A Novel Electrochemiluminescence Immunosensor Based on Resonance Energy Transfer between g-CN and NU-1000(Zr) for Ultrasensitive Detection of Ochratoxin A in Coffee

In this study, an electrochemiluminescence (ECL) immunosensor based on nanobody heptamer and resonance energy transfer (RET) between g-C3N4 (g-CN) and NU-1000(Zr) was proposed for ultrasensitive ochratoxin A (OTA) detection. First, OTA heptamer fusion protein was prepared by fusing OTA-specific nanometric (Nb28) with a c-terminal fragment of C4 binding protein (C4bpα) (Nb28-C4bpα). Then, Nb28-C4bpα heptamer with the high affinity used as a molecular recognition probe, of which plenty of binding sites were provided for OTA-Apt-NU-1000(Zr) nanocomposites, thereby improving the immunosensors’ sensitivity. In addition, the quantitative analysis of OTA can be achieved by using the signal quenching effect of NU-1000(Zr) on g-CN. As the concentration of OTA increases, the amount of OTA-Apt-NU-1000(Zr) fixed on the electrode surface decreases. RET between g-CN and NU-1000(Zr) is weakened leading to the increase of ECL signal. Thus, OTA content is indirectly proportional to ECL intensity. Based on the above principle, an ultra-sensitive and specific ECL immunosensor for OTA detection was constructed by using heptamer technology and RET between two nanomaterials, with a range from 0.1 pg/mL to 500 ng/mL, and the detection limit of only 33 fg/mL. The prepared ECL-RET immunosensor showed good performance and can be successfully used for the determination of OTA content in real coffee samples, suggesting that the nanobody polymerization strategy and the RET effect between NU-1000(Zr) and g-CN can provide an alternative for improving the sensitivity of important mycotoxin detection.


Introduction
Ochratoxin A (OTA) is one of the significant mycotoxins, a toxic secondary metabolite produced by Aspergillus and Penicillium, which is widely distributed in nature [1]. It is easy to contaminate many types of foods, including grains, coffee beans, and spices. Furthermore, OTA with the stable chemical properties is not easily degraded quickly, making it difficult to eliminate once it enters the body through food [2]. Long-term OTA accumulation will cause irreversible harm to humans and animals, leading to deformities and cancer [3,4]. Today, OTA is a major mycotoxin that International Agency for Research on Cancer (IARC) has designated as a human IIB carcinogen. To reduce the risk of OTA to human health, organizations and countries have established the maximum limit of OTA in food [5].
Coffee is a significant cash crop in China, particularly in Hainan province. However, due to Hainan's warm climate, the coffee is susceptible to OTA contamination. As a result, coffee pollution has caused widespread concern among researchers [6,7]. In our country, the maximum OTA limit for coffee beans or coffee powder is 5.0 µg/kg, and the maximum

Carboxylation and Activation of g-CN
The carboxylation of g-CN was carried out using the method reported in the literature [31]. First, 1 g g-CN powder was placed in 100 mL 5 M HNO 3 and refluxed at 125 • C for 24 h. After natural cooling to 25 • C, the refluxed products were centrifuged. Then, products were washed with ultrapure water. Carboxylated g-CN was obtained 12 h after being vacuum-dried at 35 • C. Then, a mixture of 1.5 mL 0.4 M EDC and 0.1 M NHS was added to 3 mg g-CN and shaken for 6 h on a 200 rpm on a thermostatic shaker before being washed and centrifuged 3 times. The activated g-CN was re-dispersed in 1.5 mL 0.001% chitosan acetic acid solution and stored at 4 • C for further use.

