Multifunctional MOF-based Electrochemiluminescent Nanocubes for an Ultrasensitive Biosensing Strategy of B. Pseudomallei Determination Coupled With 3D Magnetic Walking Nanomachine


 Burkholderia pseudomallei (B. pseudomallei) can cause melioidosis that is usually fatal. A reliable and rapid detection method is greatly needed for disease surveillance and diagnosis. Herein, an ultrasensitive electrochemiluminescence (ECL) biosensor was constructed for accurate determination of B. pseudomallei coupled with multifunctional Au@Co-MOF@ABEI nanocubes and 3D magnetic walking nanomachine. The synthesized nanocubes could not only immobilize enormous ABEI but exhibit superior peroxidase-like activity to decompose H2O2 to produce plentiful reactive oxygen species (ROSs) that could further react with ABEI, so that the enhanced ECL signals were achieved. Meanwhile, the target-activated walking nanomachine was efficiently driven by Exonuclease III (Exo III) for further improving the sensitivity of the biosensor. As a result, the fabricated ECL biosensor could detect pathogenic gene down to 60.3 aM with a linear range from 100.0 aM to 100.0 pM. Moreover, the biosensing platform successfully achieved the determination of B. pseudomallei down to 9.0 CFU mL−1 in serum samples. This work exhibited an ultrasensitive and specific performance for B. pseudomallei detection, which would become a versatile tool in the early diagnosis and treatment of melioidosis.


Introduction
B. pseudomallei, the pathogen of melioidosis, has been one of the most widespread bacteria responsible for extremely poisonous and tropical diseases [1][2][3]. It is estimated that B. pseudomallei is endemic in approximately 45 countries including multiple subtropical and tropical regions [4,5]. Considering its higher fatality, early and sensitive detection of B. pseudomallei is urgently needed for disease surveillance and diagnosis melioidosis. At present, several traditional methods have been utilized for the determination of B. pseudomallei, such as conventional culture methods [6], polymerase chain reaction (PCR) [7], and enzyme-linked immunosorbent assay [8]. Although these approaches are powerful and reliable, most of them are expensive, labor-intensive, and time-consuming [9,10]. Therefore, a detection method which can simple, cost-effective, and sensitive determination of B. pseudomallei in the early period of infection is urgently required.
In recent years, ECL biosensor has been veri ed to be an alternative platform for bacteria sensing, resulting from the merits of simple operation and low cost [11][12][13]. N-(4-aminobutyl)-N-(ethylisoluminol) (ABEI), as a typical derivative of luminol, has been widely used in the ECL reaction system (ABEI/H 2 O 2 ) due to its good stability, low toxicity, and easy modi cation [14][15][16]. Given the consensus that nanozymes can signi cantly enhance the ECL emission of ABEI/H 2 O 2 by loading enormous ABEI and catalyzing H 2 O 2 to produce ROSs owing to their high surface area and good peroxidase-like activity, various nanozymes have been employed [17][18][19][20]. For instance, Yuan' group proposed ABEI functionalized Fe-MIL-101 that is one of metal-organic frameworks (MOFs), revealing an enhanced ECL performance [21]. Furthermore, Jin' group fabricated luminol-capped Ag nanoparticles (NPs)@Co-MOF, demonstrating that metal-doped MOF could further improve the ECL intensity of luminophore [22]. Inspired by these valuable reports, we speculate that Au NPs coated Co-MOF is capable of enhancing the emission of ABEI, which contributes to develop high-performance ECL biosensors.
Biological systems take advantage of highly complicated and hierarchical machineries to operate various functions [23,24]. Inspired by natural phenomenon, 3D DNA walking nanomachines have been designed and applied in the eld of biosensing for the rapid transduction and ampli cation of signals because of the increased local concentration of DNA strands [25][26][27]. Usually, the autonomous motion of the nanomachines is propelled by DNAzyme, protease, and strand displacement [28][29][30]. It is worth noting that the signal ampli cation ability of DNAzyme and strand displacement-propelled nanomachines is restricted owing to the low catalytic e ciency of DNAzyme and the random collision of drive strand and track [31,32]. In contrast, protease-powered nanomachines are promising alternatives for constructing 3D magnetic walking nanomachines that can achieve rapid walking processivity and avoid the interference from environmental factors.
