA) = − 262.69v1/2(V s− 1)1/2-12.26 (R2 = 0.998), respectively. This oxidation–reduction reaction of SPCE/ZrGA indicates a diffusion-controlled process. Furthermore, the electroactive surface area (A) of the three electrodes was calculated using the Randles–Sevcik equation: [28]

$$_}=2.69\times ^^}A^}^}_$$

(1)

where Ip is the peak current (A), n (= 1) is the number of electronic transfers, D (= 6.7 ± 0.02 × 10–6 cm2 s− 1) is the diffusion coefficient, υ = 0.05 V s− 1, and C0 (= 5 × 10− 6 mol cm–3) is the [Fe(CN)6]3−/4−concentration. According to Eq. (1), SPCE/ZrGA has an electroactive surface area of 0.191 cm2, which is 1.3 times that of SPCE/Zr-MOF-rGO (0.148 cm2) and 2.5 times that of SPCE (0.076 cm2). In addition, Fig. 4c shows the variation of the anode and cathode peak potentials with the logarithm of the scan rate (lg(v)). The equation of line for the anode peak potential is Ep.a.(V) = 0.0582 lgv (V s− 1) + 0.277 (R2 = 0.988), and that of the cathode is Epc(V) = -0.0698 lgv (V s− 1) + 0.011 (R2 = 0.981). Based on the Laviron theory: [44]

$$lg\frac_}_}=lg\frac,$$

(2)

$$\begin lg = \alpha \,}\left( \right) + \left( \right)l\,g\alpha - l\,g \\ \frac}} - \frac} \right)nF\vartriangle }}, \\ \end$$

(3)

where ka and kc are the slopes of Ep.a.–lg(v) and Epc–lg(v) lines, respectively, R is the gas constant (8.314 J·(mol·K) −1), T is the absolute temperature (298 k), F is the Faraday’s constant (96,493 C·mol− 1), and n is the number of electronic transfers (= 1). From Eqs. (2) and (3), α = 0.455, which is the charge transfer coefficient, and ks = 1.243 s− 1, which is the apparent electron transfer rate constant. The electron transfer rate ks obtained here is higher or comparable to the values reported in the literature (Table S3), indicating that ZrGA exhibits accelerated electron transfer.

Fig. 4

Electrochemical performance of the SPCE/ZrGA/mDNA-J bioplatform: (a) Cyclic voltammetry (CV) curves of SPCE/ZrGA at different scan rates in 5 mM [Fe (CN)6]3−/4− containing 0.1 M KCl. (b) Variation of anodic and cathodic peak currents with scan rate. (c) Curve fitting of the logarithm of the scan rate (lgv) versus anode/cathode potential. (d) Flow diagram of miRNA detection by the proposed SPCE/ZrGA/mDNA-J bioplatform. (e) CV curves of bare SPCE, SPCE/Zr-MOF-rGO, SPCE/ZrGA, SPCE/ZrGA/mDNA-J, and SPCE/ZrGA/mDNA-J/miRNA/H1-H2 in 5 mM [Fe (CN)6]3−/4− containing 0.1 M KCl. Feasibility study: (f) Flow diagram of miRNA detection by mDNA jumpers. (g) Square wave voltammetry (SWV) responses of the prepared electrode in the presence of 200-nM miR-21 and 60-mM Na+ (0.01 M PBS solution; scan rate: 50 mV s− 1)

Figure 4d shows the assembly process of the SPCE/ZrGA/mDNA-J bioplatform, and CV was employed to characterize the electrochemical behavior of the electrode during its modification on WEA. The CV curves of different modified electrodes in a 5 mM [Fe (CN)6]3−/4− detection solution containing 0.1 M KCl are shown in Fig. 4e. After the Zr-MOF-rGO nanocomposite was modified on the surface of the SPCE electrode (red curve), the current response was significantly higher than that of the bare SPCE electrode (black curve). This is because, in addition to the good conductivity of the Zr-MOF-rGO nanocomposite, it greatly increased the specific surface area of the electrode. Notably, the current signal response of ZrGA (blue curve) was approximately three times that of Zr-MOF-rGO, indicating that Zr-MOF-rGO provided a large specific surface area to support Au NPs and could immobilize numerous capture probes. When the tentacles of mDNA-Js were fixed on the electrode surface by Au–S bonds (SPCE/ZrGA/mDNA-J, purple curve), the current response significantly decreased. This is attributed to the electrostatic repulsion between the self-negatively-charged phosphate skeleton and Fe2+/3+ in the solution, which hindered current diffusion between [Fe(CN)6]3−/4− and the electrode surface, thereby decreasing the redox peak current. Due to the hybridization and polymerization of more non-electroactive DNA chains, the current response further decreased after HCR (SPCE/ZrGA/mDNA-J/miRNA/H1-H2, green curve). This is because the DNA hybridization double chains formed on the electrode surface further hinder current diffusion from [Fe(CN)6]3−/4− to the electrode surface, thereby decreasing the current response signal.