Expression, Purification, and Identification of Nb28-C4bpα Fusion Proteins
As previously reported, we expressed Nb28-C4bpα fusion protein [11]. To obtain the fusion proteins, the E. coli Rosetta chemically competent cells containing the vector pET25b-Nb28-C4bpα was used for auto-induction. First, the strain was inoculated in LB medium (containing 100 µg/mL of ampicillin) and then incubated overnight at 37 • C with shaking at 250 rpm until OD 600 reached 0.5-0.7. Bacterial cells were collected by centrifugation after Nb28-C4bpα was expressed in 25 • C conditions with intense shaking overnight. The cells were then resuspended in a 20 mL equilibration buffer (1 mM PMSF, 8 mM Na 2 HPO 4 , 2 mM KH 2 PO 4 , 2.6 mM KCl, 136 mM NaCl, and 60 mg lysozyme). The resuspended E. coli cells were subjected to ultrasound in an ice bath to prevent degradation or denaturation of the target proteins. Soluble Nb28-C4bpα fusion proteins were obtained by centrifugation (8000× g, 4 • C) and filtered through a syringe filter with a 0.22 µm pore size. Finally, Ni-NTA Sepharose and PBS were used to purify and perform dialysis fusion on proteins. The purity and concentration of fusion protein were analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and microdroplet ultramicro spectrophotometer, respectively.

Fabrication
Process of ECL-RET Immunosensor between g-CN and NU-1000(Zr) Scheme 1 describes the assembly and recognition process of the fabricated ECL-RET immunosensor. First, glassy carbon electrodes (GCE) were polished and cleaned to obtain a bright mirror surface. Then, 5 µL activated g-CN (2 mg/mL) nanosheets were added dropwise to the surface of GCE and dried naturally. To capture the target molecule, 5 µL Nb28-C4bpα heptamer solution was coated, and incubated at 37 • C for 1 h. Then, the modified electrode was washed with ultrapure water to remove the unbound Nb28-C4bpα heptamer molecule. During this process, the amino group of Nb28-C4bpα forms a covalent bond with the carboxylated g-CN. Nb28-C4bpα/g-CN/GCE was then incubated with BSA solution (5 µL, 1%) at 37 • C for 0.5 h to block the non-specific binding sites. After careful cleaning, the obtained BSA/Nb28-C4bpα/g-CN/GCE modified electrode incubated in OTA concentrations of 37 • C for 2 h, then gently washed with distilled water and dried naturally. Finally, 5 µL OTA-Apt-NU-1000(Zr) was dripped onto the modified electrode and incubated at 37 • C for 1 h, and then rinsed with ultrapure water. The OTA-Apt-NU-1000(Zr)/OTA/BSA/Nb28-C4bpα/g-CN/GCE ECL-RET immunosensor was stored at 4 • C for future use.

Characterizations of g-CN and NU-1000(Zr)
g-CN nanomaterial provides an ECL signal for the immunosensor, and its morphology has an essential influence on the performance of the ECL sensor. The morphology of synthesized g-CN was characterized by SEM. As seen from Figure 1A,B that g-CN nanomaterial has a lamellar structure with a large number of pores between the lamellar layers, which is similar to those reported literatures [34,35]. This structure effectively increases the contact area between g-CN and the S2O8 2− solution, resulting in an obvious ECL signal. In addition, EDS and mapping analysis show that only C and N elements were present in g-CN and evenly distributed. As for NU-1000(Zr), its morphology was characterized by SEM ( Figure 1A  As can be seen from Figure 2, NU-1000(Zr) is a uniform square rod structure with a length of about 1 µm and a thickness of 150 nm. It can be seen that C, N, and Zr elements are uniformly distributed in NU-1000(Zr) in EDS and mapping.

Characterizations of g-CN and NU-1000(Zr)
g-CN nanomaterial provides an ECL signal for the immunosensor, and its morphology has an essential influence on the performance of the ECL sensor. The morphology of synthesized g-CN was characterized by SEM. As seen from Figure 1A,B that g-CN nanomaterial has a lamellar structure with a large number of pores between the lamellar layers, which is similar to those reported literatures [34,35]. This structure effectively increases the contact area between g-CN and the S 2 O 8 2− solution, resulting in an obvious ECL signal. In addition, EDS and mapping analysis show that only C and N elements were present in g-CN and evenly distributed. As for NU-1000(Zr), its morphology was characterized by SEM ( Figure 1A,B) and TEM ( Figure 1C-E).

