2.1 Principle of the proposed ECL biosensor for SARS-CoV-2 monitoring
In our strategy, ECL emitting material (Ti3C2/PEI-Ru@Au) was coated onto the GCE surface, and the ECL signals were used as the signal output of the sensor. Then, we modified the automated molecular machine on the surface of Ti3C2/PEI-Ru@Au. The inverted tetrahedron has only one vertex modified by thiol, while possessing three capture probes where two of the strands (H and P) are partially complementarily paired with H’ and P’, respectively. The one remaining capture probe (S’-S) forms a hairpin structure by itself and its S’ end carries the quenching motif ferrocene (Fc) of the ECL signal. When the target DNA (T) binds to the exposed toehold sequence of the anticodon H’, H is released, which in turn binds to the toehold sequence of the anticodon P’, which allows P to be released. Immediately afterward, P binds to the anticodon S’-S, producing a site that is recognized and cleaved by the nuclear endonuclease (Nt.BbvCI), resulting in the release of quenched motif Fc and reuse of the P. In this process, the repeatedly generated P enables the cyclic cleavage of the S’-S and the cyclic release of the Fc, resulting in the generation of a transduction signal that allows the assessment and quantitative analysis of target concentration.
2.2 Characterizations of the Ti3C2/PEI-Ru@Au
We have characterized the synthesis of Ti3C2/PEI-Ru@Au. Fig. 1A and Fig. 1B show the TEM morphologies of Ti3C2 and Ti3C2/PEI-Ru@Au, respectively. We can see that Ti3C2 shows monolayers or multilayers of nanosheets. After the modification, it can be seen that the gold nanoparticles are uniformly distributed on the Ti3C2 surface. The magnified Ti3C2/PEI-Ru@Au TEM image (Fig. 1C) and its inset show that the gold nanoparticles are uniformly distributed and the particle size is approximately 4.3 nm. To further validate the synthesis of Ti3C2/PEI-Ru@Au, we characterized the elemental distribution (Figs. S2A-H). The elemental distribution also shows that Au particles as well as Ru(dcbpy)32+are uniformly distributed on the Ti3C2 nanosheets.
We have performed XPS analysis of Ti3C2/PEI-Ru@Au nanocomposite. The XPS survey spectrum of the complex (Fig. 1D) exhibits Ti 2p and C 1s peaks, which are the elemental peaks of Ti3C2. The formation process of Ti3C2 produces F and Cl elements; therefore, F 1s and Cl 2p peaks are observed. The complex also has Ru 3p and Ru 3d peaks, and Au 4f characteristic peaks, as a result of the introduction of Ru(dcbpy)32+ and the reduction of Au3+ to gold nanoparticles. We also elaborated the bonding interaction between the complexes by fitting C 1s, O 1s, N 1s and Au 4f (Figs. 1E-H). Four fitted peaks appear in the C 1s peak of Ti3C2/PEI-Ru@Au, which mainly correspond to the characteristic bonds of O-C=O, C-O/C-N, C-C and C-Ti. Similarly, the XPS spectra at O 1s show the typical bonds of C-Ti-(OH), C-Ti-Ox and TiO2. The XPS spectra at N 1s also show the characteristic bonds of R-NH2, R-NH-R, R=N-R and N-Ti. The double peaks of Au at Au 4f7/2 (82.83 eV) and Au 4f7/2 (86.51 eV) also verified the presence of gold particles. All these XPS analysis data indicate the formation of the Ti3C2/PEI-Ru@Au complexes. We also verified the synthesis of the complex by UV-vis spectral analysis as depicted in Fig. 1I. It showed that the Ti3C2/PEI-Ru@Au complex has both Ru-PEI (characteristic peaks at 300 nm and 475 nm) and Au nanoparticles (characteristic absorption peak at 525 nm) peaks, also verifying the synthesis of Ti3C2/PEI-Ru@Au complex.