Herein, combining 3D Exo III-propelled magnetic walking nanomachine and Au@Co-MOF@ABEI nanocubes, a pragmatic ECL biosensing platform for an ultrasensitive detection of B. pseudomallei was established (Scheme 1). For process A, multifunctionalized Au@Co-MOF nanocubes were synthesized by a simple wet-chemical approach and then used to loading numerous ABEI for forming Au@Co-MOF@ABEI nanocubes. For process B, the designed 3D magnetic walking nanomachine consisted of three aspects: the biotin-modi ed DNA duplex probes with the exposed 3' end (PP/DWT as a walker and S3/DWT1 as DNA track) immobilized on the same magnetic bead (MB), target B. pseudomallei gene as energy input, and Exo III as a driver. The presence of target gene triggered the release of the DWT from PP/DWT duplex, which could then displace S3 to form DWT/DWT1 with blunt 3'end. Subsequently, Exo III digested the DWT1 of DWT/DWT1 and released the DWT again. In this way, the DWT autonomously moved along the designed 3D track, resulting in the release of numerous S3. For process C, the ECL biosensor was xed on the glassy carbon electrode (GCE) which was deposited with Au@Co-MOF@ABEI nanocubes to achieve a high ECL emission ("signal-on" state). Subsequently, dopamine-labeled S2 strands (S2-DA) bound with S1 to su ciently quench the initial ECL signal ("signal-off" state). Finally, the released S3 displaced S2-DA to recover the quenched ECL signal ("signal-on" state). Bene ting from Au@Co-MOF@ABEI nanocubes and 3D magnetic walking nanomachine, the ECL biosensor for B. pseudomallei gene determination exhibited remarkable analytical performance, which provided a novel platform for pathogenic bacteria determination.
The oligonucleotide DNAs were synthesized and puri ed by Sangon Biotechnology Co., Ltd. (Shanghai, China). All sequences are listed in Table S1

Apparatus
The multifunctional MPI-E analyzer (Xi'An Remax Electronic Science & Technology Co., Ltd. Xi'An, China) was utilized to carry out ECL analysis with the voltage of the photomultiplier tube at 800 V. Cyclic voltammetry (CV) and electrochemical impedance spectroscopy (EIS) measurements were taken with a CHI660D electrochemistry workstation (Shanghai Chenhua Instruments, China). Three kinds of electrodes were included in the measuring system, referred to a modi ed GCE (working electrode), a platinum wire (counter electrode), and an Ag/AgCl electrode (reference electrode). Fluorescence signals were obtained using the Cary Eclipse Fluorescence spectrophotometer (Agilent, California). The PCR analysis was performed with a Cycler thermal cycler (Bio-Rad Laboratories, USA). X-ray photoelectron spectroscopy (XPS) spectra was registered with an Thermo ESCALAB 250Xi (USA). X-ray diffraction (XRD) patterns were gained with a D8 ADVANCE X-ray diffractometer (Bruker, Germany) to con rm the structure of the materials. The morphology characterization of different nanomaterials was characterized by Oxford xmax energy dispersive spectroscopy (EDS, Berlin, Germany) and SU8010 scanning electron microscopy (SEM, Berlin, Germany).