Characterizations of g-CN and NU-1000(Zr)
g-CN nanomaterial provides an ECL signal for the immunosensor, and its morphology has an essential influence on the performance of the ECL sensor. The morphology of synthesized g-CN was characterized by SEM. As seen from Figure 1A,B that g-CN nanomaterial has a lamellar structure with a large number of pores between the lamellar layers, which is similar to those reported literatures [34,35]. This structure effectively increases the contact area between g-CN and the S2O8 2− solution, resulting in an obvious ECL signal. In addition, EDS and mapping analysis show that only C and N elements were present in g-CN and evenly distributed. As for NU-1000(Zr), its morphology was characterized by SEM ( Figure 1A  As can be seen from Figure 2, NU-1000(Zr) is a uniform square rod structure with a length of about 1 µm and a thickness of 150 nm. It can be seen that C, N, and Zr elements are uniformly distributed in NU-1000(Zr) in EDS and mapping. As can be seen from Figure 2, NU-1000(Zr) is a uniform square rod structure with a length of about 1 µm and a thickness of 150 nm. It can be seen that C, N, and Zr elements are uniformly distributed in NU-1000(Zr) in EDS and mapping.
The elongated rod structure provides a large number of binding sites for OTA-Apt, thereby immobilizing more OTA molecules. When a large amount of NU-1000(Zr) are fixed on the surface of the ECL immunosensor, the ECL-RET efficiency between g-CN and NU-1000(Zr) is enhanced, leading to a significant change in the ECL signal.  The elongated rod structure provides a large number of binding sites for OTA-Apt, thereby immobilizing more OTA molecules. When a large amount of NU-1000(Zr) are fixed on the surface of the ECL immunosensor, the ECL-RET efficiency between g-CN and NU-1000(Zr) is enhanced, leading to a significant change in the ECL signal.

Expression, Purification, and Identification of Nb28-C4bpα
Nb28-C4bpα fusion protein was developed to improve the sensitivity of Nb28-based immunoassay. Purification and polymerization of Nb28-C4bpα fusion protein were analyzed by SDS-PAGE. By running SDS-PAGE under non-reduction conditions, two clear target bands with definite monomers and heptamers were observed.
SDS-PAGE was used to characterize the purified Nb28 monomer and Nb28-C4bpα heptamer. Figure 3 shows that both Nb28 monomer and Nb28-C4bpα heptamer have only one band, suggesting high antibody purity. SDS-PAGE also detected an obvious target protein band of Nb28 monomer with a molecular weight of 30 kDa approximately. The band length of the Nb28-C4bpα heptamer target protein exceeds 210 kDa. The Nb28-C4bpα fusion protein was proved to be assembled into heptamer by intermolecular disulfide bonds.

Expression, Purification, and Identification of Nb28-C4bpα
Nb28-C4bpα fusion protein was developed to improve the sensitivity of Nb28-based immunoassay. Purification and polymerization of Nb28-C4bpα fusion protein were analyzed by SDS-PAGE. By running SDS-PAGE under non-reduction conditions, two clear target bands with definite monomers and heptamers were observed.
SDS-PAGE was used to characterize the purified Nb28 monomer and Nb28-C4bpα heptamer. Figure 3 shows that both Nb28 monomer and Nb28-C4bpα heptamer have only one band, suggesting high antibody purity. SDS-PAGE also detected an obvious target protein band of Nb28 monomer with a molecular weight of 30 kDa approximately. The band length of the Nb28-C4bpα heptamer target protein exceeds 210 kDa. The Nb28-C4bpα fusion protein was proved to be assembled into heptamer by intermolecular disulfide bonds.

