3.1. Morphological and elemental analysis of the Ti3C2Tx MXene
Transmission electron microscopy (TEM) was employed to study the morphology of the used Ti3C2Tx MXene sample, which presented a remarkably large flake and some stacked little fragments (Fig. 1A). Moreover, the high-angle annular dark-field (HAADF)-STEM image showed no observable spot on the flake, suggesting the uniform distribution of the elements (Fig. 1B). As shown in Fig. 1C-1E, STEM-EDS elemental mappings of C, Ti, and O presented outlines well matched with the HAADF-STEM image, which visually displayed the elemental composition of the Ti3C2Tx MXene.
Energy dispersive X-ray spectroscopy (EDX) was also utilized to analyze the elemental composition of the Ti3C2Tx MXene (see in Supplementary Material, Fig. S1), which involved of C, O, Ti, F and Al elements. Among them, F and Al were mainly from the residual impurities and their contents were notably lower than C, O, and Ti.
Atomic force microscope (AFM) was employed to further study the morphology of Ti3C2Tx MXene. As shown in Fig. 1G and 1H, AFM image presented sheets with thickness of about 4 nm, corresponding to the thickness of 3 layers.
3.2. Electrocatalytic activity of Ti3C2Tx MXene for phenolic compound oxidation
The electrocatalytic activity for phenolic compound oxidation of Ti3C2Tx MXene was confirmed by testing the electrocatalytic performances with different phenolic substrates, including 1-naphthol, 4-nitrophenol, and β-estradiol. As shown in Fig. 2 the modification of MXene significantly improved the oxidation currents for all the three phenolic compounds (DPV curves were seen in Supplementary Material, Fig. S2), indicating the favourable and comprehensive electrocatalytic activity of Ti3C2Tx MXene for phenolic compound oxidation. Moreover, it's notable that the oxidation peaks presented distinct shifts to lower potential, revealing the electron transfer between Ti3C2Tx MXene and phenolic compound.
To further profile the unique electrocatalytic activity of Ti3C2Tx MXene, several nanomaterials including MoS2 nanosheets, WS2 nanosheets, Fe3O4 nanoparticles, TiO2 nanoparticles and bulk Ti3AlC2 were separately employed as constrast with 1-naphthol as substrate. It's notable that only Ti3C2Tx MXene presented observable electrocatalytic activity for 1-naphthol oxidation, indicating that the unique electrocatalytic activity was the intrinsic property of Ti3C2Tx MXene.
To quantitatively present the electrocatalytic activity of Ti3C2Tx MXene for 1-naphthol oxidation, DPV curves were measured with the addition of 1-naphthol at different concentrations. As shown in Fig. 3A, the peak current increased with the increasing concentration of 1-naphthol at low concentrations until reaching about 90 µA. According to Faraday's laws of electrolysis and Michaelis-Menten equation, the fitting curve was achieved with hyperbola function, where the Michaelis constant was calculated to be 0.22 mM, indicating that Ti3C2Tx MXene possessed strong affinity to 1-naphthol . Moreover, the catalytic activity of Ti3C2Tx MXene for 1-naphthol oxidation in homogeneous phase solution was explored using hydrogen peroxide as oxidant. Fig. 3B displayed the real-time absorbance of the aqueous solutions at 387.5 nm, including 1-naphthol (curve a), mixture of 1-naphthol and hydrogen peroxide (curve b), mixture of 1-naphthol, hydrogen peroxide and Ti3C2Tx MXene (curve c), respectively. It could be seen that the reaction ratio was significantly improved with the addition of Ti3C2Tx MXene, indicating that Ti3C2Tx MXene could efficiently catalyze the 1-naphthol oxidation by hydrogen peroxide as well  .
To survey the quantitative relation between electrocatalytic activity and amount of Ti3C2Tx MXene, Ti3C2Tx MXene solutions at different concentrations were employed to modify GCE to test the electrocatalytic activities for 1-naphthol oxidation, respectively. As shown in Fig. 3C, the Faradic current increased linearly with the increasing concentration of Ti3C2Tx MXene in low concentration range and reached a constant in high concentration range, indicating that the electrocatalytic activity was in direct proportion to the total modified area of Ti3C2Tx MXene flakes but not further improved by the stacking of the flakes.