Fabrication of 3D magnetic walking nanomachine
First, 10 µL of biotin-modi ed DNA walker stand (DWT, 10 µM) and 10 µL of protecting probe (PP, 10 µM) were dissolved in the 100 µL of 20 mM Tris-HCl buffer at 37 ºC for 1 h to form PP/DWT duplex strand (PWT, 1 µM). Meanwhile, 10 µL of biotin-modi ed DNA track stand (DWT1, 10 µM) and 10 µL of output stand (S3, 10 µM) were also dissolved in the 100 µL of 20 mM Tris-HCl buffer at 37 ºC for 1 h to form S3/DWT1 double-stranded track probes (SDs). Second, 10 µL of SMB was washed three times with washer buffer 1 and then redispersed in 10 µL of 20 mM Tris-HCl buffer. Third, to combine DNA strands with SMB, 100 µL of PWT and SDs at the molar ratio of 1 to 20 was su ciently mixed with 10 µL of the redispersed SMB for 1 h at 37°C. Subsequently, the formed PWT-SDs-SMB was successively washed three times with 20 mM Tris-HCl buffer after magnetic separation. Then, the 3D magnetic walking nanomachine was rstly initiated by adding 10 µL of target gene with various concentrations, 5 µL of PWT-SDs-SMB, 10 µL of 10 × NEBuffer 1 (100 mM bis Tris Propane-HCl, 100 mM MgCl 2 , 10 mM dithiothreitol), and 72.5 µL of ultrapure water for 30 min at 37°C. After that, 2.5 µL of 100 U mL −1 Exo III was mixed for brief vortex and further incubated at 37°C for 1.5 h to perform the motion of walker. Finally, numerous S3 were obtained after magnetic separation, and the Exo III was inactivated at 70 ºC for 30 min.

Preparation of Au@Co-MOF@ABEI nanocubes
The preparation steps of Co-MOF were described according to previously reported method with minor alteration [33]. Typically, 5 mL of aqueous solution A was obtained by mixing up 0.0375 g Co(NO 3 ) 2 ·6H 2 O, 0.0551 g Na 3 C 6 H 5 O 7 ·2H 2 O, and 5 mL of ultrapure water. Meanwhile, 7.5 mL of aqueous solution B was acquired by adding 0.033 g K 3 [Co(CN) 6 ] into 7.5 mL of ultrapure water. Then, solution A and solution B were mixed for brief ultrasound at room temperature (RT) overnight. In the following, the precipitate with pink color was washed with absolute ethanol by centrifugation (12000 rpm, 5min) at least three times. Finally, the pink Co-MOF powders were collected after drying at 60 ºC for 2 h.
The preparation of Au@Co-MOF@ABEI nanocubes was depicted as follows. In brief, the synthesized Co-MOF (0.025 g) and 1 ml of APTEs were mixed into 20 mL of ultrapure water and stirred for 1.5 h to decorate Co-MOF with abundant amino groups. After that, 0.25 mL of HAuCl 4 ·6H 2 O (1.0 wt%) and 0.00125 g PVP were mixed into the preceding solution and stirred for 15 min. In the following, 10 µL of fresh NaBH 4 (1.9 M) was dropped into the mixed solution and stirred for 1.5h to obtain Au@Co-MOF.
Subsequently, the resulted Au@Co-MOF was obtained by centrifugation and washed with absolute ethanol for three times. After that, the Au@Co-MOF was re-suspended in 20 mL of ultrapure water. Then, 1 mL of ABEI (10 mM) was added into the Au@Co-MOF solution with full stirring for 30 min. The mixed solution was washed with ultrapure water to eliminate unreacted ABEI. Finally, the obtained Au@Co-MOF@ABEI nanocubes were dissolved in 0.1% chitosan solution and kept at 4°C for further use.