Feasibility Analysis of RET between g-CN and NU-1000(Zr)
The fluorescence emission spectrum (FL emission spectrum) of g-CN and UV-visible absorption spectrum (UV-vis) of NU-1000(Zr) were used to verify the existence of effective ECL-RET between the two nanomaterials. Figure 4 illustrates a strong FL emission

Feasibility Analysis of RET between g-CN and NU-1000(Zr)
The fluorescence emission spectrum (FL emission spectrum) of g-CN and UV-visible absorption spectrum (UV-vis) of NU-1000(Zr) were used to verify the existence of effective ECL-RET between the two nanomaterials. Figure 4 illustrates a strong FL emission peak of g-CN at 455 nm, the FL spectrum of g-CN overlaps with the UV-vis of NU-1000(Zr) at 400 nm-500 nm. According to the literature, the ECL emission spectrum of g-CN has a strong ECL emission peak at 400-600 nm, which overlaps almost completely with the FL spectrum, which is feasible in principle [24,36,37].

Feasibility Analysis of RET between g-CN and NU-1000(Zr)
The fluorescence emission spectrum (FL emission spectrum) of g-CN and UV-vis absorption spectrum (UV-vis) of NU-1000(Zr) were used to verify the existence of ef tive ECL-RET between the two nanomaterials. Figure 4 illustrates a strong FL emis peak of g-CN at 455 nm, the FL spectrum of g-CN overlaps with the UV-vis of N 1000(Zr) at 400 nm-500 nm. According to the literature, the ECL emission spectrum o CN has a strong ECL emission peak at 400-600 nm, which overlaps almost comple with the FL spectrum, which is feasible in principle [24,36,37].

Electrochemical and ECL Behaviors of the ECL-RET Immunosensor
The construction process of the modified electrode in 5 mM K3Fe(CN)6 containing M KCl was characterized by CV curves. As exhibited in Figure 5a, bare GCE showe quasi-reversible redox peak in the presence of a redox probe (curve a). After modifica of g-CN on the GCE surface (curve b), there was little effect on electron transfer, and change of redox current can be ignored. When Nb28-C4bpα heptamer was immobili on the modified electrode (curve c), the peak current was markedly decreased, which m be attributed to the insulation barrier generated by Nb28-C4bpα heptamer and its h