A feasibility study was conducted by measuring the change in the target miRNA-induced electrochemical signal in the presence of 200 nM miR-21 (target) and 60 mM Na+ in a 0.01 M PBS solution. The detection procedure is depicted in Fig. 4f g. miR-21 was selected herein to demonstrate the performance of the proposed platform because it is a prevalent circulating miRNA biomarker overexpressed in lung cancer. The target miR-21 serves as a “toehold” to initiate interactions with the blue domain of Lock. Subsequently, the complementary double chain (△Genzyme strand: Lock = − 26.62 kcal mol− 1) formed partly by enzyme strands and partly by Lock is opened and forms an miR-21–Lock complex (△GmiR-21: Lock = − 33.35 kcal mol− 1). Thus, the red domain of the enzyme strand is no longer occluded and can bind to the red domain of the hairpin substrate strand (△Gsubstrate strand = − 5.76 kcal mol− 1), then opens the hairpin substrate strand and forms a Na+-specific DNAzyme (△Genzyme strand: substrate strand = − 34.06 kcal mol− 1). With the addition of Na+, DNAzyme is activated to specially cut the substrate strand. The leaving sticky end opens hairpin H1 (△GH1 = − 3.75 kcal mol− 1) and forms a sticky-end–H1 double chain (△Gsticky end: H1 = − 25.10 kcal mol− 1), at which a new sticky end opens hairpin H2 (△GH2 = − 6.19 kcal mol− 1) and forms a sticky-end–H1–H2 double chain (△Gsticky end: H1:H2 = − 43.97 kcal mol− 1). The constant existence of sticky ends induces HCR (H1 and H2 structures, Figure S4).

Figure 4 g shows square wave voltammetry (SWV) curves of the SPCE/ZrGA/mDNA-J bioplatform under different detection conditions. There was almost no current response for the SPCE/ZrGA/mDNA-J bioplatform in the presence of only 60 mM Na+ (black curve) or miR-21 (target, blue curve), indicating that HCR was not induced. A significant SWV response peak (Fc, 0.52 V) was observed in the presence of miR-21 and Na+ (red curve), indicating that HCR was induced and DNA hairpins with Fc (H1-Fc and H2-Fc) were opened to hybridize with each other.

Detection of sEV-miR-21

Optimization of conditions: To further optimize the SPCE/ZrGA bioplatform, we first optimized the mDNA-J concentration and incubation time immobilized on the sensing surface of SPCE/ZrGA. When a parameter is optimized, other parameters would be optimal. The detection sensitivity of the SPCE/ZrGA bioplatform for biomolecules largely depends on the number of fixed mDNA-Js. Figure S5 shows that the SWV current signal detected by 0.2 µM miR-21 at an mDNA-J concentration range of 0.2–1.2 µM rapidly increased at the beginning and reached a maximum at 1.0 µM and then remained constant. As the incubation time increased from 30 to 150 min, the generated current signal increased almost linearly within the first 120 min and then remained constant afterward (Figure S6). In addition, the time for H1–H2 HCR contributed considerably to the total detection time (Figure S7). In the time range of 30–150 min, the SWV response signal increased almost linearly within the first 60 min and then stabilized, indicating that the reaction reached a steady state. Therefore, for an mDNA-J concentration of 1.0 µM, an mDNA-J incubation time of 150 min and an HCR time of 60 min were selected for subsequent experiments.

Fig. 5

(a) Schematic of the SPCE/ZrGA/mDNA-J bioplatform for the detection of miR-21. (b) SWV responses to different concentrations of miR-21: (a) 0 M, (b) 100 aM, (c) 1 fM, (d) 100 fM, (e) 10 pM, (f) 1 nM, (g) 100 nM, and (h) 200 nM. (c) Calibration plots of (b). (d) Current responses of the SPCE/ZrGA/mDNA-J electrode with the target (miR-21), interfering miRNAs (miR-155, miR-26a, miR-192, miR-10b, single-base mismatch (SM) miR-21, three-base mismatch (TM) miR-21), and different cofactors (K+, Zn2+, Mg2+, and Fe3+). Error bars: SD; n = 3. (e) Reproducibility of the proposed SPCE/ZrGA/mDNA-J biosensor (n = 8). The error bars indicate standard deviations for five measurements