2.3 Characterization of the inverted DNA tetrahedral scaffold
We also characterized the inverted DNA tetrahedra by AFM (Fig. 2A). It can be clearly seen that the DNA tetrahedra formed self-assembled by four strands (H-T1, P-T2, S’-S-T3, T4) appear granular and uniform in size, which also confirms the successful preparation of our tetrahedral probes. Meanwhile, we characterize the heights of the tetrahedral particles relative to the silicon wafer base (control) as shown in Figs. 2B-2F. We define the silicon wafer substrate as the control, and the heights of the tetrahedra relative to the substrate as the actual height of the DNA tetrahedra. These characterization and analysis tools all demonstrate the successful synthesis of inverted tetrahedra with relatively homogeneous dimensions.
2.4 Feasibility of DNA reactions exploited by the automated molecular machine
We next verified the possibility of cascade reactions on the automated molecular machine as shown in Fig. S5A. It is shown that when T and H’-H-T1 are present together that they can be effective against P’-P-T2, eventually producing three major new bands: T-H’, T1-H-P’ and P-T2. However, when T is not present, H’-H-T1 and P’-P-T2 do not react with each other. Likewise, T does not act on P’-P-T2 when the H’-H-T1 intermediate is absent.
Upon the addition of H-T1 with P’-P-T2 to S’-S-T3 with NE, we observed the disappearance of S’-S-T3 and the appearance of three new bands: P-T2, T1-H-P’ and cleaved S’-S-T3, which indicates the successful linkage of the nicking reaction to the previous one proved by Fig. S5B. Owing to the essential H-T1 linkage, the modular cascade reaction on the DNA molecular machine proceeds smoothly. At the same time, the newly generated P-T2 is available for the continuous action of S’-S-T3 until its depletion. These evidences provided by the two images suggest the possibility that the reaction occurs as expected on the automated molecular machine.
2.5 Monitoring performance, specificity and stability of this automated molecular machine for target monitoring
In our testing experiments, the automated molecular machine exhibits highly sensitive ECL variations for target concentrations as shown in Fig. 3A and 3B. The increased ECL intensity (ΔECL intensity) increases with the logarithmic value of the target concentration in the range from 1 pM to 1 nM. The linear equation of the calibration plot is y = 431.5 + 1255.9lgCtarget, where y represents ΔECL intensity. The calculated limit of detection (LOD) was 0.68 pM according 3σ method. In Table S2, we compared our strategy in terms of detection methods, sensitivity and detection range with other viral DNA detection methods. It can be noted that our detecting method performs well in these three aspects. Besides, this method as a fundamental strategy in combination with other signal amplification techniques such as hybridization chain reaction (HCR), loop-mediated isothermal amplification (LAMP), catalytic hairpin assembly (CHA) could achieve higher sensitivity. We further evaluated the specificity of DNA automated molecular machine for the target monitoring. The sensing performances of three non-specific strands with 10-fold target concentration, including Bat SARS-related CoV isolate bat-SL-CoVZC45 target (M1), BM48-31/BGR/2008 target (M2), and SARS-CoV target. As shown in Fig. 3C, we can see that the ECL intensity of the target is higher than the DNA intensity of the various mismatched strands, which also illustrates that our automated molecular machine has superb accuracy. Also, our automated molecular machine has excellent reproducibility as illustrated in Fig. 3D. The automated molecular machine exhibits extremely high signal stability by performing continuous potential scans for 15 cycles, with a relative standard deviation (RSD) of only 2.54% and 2.13% for 50 pM and 0.5 nM, respectively.
2.6 Saliva environmental and clinical human serum dilution sample determinations
Since viral DNA are usually obtained from viral RNA by techniques such as RT-PCR or RT-RPA during actual virus detection, there are some non-specific substances that interfere with the diagnosis of COVID-19. Therefore, we discussed the feasibility of the assay under complex environments and clinical specimens. The pharyngeal swabs method usually collects viral or bacteria samples in saliva. To verify the ability of the DNA automated molecular machine to monitor targets in the saliva environment, we added targets with different concentrations to 20-fold or 50-fold human saliva dilutions and verified their feasibility. As shown in Table S3, after repeated experiments (n=3), we obtained recoveries between 96.4 and 104.8%, which further validates the reliability of our DNA automated molecular machine platform for target monitoring. During the experiments, targets in the range of 5 pM to 5 nM were successfully monitored, demonstrating the universality of our DNA automated molecular machine in the complex saliva environments.