Fabrication of the switching ECL biosensor
Prior to use, GCE was polished with 0.3 and 0.05 µm Al 2 O 3 slurry orderly, and sonicated with ultrapure water to attain a mirror-like surface. After drying in the nitrogen, the cleaned GCE was decorated by 10 µL of Au@Co-MOF@ABEI nanocubes and dried at RT to form nano lm. Then, 10 µL of S1 (1 µM) was added to the modi ed electrode at 4°C overnight. Afterward, the modi ed electrode was incubated with 5 µL of MCA (1 µM) to block the remaining active sites for 30 min. Subsequently, 10 µL of S2-DA was dropped onto the modi ed electrodes at 37°C for 1 h. After the hybridization between S1 and S2-DA, the excess S2-DA was eliminated by the washer buffer 2. Then, 10 µL of the released S3 was hybridized with the S2-DA for 1.5 h. Washer buffer 2 was used to clean the electrode after each construction step, respectively, and the obtained electrode was preserved at 4°C for later use. Finally, the established ECL biosensor was tested in 2 mL PBS (0.1 M, pH 8.0) including H 2 O 2 (3 mM) to record the ECL signal.

Results And Discussion
Characterization of Au@Co-MOF@ABEI nanocubes The size and surface morphologies of the as-synthesized nanomaterials were characterized by SEM. As showed in Fig. 1A, the Co-MOFs illustrated a nanocube with a side length of about 500 ± 10 nm and a smooth surface, which was favorable for further modi cation. Fig. 1B showed that Au NPs were coated on Co-MOFs and had an average diameter of 14 ± 2 nm (Fig. 1C), which agreed with the typical XRD pattern of Co-MOF and Au@Co-MOF (Fig. S1). These results indicated the successful synthesis of Au@Co-MOF@ABEI nanocubes. In addition, the element distribution of Au@Co-MOF@ABEI nanocubes was veri ed by EDX elemental mapping with the high-angle annular dark-eld scanning electron microscopy (HAADF-SEM) in Fig. 1D. As shown in Fig. 1(E-I), the occurrence of C, N, O, Au, and Co elements showed that Au NPs on the surface of Co-MOFs were successfully synthesized. Subsequently, XPS analysis was utilized to further investigate the electronic state and elemental composition of the Au@Co-MOF@ABEI. As illustrated in Fig. 1(J-L), the peaks at 285.0, 398.4, and 531.8 eV belonged to C1s, N1s, and O1s, respectively. The characteristic peaks at 84.3 eV and 88.0 eV were originated from Au 4f peaks, manifesting that the metallic Au adhered to the surface of the nanocubes [34]. As revealed in the high-resolution spectra of Co 2p, the characteristic peaks at 799.9, 797.1, 784.4, and 781.9 eV could be indexed to metallic Co [35], which was in accord with the SEM-EDX mapping results, further indicating that Au@Co-MOF@ABEI nanocubes were successfully synthesized as expected.
The mechanism of the ECL biosensor To con rm the superiority of the Au@Co-MOF@ABEI nanocubes in ECL intensity, a contrast experiment was implemented by comparing the performance of Au@Co-MOF@ABEI to those of Co-MOF@ABEI, Au@ABEI, and ABEI. As illustrated in Fig. 2A, the ECL intensity obtained by diverse MOF-based nanomaterials followed the order of Au@Co-MOF@ABEI nanocubes > Co-MOF@ABEI > Au@ABEI > ABEI. Speci cally, only a weak ECL signal (1648 a.u.) was detected when ABEI was dropped onto the cleaned electrode (curve a), the Au@ABEI-modi ed GCE produced ECL signal about 3293 a.u. (curve b), and the Co-MOF@ABEI-modi ed GCE generated ECL signal about 6872 a.u. (curve c). However, when Au@Co-MOF@ABEI nanocubes was preferred to act as ECL luminophore, a remarkably higher ECL signal (12392 a.u.) was observed by the ECL biosensor (curve d). These results indicated that the multifunctionalized Au@Co-MOF@ABEI nanocubes containing Au NPs and Co-MOF could greatly enhance the ECL of ABEI.
In addition, the "on-off-on" mode was investigated. As displayed in Fig. 2B, the Au@Co-MOF@ABEI nanocubes was rstly immobilized on the electrode surface to obtain an enhanced ECL signal (12362 a.u.), owing to the decomposition of Au@Co-MOF toward H 2 O 2 to produce many ROSs including superoxide radical (O 2 . − ) and hydroxyl radical (OH.) that could further react with ABEI ("signal-on" state).