Figure 5a shows the response of the SPCE/ZrGA/mDNA-J bioplatform for the detection of miR-21 in the presence of 200 nM miR-21 (target) and 60 mM Na+. Under optimal experimental conditions, the current responses of the SPCE/ZrGA/mDNA-J bioplatform to the target miR-21 at various concentrations (0–0.2 µM) were examined using the SWV method (Fig. 5b). With an increase in the concentration of the target miR-21, the Fc signals gradually increased. The relationship between the concentration and the current signal was fitted to a linear function (Fig. 5c). For the concentration range of 100 aM–0.2 µM, the equation for the linear fitting is y = 5.84 logC miR-21 + 95.54 (R2 = 0.997), where C is the concentration of the targeted miR-21. The calculated LOD for miR-21 is 34.6 aM (S/N = 3). Considering the detection range and LOD, the performance of the SPCE/ZrGA/mDNA-J bioplatform is comparable to or better than that of previously reported biosensors (Table S4). Notably, this bioplatform is promising for POC applications because it does not require target amplification, making it less time-consuming and easy to operate (it can be operated by merely dropping 10 µL of a reactive solution).

Specificity and reproducibility of the proposed SPCE/ZrGA/mDNA-J bioplatform

To further investigate the selectivity of the SPCE/ZrGA/mDNA-J bioplatform, we introduced a variety of control targets including mismatched targets based on binding free energy changes via the NUPACK and different cofactors (K+, Zn2+, Mg2+, and Fe3+) to conduct anti-interference experiments. As shown in Fig. 5d, the peak current change (△I) was highest after the hybridization of the complementary target (miR-21) with 60 mM Na+, whereas other interfering miRNAs and different cofactors produced very weak current signals. This is attributed to the high specificity of chain substitution reactions. The reproducibility of the electrode was further studied (Fig. 5e). We measured the SWV current response of 0.2 µM miR-21 in eight ZrGA biosensors, and the calculated relative standard deviation is 2.49% (miR-21), indicating good repeatability.

sEV-miR-21 clinical sample detection and comparison

The morphology, characteristic proteins, and particle-size distribution of the extracted sEVs were analyzed. TEM revealed that the sEVs have a cup-shaped membrane structure (Fig. 6a and S8), which is consistent with previous reports [48, 49]. For the protein expression, western blot (WB) experiments confirmed the presence of characteristic protein markers, CD63, CD81, and CD9, on the sEVs membrane (Fig. 6b), corresponding to the 32, 20, and 23 kDa bands, respectively, which is consistent with previous reports. Furthermore, NP tracking analysis (NTA) revealed that about 98% of the sEVs from serum specimens have a particle size of 30–250 nm with a mean of 105.2 nm (Fig. 6c). Therefore, the extracted sEVs maintained a good membrane structure and significant distribution of specific proteins, demonstrating the effective extraction of sEVs from serum specimens.

Fig. 6

(a) TEM image of isolated sEVs derived from clinical blood samples. The scale bar is 100 nm. (b) Western blot bands of CD63, CD81, and CD9 on the sEV membrane, NC: PBS. (c) Nanoparticle tracking analysis (NTA) result of sEVs. (d) and (e) Validation of clinical differentiation for sEVs miR-21 in clinical samples from healthy individuals (10 samples as a control) and non-small-cell lung cancer (NSCLC) patients (16 samples) who were in the tumor stage using qRT-PCR and our platform, respectively. The results were analyzed by an unpaired, two-tailed Student’s t-test (two groups) or ANOVA (three or more groups) followed by Bonferroni’s correction if needed. ***: p < 0.001, ****: p < 0.0001. (f) Comparison between our platform and qRT-PCR towards sEVs miR-21 detection. (g) Correlation between the results of sEVs miR-21 detection measured using the proposed sensor and qRT-PCR

To confirm the applicability of the proposed SPCE/ZrGA/mDNA-J portable bioplatform to clinical samples, we explored its response to sEV-miR-21 in total RNA extracted from sEVs of clinical plasma samples. We analyzed 26 clinical blood samples (10 samples from healthy individuals and 16 samples from non-small-cell lung cancer (NSCLC) patients) using the proposed bioplatform and quantitative real-time PCR (qRT-PCR), and both techniques showed similar results (Fig. 6d and e). The signal of the proposed SPCE/ZrGA/mDNA-J bioplatform in the clinical sample tests increased proportionally with NSCLC staging in the samples collected from NSCLC patients. Furthermore, a much higher expression of miR-21 was observed in stage IV patients compared with that of healthy individuals, which is consistent with previous reports [28, 50]. Moreover, the proposed bioplatform and classic qRT-PCR showed similar results in differentiating samples from NSCLC patients (Fig. 6f), indicating good consistency between the proposed bioplatform and qRT-PCR (R2 = 0.988) (Fig. 6g). These results show that the proposed bioplatform has high accuracy and applicability and can accurately reflect NSCLC staging in clinical samples.