Electrochemical and ECL Behaviors of the ECL-RET Immunosensor
The construction process of the modified electrode in 5 mM K 3 Fe(CN) 6 containing 0.1 M KCl was characterized by CV curves. As exhibited in Figure 5a, bare GCE showed a quasi-reversible redox peak in the presence of a redox probe (curve a). After modification of g-CN on the GCE surface (curve b), there was little effect on electron transfer, and the change of redox current can be ignored. When Nb28-C4bpα heptamer was immobilized on the modified electrode (curve c), the peak current was markedly decreased, which may be attributed to the insulation barrier generated by Nb28-C4bpα heptamer and its high resistance to electron transfer at the electrode/electrolyte interface. After using BSA to block the non-specific active site of the above electrode (curve d), the peak current is reduced due to the blocking effect of this protein on interfacial electron transfer. When OTA was incubated on the modified electrode (curve e), the peak current of CV continued to decrease due to the blocking effect on the redox probe (curve e). After OTA-Apt-NU-1000(Zr) was assembled to the modified electrode, the peak current decreased significantly (curve f) due to the fact that the nanomaterials with lower conductivity have a great blocking effect on the redox probe. The result of the CV signal demonstrates that the modified electrode is successfully constructed. duced due to the blocking effect of this protein on interfacial electron transfer. When OTA was incubated on the modified electrode (curve e), the peak current of CV continued to decrease due to the blocking effect on the redox probe (curve e). After OTA-Apt-NU-1000(Zr) was assembled to the modified electrode, the peak current decreased significantly (curve f) due to the fact that the nanomaterials with lower conductivity have a great blocking effect on the redox probe. The result of the CV signal demonstrates that the modified electrode is successfully constructed. To further verify the quenching effect of NU-1000(Zr) on the signal of g-CN by ECL-RET, the ECL response of different types of modified electrodes was measured in the detection solution. The reaction mechanism of ECL-RET is based on a g-CN-K2S2O8 system as follows. In this reaction system, firstly, g-CN − , SO4 2− , and SO4 − are generated from g-CN and S2O8 2− , respectively, and g-CN − reacts with SO4 − to form the excited state g-CN * . The excited state g-CN * (g-C3N4*) is unstable, and a strong cathode ECL signal will be emitted when g-CN * returns to the ground state of g-CN. As shown in Figure 5b, after g-CN assembly, the modified electrode has strong ECL signal in 0.1 M PBS containing 20 mM K2S2O8 and 0.1 M KCl. In the process of electrode modification, Nb28-C4bpα heptamer and BSA macromolecular proteins have a certain shielding effect on the signal of g-CN on the electrode surface, resulting in the reducing of ECL signal, while OTA with a small molecular weight has a slightly different effect on ECL signal. When the electrodes were directly modified with OTA-Apt-NU-1000(Zr), the ECL-RET interaction between g-CN and NU-1000(Zr) resulted in a large decrease in the ECL signal. If a certain concentration of OTA is incubated with modified electrode prior to incubation of OTA-APT-NU-1000(Zr), OTA will first occupy some the specific binding site of Nb28-C4bpα heptamer, and OTA-Apt-NU-1000(Zr) will bind to the excess site of Nb28-C4bpα heptamer on the electrode surface. As the amount of OTA-Apt-NU-1000(Zr) bound on the electrode surface is reduced, the ECL-RET between g-CN and NU-1000(Zr) is weakened, thereby reducing the ECL signal degradation. To further verify the quenching effect of NU-1000(Zr) on the signal of g-CN by ECL-RET, the ECL response of different types of modified electrodes was measured in the detection solution. The reaction mechanism of ECL-RET is based on a g-CN-K 2 S 2 O 8 system as follows. In this reaction system, firstly, g-CN − , SO 4 2− , and SO 4 − are generated from g-CN and S 2 O 8 2− , respectively, and g-CN − reacts with SO 4 − to form the excited state g-CN * . The excited state g-CN * (g-C 3 N 4 *) is unstable, and a strong cathode ECL signal will be emitted when g-CN * returns to the ground state of g-CN. As shown in Figure 5b, after g-CN assembly, the modified electrode has strong ECL signal in 0.1 M PBS containing 20 mM K 2 S 2 O 8 and 0.1 M KCl. In the process of electrode modification, Nb28-C4bpα heptamer and BSA macromolecular proteins have a certain shielding effect on the signal of g-CN on the electrode surface, resulting in the reducing of ECL signal, while OTA with a small molecular weight has a slightly different effect on ECL signal. When the electrodes were directly modified with OTA-Apt-NU-1000(Zr), the ECL-RET interaction between g-CN and NU-1000(Zr) resulted in a large decrease in the ECL signal. If a certain concentration of OTA is incubated with modified electrode prior to incubation of OTA-APT-NU-1000(Zr), OTA will first occupy some the specific binding site of Nb28-C4bpα heptamer, and OTA-Apt-NU-1000(Zr) will bind to the excess site of Nb28-C4bpα heptamer on the electrode surface. As the amount of OTA-Apt-NU-1000(Zr) bound on the electrode surface is reduced, the ECL-RET between g-CN and NU-1000(Zr) is weakened, thereby reducing the ECL signal degradation.
Because the protein on the electrode reduces the efficiency of electron transport and inhibits the ECL reaction to form the excited state g-CN * , when NU-1000(Zr) were assembled to the modified electrode, the ECL signal significantly decreased, indicating that the energy of the g-CN donor is transferred to the receptor NU-1000(Zr) due to the RET on the electrode.