Next, S1/S2-DA duplex probes were dropped onto the modi ed electrode, leading to the obvious decrease of the initial ECL signal (1692 a.u.), resulting from the quenching effect of DA toward ABEI luminescence ("signal-off" state). Then, S2-DA was replaced by the output S3, leading to the recovery of the ECL response (9540 a.u.), indicating that the 3D magnetic walking nanomachine was successfully fabricated ("signal-on" state). Therefore, these results indicated that the established "on-off-on" ECL biosensor possessed the advantages of excellent ECL performance and obvious signal-to-noise ratio (S/N).  Fig. 2C, an initial CV peak was obtained on the bare GCE (curve a). When the Au@Co-MOF@ABEI (curve b) was immobilized onto the electrode, the CV peak decreased, given a poor conductivity of the nanocubes as an ECL luminophore. After the S1 was immobilized onto the GCE/Au@Co-MOF@ABEI, the current signal decreased due to the electronic repulsion between the negatively charged DNA and [Fe(CN) 6 ] 3−/4− (curve c). As expected, the current response was greatly reduced after hybridizing with the S2-DA (curve d), because there was repulsive effect from abundant S2-DA. Subsequently, peak current dramatically increased and characteristic peak gap was signi cantly separated (curve e), because S3 was dropped on the prepared electrode to displace S2-DA, which largely decreased negative charges of the electrode surface. The CV results were fairly consistent with the results obtained from EIS (Fig. 2D), in which the R et values changed upon the immobilization, hybridization, and displacement process. These results indicated the feasibility of the constructed biosensor.

Optimization of the experimental conditions
Firstly, the 3D magnetic walking nanomachine was successfully fabricated and veri ed the feasibility by uorescent experiment (Fig. S2). Then, experimental parameters of the walking nanomachine were optimized ( Fig. S3A and 3B). On this basis, to obtain excellent analytical capability for B. pseudomallei gene detection, different experimental conditions of the ECL biosensor were further optimized. the ECL intensity and S/N were used to evaluate the performance of the developed biosensor. As shown in Fig. 3A, the ECL intensity and S/N for 10.0 nM target gene increased with the increasing the molar ratio of PWT to SDs and then decreased obviously after 1:20, which was then chosen as the optimal ratio. As could be seen in Fig. 3B, the ECL intensity and S/N increased with the increasing concentration of Exo III in the range from 100 to 250 U mL −1 , and then showed a stable tendency. Therefore, 250 U mL −1 was used as the optimized concentration of Exo III.
Furthermore, the concentration of DA was vital in the quenching effect of the ECL responses. As exhibited in Fig. 3C, after introducing 0.31 µM DA, the S/N was maximum. As shown in Fig. 3D, the incubation time of S3 and S2-DA was studied, and the results indicated that the ECL signals increased with the incubation time from 0.5 to 1.5 h, while the signal slightly increased when the incubation time was longer than 1.5 h. Therefore, 1.5 h was selected as the optimized time.

Analytical performance of the ECL biosensor
Under the optimal experimental conditions, the dependence of ECL intensity was investigated to evaluate the analytical performance of the ECL Biosensor. Fig. 4A showed the good linear relationship between the ECL intensity and concentration of B. pseudomallei gene in the range of 0, 100.0 aM to 100.0 pM (curve a-h). As shown in Fig. 4B, The corresponding calibration plot between the ECL intensity and the logarithm (lg) of the B. pseudomallei gene concentration showed a linear regression equation of I = 1362.9 lgc -627.9 (R 2 = 0.9958). The limit of detection (LOD) of 60.3 aM was evaluated based on a 3σb/slope, the σb stands for the standard deviation of three blank samples, which was about 2 orders of magnitude lower than that (4.97 fM) of the individual ECL system without the nanomachine ( Fig. S4A and B). In addition, the LOD of the integral ECL platform was about 5 orders of magnitude lower than that (0.34 pM) of the individual walking nanomachine ( Fig. S4C and D). The ECL biosensor exhibited a lower LOD, resulting from multifunctionalized Au@Co-MOF@ABEI nanocubes and effective magnetic walking nanomachine.