3.3. Mechanism study for the catalytic activity of Ti3C2Tx MXene
The first principle calculations are performed to explain the catalytic activity of Ti3C2Tx MXene for 1-naphthol oxidation. The free energies of 1-naphthol adsorption were first calculated to determine whether 1-naphthol could be adsorbed by Ti3C2Tx MXene. The optimized geometries of 1-naphthol on and Ti3C2Tx MXene were given in Fig. S3 (see in Supplementary Material), which showed 1-naphthol molecule adsorbed on the plane surface of Ti3C2Tx MXene in a "lying-down" or "standing-up" manner, with adsorption energies of -1.18476 and -0.75625 eV, respectively, indicating relative strong physical adsorption. As a result, the "lying-down" adsorption of 1-naphthol on Ti3C2Tx MXene was critical in the electrocatalytic oxidation. To further investigate the origin of 1-naphthol adsorption, the differential charge density of 1-naphthol adsorbed on Ti3C2Tx MXene was calculated. As shown in Fig. 3D, the changes in charge density caused by 1-naphthol adsorption mainly came from hydroxyl group and oxygen atoms. Moreover, changes in charge density were also found on the aromatic rings, indicating that the aromatic structure played a role in leading to the "lying-down" adsorption mode of 1-naphthol on Ti3C2Tx MXene.
3.4. Characterization of the DNA walking machine
To quantify the efficiency of DNA Walking machine, fluorescence dynamics experiments were carried out to verify it. Firstly, AuNPs and ssDNA-functionalized AuNPs were prepared according to previous literature with little adjustment . In short, sodium citrate (3 mL, 1%) was added rapidly to the boiling solution of HAuCl4 (100 mL, 1%). After the color changed from pale yellow to wine-red, the mixture was stopped heating and cooled to room temperature (RT) with continued stirring. The AuNPs had an average particle size of 13 nm and were stored at 4°C for further use. The preparation method of ssDNA (support probe and walker-protect dsDNA) -functionalized AuNPs is as follows. Firstly, the 1428 µL denature-supporting probe (2 µM) and 72 µL denature-walker-protect dsDNA (2 µM) were mixed with 1 µL acetic acid (500 mM, pH 5.2) and 0.5 µL TCEP (100 mM) at RT for 1 h, respectively. Then the mixture was added into 1 mL AuNP solution, and the resultant solution was stored in a drawer at RT for at least 16 h. After 25 µL Tris-acetate (500 mM, pH 8.2) was added to the mixture, 250 µL NaCl (1 M) was dropwise added to the mixture every three hours (30, 40, 50, 60, 70 µL were added respectively). Subsequently, the resulting mixture was stored in a drawer overnight. Lastly, the mixture was centrifuged (10 000 rpm, 10 min) to remove the excess reagents, and the red precipitate was washed and dispersed in DNA preparation solution for further use.
Secondly, fluorescence kinetics curve was shown in Fig. 4A to prove the cutting efficiency of Nt.BsmAI nicking endonuclease (Nt.BsmAI). The curve a, b and c showed corresponding changes when different concentrations of target BCR/ABL fusion gene and 10 U Nt.BsmAI were added into the ssDNA (support probe and walker-protect dsDNA) -functionalized AuNPs solution, respectively. Curve d was the blank control. The slope of the curve reflects the reaction rate of enzyme shearing, which is correlated with the concentration of target gene. As can be seen from the figure, the reaction rate is fast. At the same time, when the reaction reached 2 hours, the shearing enzyme still did not reach the maximum shearing value, which means that the shearing enzyme has not been completely reacted. Thus, the cutting efficiency of the Nt.BsmAI is excellent. In addition, the time of releasing hairpin structure DNA can also be known from the curve. The sharply rising stage in the curve mainly the enzymatic cleaving on the prehybridized Support DNA-Walker. It can be seen from the figure that it takes about 6-8 minutes to cleave and release the hairpin. The slope of tangent at t = 0s was calculated with the fitting curves indicating the cleaving rate was highly related to the concentration of substrate in Fig. 4B. The slope of tangent at t = 0s and logarithmic value of target BCR/ABL fusion gene concentrations presents well linear dependence range from 2 pM to 2 µΜ with pearson correlation coefficient of 0.99192, which is corresponding to the kinetic characteristic of first-order reaction.