Optimization of Experimental Conditions
In immuno-recognition experiments, concentrations of Nb28-C4bpα heptamer, incubation time of Nb28-C4bpα heptamer and OTA-Apt-NU-1000(Zr) directly affect the ECL reaction in the assembly process, which ultimately influence the performance of the fabricated ECL-RET immunosensor.
For highly sensitive immune recognition of OTA, the concentration of Nb28-C4bpα heptamer is a key factor affecting specificity recognition efficiency. After incubation with different concentrations of Nb28-C4bpα, the modified electrode was incubated with low concentrations of OTA and OTA-Apt-NU-1000(Zr) complexes. During the experiment, the concentrations and incubation time of OTA and OTA-Apt-NU-1000(Zr) complexes were kept consistent. When the concentration of Nb28-C4bpα was higher, the amount of OTA-Apt-NU-1000 (Zr) complex bound on the electrode surface was higher, causing stronger ECL-RET effect, leading to a decrease in ECL signal. The ECL signal reached equilibrium at 10.0 µg/mL, suggesting that Nb28-C4bpα heptamer captured on the modified electrode was saturated. Thus, the optimal concentration of Nb28-C4bpα heptamer was 10.0 mg/mL.
In addition, the incubation time of Nb28-C4bpα heptamer also had a significant impact on the performance of the ECL-RET immunosensor. As shown in Figure 6b, with the incubation time of Nb28-C4bpα heptamer increase, the ECL signal displayed a downward trend. A rough equilibrium was reached at 60 min, indicating that sufficient Nb28-C4bpα heptamer was trapped on the electrode surface. Thus, 60 min is chosen as the optimal time for the binding of the Nb28-C4bpα heptamer.
Because the protein on the electrode reduces the efficiency of electron transport and inhibits the ECL reaction to form the excited state g-CN * , when NU-1000(Zr) were assembled to the modified electrode, the ECL signal significantly decreased, indicating that the energy of the g-CN donor is transferred to the receptor NU-1000(Zr) due to the RET on the electrode.

Optimization of Experimental Conditions
In immuno-recognition experiments, concentrations of Nb28-C4bpα heptamer, incubation time of Nb28-C4bpα heptamer and OTA-Apt-NU-1000(Zr) directly affect the ECL reaction in the assembly process, which ultimately influence the performance of the fabricated ECL-RET immunosensor.
For highly sensitive immune recognition of OTA, the concentration of Nb28-C4bpα heptamer is a key factor affecting specificity recognition efficiency. After incubation with different concentrations of Nb28-C4bpα, the modified electrode was incubated with low concentrations of OTA and OTA-Apt-NU-1000(Zr) complexes. During the experiment, the concentrations and incubation time of OTA and OTA-Apt-NU-1000(Zr) complexes were kept consistent. When the concentration of Nb28-C4bpα was higher, the amount of OTA-Apt-NU-1000 (Zr) complex bound on the electrode surface was higher, causing stronger ECL-RET effect, leading to a decrease in ECL signal. The ECL signal reached equilibrium at 10.0 µg/mL, suggesting that Nb28-C4bpα heptamer captured on the modified electrode was saturated. Thus, the optimal concentration of Nb28-C4bpα heptamer was 10.0 mg/mL.
In addition, the incubation time of Nb28-C4bpα heptamer also had a significant impact on the performance of the ECL-RET immunosensor. As shown in Figure 6b, with the incubation time of Nb28-C4bpα heptamer increase, the ECL signal displayed a downward trend. A rough equilibrium was reached at 60 min, indicating that sufficient Nb28-C4bpα heptamer was trapped on the electrode surface. Thus, 60 min is chosen as the optimal time for the binding of the Nb28-C4bpα heptamer.  Subsequently, the binding time of OTA-Apt-NU-1000(Zr) also has a significant impact on the performance of the ECL-RET immunosensor. In Figure 6c, the influence of the binding time of OTA-Apt-NU-1000(Zr) is presented. Results showed that the intensity of ECL decreased significantly with increasing incubation time of OTA-Apt-NU-1000(Zr), and reached a stable level after the continuous increase of OTA-Apt-NU-1000(Zr) incubation duration for 60 min. Thus, 60 min is selected as an appropriate incubation time in the experiment.