Moreover, these results indicated that the "off-on" mode possessed the superiorities of high sensitivity, obvious S/N, and wide range of detection. Therefore, the constructed ECL biosensor was a versatile tool for an ultrasensitive detection of B. pseudomallei gene.
The stability of the ECL biosensor is important for clinical diagnosis. As showed in Fig. 4C, an ECL signal for target gene within 100 s was obtained with the relative standard deviation (RSD) of 2.15%, manifesting that the biosensor obtained eminent stability. In order to evaluate the speci city of the constructed ECL biosensor, nine diverse mismatched synthetic DNAs were examined. As shown in Fig. 4D, the ECL intensity for T1 was the highest among other mismatched DNAs, containing T1 (1-base mismatch), T2 (2-base mismatch), T3 (3-base mismatch), T4 (4-base mismatch), T5 (5-base mismatch), T6 (6-base mismatch), T7 (7-base mismatch), T8 (8-base mismatch), and T9 (9-base mismatch). The ECL intensity for T9 was close to that for the blank space. These results indicated that the constructed ECL biosensor could discriminate different mismatched DNA effectively and showed good selectivity.

Real samples analysis
To evaluate whether the constructed ECL biosensor could detect practical sample, sensitively and speci cally, the B. pseudomallei were cultivated into human serum at the concentrations from 0 to 1.5 x 10 8 CFU mL −1 . After a preprocessing step for the human serum samples, the PCR analysis was then executed by using extracted genomic DNA for each concentration of B. pseudomallei. Meanwhile, the established ECL biosensor was also carried out to detect the denatured PCR products of B. pseudomallei. The ECL intensity of the constructed ECL biosensor to different PCR samples were illustrated in Fig. 5A.
As seen in Fig. 5B, the linear regression equation between the ECL responses and B. pseudomallei concentration was expressed as I = 912.82 c + 766.20, in which I represents for ECL intensity, c stands B.
pseudomallei concentration (CFU mL −1 ) with a correlation coe cient of 0.9994. The ECL biosensor could analyze B. pseudomallei down to 9.0 CFU mL −1 in real serum samples, which was much lower than that of previous methods for pathogenic bacteria analysis (Table S3). Therefore, The established biosensing platform showed high sensitivity, speci city, and speediness, exhibiting the potential as a pragmatic tool for B.pseudomallei determination in real samples.
The stability of the ECL biosensor for B. pseudomallei detection was an important factor for clinical diagnosis in real samples analysis. As illustrated in Fig. 5C, the stability of the ECL biosensor was studied by using the RSD of the ECL intensity for ve concentrations (1.5 x 10 1 CFU mL −1 , 1.5 x 10 2 CFU mL −1 , 1.5 x 10 3 CFU mL −1 , 1.5 x 10 4 CFU mL −1 , and 1.5 x 10 5 CFU mL −1 ) of B. pseudomallei (a-e). The RSD were calculated to be 4.02%, 0.84%, 0.84%, 1.55%, and 0.21% (n = 3). This result demonstated that the developed ECL biosensor possessed good stability. In addition, to evaluate the speci city of constructed ECL biosensor, the PCR products for different types of pathogenic bacteria were detected, including 1.5 x 10 8 CFU mL −1 of blank, E. coli, S. aureus, mixture and B. pseudomallei, respectively. As show in Fig. 5D, the ECL intensity of B. pseudomallei detection was much larger than those of other detection, while the intensity of E. coli and S. aureus detection were close to that of blank, and these results indicated that the constructed ECL biosensor showed good speci city for an ultrasensitive detection of B. pseudomallei. Therefore, the high stability and speci city could further con rm the practicality of the developed biosensor.