Thirdly, as for the time of hairpin structure intermediate DNAs hybridized with the thiolated capture DNA on the electrode surface to form the sensing interface, we also conducted relevant optimization experiments. As shown in Fig. 4C, the time needed was only 1h.
Lastly, we have consulted relevant literature, which shows that the amount of DNA loading on 15 nm gold nanoparticles is 20-30 when the concentration of NaCl is 140mM . According to the concentration and dosage of AuNPs, the amount of ssDNA fabricated on the interface of this sensor was calculated to be about 1×1010.
3.5. Optimization of experimental conditions
To achieved optimal analytical performance of the biosensor, some critical experimental conditions were optimized, including the ratio of walker probe to support probe, the cleaving time of Nt.BsmAI nicking endonuclease, the pH of DEA buffer and the cleaving temperature of Nt.BsmAI nicking endonuclease. As shown in Fig. 5, optimal ratio of walker probe to support probe, cleaving time, pH of DEA buffer and temperature were achieved to be 1:20, 120 min, 9.6 and 37°C, respectively.
3.6. Analytical performance of the proposed electrochemical biosensor
To estimate analytical performance of the biosensor, the current responses toward BCR/ABL fusion gene at different concentrations were recorded under the optimal conditions through DPV measurements. As shown in Fig. 6A, the detection signal increased with the increasing concentration of target BCR/ABL fusion gene. The corresponding calibration plots of the peak currents showed a strong linear relationship to the logarithm value of target BCR/ABL fusion gene concentrations range from 0.2 fM to 20 nΜ with pearson correlation coefficient of 0.99836 (Fig. 6B). The linear regression equation was I = 1.00012 × lg (c/pM) + 11.23074 (c and I stood for the concentration of target BCR/ABL fusion gene and corresponding peak current value, respectively). The limit of detection was obtained based on three times the average standard deviation corresponding to blank sample detection, which was calculated to be 0.05 fM. Comparisons of this biosensor with some reported works for BCR/ABL fusion gene detection are shown in Table S2, which highlighted the excellent sensitivity of this method in BCR/ABL fusion gene detection due to the cascading catalytic strategy and DNA walking machine for signal amplification.
Moreover, the specificity of the biosensor was evaluated by using 3 different DNA oligonucleotides as references, including a single-base-mismatched strand (B1), a two-base-mismatched strand (B2) and a noncomplementary strand (B3), all at concentrations of 20 fM. As depicted in Fig. 6C, the response signals of the single-base-mismatched strand and two-base-mismatched strand were much lower than the response signal of the target, revealing the good capacity of the biosensor to distinguish base-mismatch. The response signal of noncomplementary sequences were approximate to the blank solution, indicating that the biosensor presented good selectivity for DNA detection.
To evaluate the stability of the proposed biosensor, the modified electrodes were stored at 4°C before use. As presented in Fig. 6D, there were no obvious differences during the first 5 days of storage, and the current changes were less than 1.58%. After 20 days of storage, the designed biosensor retained 89.40% of its initial current response, indicating that the proposed biosensor offers satisfactory stability for target BCR/ABL detection.
3.7. Detection of BCR/ABL fusion gene in human serum samples
To further validate the applicability of the biosensor to complex biological matrix in clinical application, different concentrations of target BCR/ABL fusion gene were added to 10-fold-diluted clinical serum samples and tested with the proposed biosensor. The detection results of BCR/ABL fusion gene in human serum samples are summarized in Table S3. Satisfactory recovery values were obtained ranging from 93.60–110.42% with relative standard deviations (RSD) between 0.27% and 0.64%. In addition, we extracted RNA from clinical serum of BCR/ABL positive patients using spin columns CB3 according to the manufacture's protocol and tested with the proposed biosensor. The detection concentrations of BCR/ABL were compared with clinical results (by reverse transcription PCR), which were summarized in Table S4. It could be seen that the proposed biosensor achieved results well matched with the clinical assay, manifesting the application potential of the proposed biosensor in clinical diagnosis.