OTA Detection Performance of ECL-RET Immunosensor
Under optimized conditions, the quantitative detection of OTA by ECL-RET immunosensor is evaluated. In Figure 7a, when the OTA concentration increases from 0.1 pg/mL to 500 ng/mL, the high concentration of OTA caused a decrease in the binding amount of the complex OTA-Apt-NU-1000(Zr) on the electrode surface, and weakened the efficiency of ECL-RET, thus the ECL intensity was high. The ECL signal has a good linear relationship with the logarithm of OTA concentration. The linear equation is I = 1277.65 LogC OTA (ng/mL) + 6345.32 (R 2 = 0.9904), and the detection limit is 33 fg/mL (S/N = 3). In addition, this ECL-RET immunosensor was comparable to the previously reported ECL sensor, as shown in Table 1. In comparison with literatures, our constructed ECL-RET immunosensor has a wider linear range and a lower limit of detection. The high sensitivity of the proposed ECL-RET immunosensor can be attributed to the high RET efficiency between g-CN and NU-1000(Zr) and nanobody heptamer with high affinity. Nb28-C4bpα heptamer can offer more active binding sites for OTA than nanobody monomer Nb28. Compared with the shielding effect of small molecules OTA, ECL signal changes by ECL-RET between g-CN and NU-1000(Zr) are more obvious and sensitive.
Under optimized conditions, the quantitative detection of OTA by ECL-RET immunosensor is evaluated. In Figure 7a, when the OTA concentration increases from 0.1 pg/mL to 500 ng/mL, the high concentration of OTA caused a decrease in the binding amount of the complex OTA-Apt-NU-1000(Zr) on the electrode surface, and weakened the efficiency of ECL-RET, thus the ECL intensity was high. The ECL signal has a good linear relationship with the logarithm of OTA concentration. The linear equation is I = 1277.65 LogCOTA(ng/mL) + 6345.32 (R 2 = 0.9904), and the detection limit is 33 fg/mL (S/N = 3). In addition, this ECL-RET immunosensor was comparable to the previously reported ECL sensor, as shown in Table 1. In comparison with literatures, our constructed ECL-RET immunosensor has a wider linear range and a lower limit of detection. The high sensitivity of the proposed ECL-RET immunosensor can be attributed to the high RET efficiency between g-CN and NU-1000(Zr) and nanobody heptamer with high affinity. Nb28-C4bpα heptamer can offer more active binding sites for OTA than nanobody monomer Nb28. Compared with the shielding effect of small molecules OTA, ECL signal changes by ECL-RET between g-CN and NU-1000(Zr) are more obvious and sensitive.  OTA-Apt-NU-1000(Zr)/OTA/BSA/Nb28-C4bpα heptamer/g-CN/GCE 0.1 pg/mL-500 ng/mL 33 fg/mL This work 1 Glutaraldehyde; 2 Chitosan; 3 quantum dots ; 4 Streptavidin horseradish peroxidase; 5 biotin-H1; 6 DNA tetrahedron-structured aptamer; 7 bipolar electrode; 8 cyanine dye probe DNA; 9 6-Mercapto-1-hexanol; 10 cadmium sulfide semiconductor; 11 Hairpin probe 1; 12 Methylene blue reference probe.  The specificity of the proposed ECL-RET immunosensor was investigated using the same concentration of other mycotoxins (10 ng/mL). With OTA as the control, OTB, OTC, AFB1, FB1, ZEN, and DON were incubated on the prepared ECL-RET immunosensor under the same conditions, respectively. Then, the changes in ECL response were recorded. As depicted in Figure 8a, we can clearly observe that only OTA can cause a significant reduction in ECL signal compared to the blank value. After other mycotoxins were incubated on the immunosensor and then modified with OTA-Apt-NU-1000(Zr), the change of ECL intensity distinctly decreased. The result showed that the binding rate of other toxins to Nb28-C4bpα was very low, so a large amount of Nb28-C4bpα can bind with OTA-APT-NU-1000(Zr), resulting in almost the same degree of quenching in ECL donors. Thus, the proposed ECL-RET immunosensor has good selectivity. Furthermore, we investigated the stability of constructed ECL-RET immunosensor constructed in 5 consecutive times cycles under optimal conditions at 0.0001, 0.1, and 10 ng/mL of OTA ( Figure 8b). As exhibited in Figure 8b (inset), the modified electrodes containing 500 ng/mL OTA obtained a relatively stable ECL curve by 14 consecutive cycles. The RSD of ECL response was only 6.176%, illustrating that the ECL-RET immunosensor had good stability for OTA detection.
As depicted in Figure 8a, we can clearly observe that only OTA can cause a significant reduction in ECL signal compared to the blank value. After other mycotoxins were incubated on the immunosensor and then modified with OTA-Apt-NU-1000(Zr), the change of ECL intensity distinctly decreased. The result showed that the binding rate of other toxins to Nb28-C4bpα was very low, so a large amount of Nb28-C4bpα can bind with OTA-APT-NU-1000(Zr), resulting in almost the same degree of quenching in ECL donors. Thus, the proposed ECL-RET immunosensor has good selectivity. Furthermore, we investigated the stability of constructed ECL-RET immunosensor constructed in 5 consecutive times cycles under optimal conditions at 0.0001, 0.1, and 10 ng/mL of OTA ( Figure 8b). As exhibited in Figure 8b (inset), the modified electrodes containing 500 ng/mL OTA obtained a relatively stable ECL curve by 14 consecutive cycles. The RSD of ECL response was only 6.176%, illustrating that the ECL-RET immunosensor had good stability for OTA detection.

Spiked Sample Analysis
Recovery experiments were conducted to validate the proposed ECL-RET immunosensor in coffee samples. First, 4 mg/mL coffee powder was suspended in water and sonicated for 30 min, and the standard concentration of OTA dissolved in 0.01 M PBS solution was added to the coffee suspension, respectively. Finally, the three samples (spiked with 0.001, 1, and 100 ng/mL OTA) were centrifuged, and the concentration of OTA was assessed in the supernatants. In Table 2, the sample recovery rates of different OTA spiked concentrations are 97.486, 100.603, and 95.784%, respectively. The RSD does not exceed 4.214%. The results reveal that the ECL-RET immunosensor can detect coffee samples with high accuracy, which provides a new way for the detection of OTA content in coffee.

Spiked Sample Analysis
Recovery experiments were conducted to validate the proposed ECL-RET immunosensor in coffee samples. First, 4 mg/mL coffee powder was suspended in water and sonicated for 30 min, and the standard concentration of OTA dissolved in 0.01 M PBS solution was added to the coffee suspension, respectively. Finally, the three samples (spiked with 0.001, 1, and 100 ng/mL OTA) were centrifuged, and the concentration of OTA was assessed in the supernatants. In Table 2, the sample recovery rates of different OTA spiked concentrations are 97.486%, 100.603%, and 95.784%, respectively. The RSD does not exceed 4.214%. The results reveal that the ECL-RET immunosensor can detect coffee samples with high accuracy, which provides a new way for the detection of OTA content in coffee.

Conclusions
In general, we presented a signal amplification ECL-RET immunosensor for ultrasensitive OTA detection using g-CN, NU-1000(Zr), and Nb28-C4bpα heptamer. In this work, the prepared Nb28-C4bpα heptamer can provide more specific capturing sites for OTA and OTA-Apt-NU-1000(Zr) than Nb28 monomers because of its high affinity. In addition, the RET between g-CN and NU-1000(Zr) can effectively display the change of ECL intensity, making up for the shortcoming that small molecules OTA have little influence on ECL signal. Thus, the synergistic effect between Nb28-C4bpα heptamer and the RET-based g-CN and NU-1000(Zr) can greatly improve the sensitivity of ECL immunosensor. The