Probe-Integrated Label-Free Electrochemical Immunosensor Based on Binary Nanocarbon Composites for Detection of CA19-9

Convenient and sensitive detection of tumor biomarkers is crucial for the early diagnosis and treatment of cancer. Herein, we present a probe-integrated and label-free electrochemical immunosensor based on binary nanocarbon composites and surface-immobilized methylene blue (MB) redox probes for detection of carbohydrate antigen 199 (CA19-9), which is closely associated with gastric malignancies. Nanocarbon composites consisting of electrochemically reduced graphene oxides and carbon nanotubes (ErGO-CNT) are electrodeposited onto an indium tin oxide (ITO) electrode surface to form a 3D nanocomposite film, which could provide high surface area to immobilize abundant MB probes, facilitate the electron transfer of MB, and therefore, improve sensitivity. Polydopamine (PDA) served as a bifunctional linker is able to immobilize anti-CA19-9 antibodies and stabilize the inner probe, conferring the sensing interface with specific recognition capacity. Electrochemical detection of CA19-9 is achieved based on the decrease of the redox signal of MB after specific binding of CA19-9 with a wide linear range of 0.1 mU/mL to 100 U/mL and a limit of detection (LOD) of 0.54 nU/mL (S/N = 3). The constructed electrochemical immunosensor has good selectivity, repeatability, reproducibility, and stability. Furthermore, determination of CA19-9 in human serum samples is also realized.


Introduction
Ultrasensitive and specific detection of tumor markers is crucial for the early diagnosis and treatment of cancer [1,2]. Generally, abnormal presence and changes of concentrations of tumor markers means that people have suffered from a deadly cancer [3]. Carbohydrate antigen 19-9 (CA19-9) is a kind of carbohydrate antigen consisting of macromolecular glycoproteins and has been shown to be closely associated with pancreatic and biliary tract cancers [4]. Usually, the concentration of CA19-9 is significant less than 37 U/mL in normal healthy human blood [5]. A slight increase of CA19-9 implies the possibility of pancreatic cancer development [6]. Up to 63.2 U/mL and 92 U/mL indicate the possibility of cholangiocarcinoma and malignant pancreatic neoplasms, respectively [7,8]. Considering that CA19-9 has been identified as the most reliable biomarker for pancreatic and biliary tract cancers, highly sensitive and convenient detection of CA19-9 is urgent [9].
In recent years, various methods have been developed to detect CA19-9 including enzyme-linked immunosorbent assays (ELISA) [10], surface-enhanced Raman scattering spectroscopy (SERS) [11], radio-immunoassay (RIA) [12], photoelectrochemistry (PEC), [13] and chemiluminescence (CL) [14]. These methods usually require complex and tedious operations, expensive instruments and professional technicians. Electrochemical immunosensor has become one of the most widely analytical methods due to the advantage of low Figure 1 illustrates the fabrication of an immunosensing interface and the reagentless determination of CA19-9 based on the electrochemical signals of the immobilized mediator. GO and CNT were pre-mixed to prepare the GO-CNT dispersion and underwent an electrochemical deposition process, leading to the reduction of GO and finally forming ErGO-CNT onto the ITO surface (ErGO-CNT/ITO). The hybrid of GO and CNT has two advantages. On the one hand, GO could enhance the solubility and dispersibility of CNT in an aqueous solution because of the π-π stacking interaction between the sidewalls of CNT and the aromatic region of GO. On the other hand, incorporated CNT serving as good electronic conductive wires is capable of promoting the electrochemical reduction of GO, decreasing the co-deposition potential. Such obtained ErGO-CNT nanocomposite could improve electroactive area and provide a binding site for cationic redox probe (e.g., MB). Due to the negative charge of carboxylated CNT and the π-conjugated structure of ErGO-CNT, MB with a positive charge can be adsorbed at the ErGO-CNT (MB/ErGO-CNT/ITO) through electrostatic and π-π interactions, giving rise to the electrochemical signals. Then, polydopamine (PDA) was formed on the MB/ErGO-CNT/ITO electrode surface through oxidative self-polymerization of dopamine (DA) in the alkaline solution, which could not only stabilize the inner layer of MB, but also act as the coupling agent for the immobilization of recognitive anti-CA19-9 antibody through Michael addition or Schiffbased reactions. After blocking the non-specific sites with bovine serum albumin (BSA), the immunosensor, denoted as Anti-CA19-9/PDA/MB/ErGO-CNT/ITO, was obtained. When CA19-9 specifically interacted with anti-CA19-9 on the surface of the electrode, the electrochemical signal was reduced due to the poor electron transport efficiency of proteins, realizing the reagentless and label-free electrochemical determination of CA19-9.
could improve electroactive area and provide a binding site for cationic redox probe (e.g., MB). Due to the negative charge of carboxylated CNT and the π-conjugated structure of ErGO-CNT, MB with a positive charge can be adsorbed at the ErGO-CNT (MB/ErGO-CNT/ITO) through electrostatic and π-π interactions, giving rise to the electrochemical signals. Then, polydopamine (PDA) was formed on the MB/ErGO-CNT/ITO electrode surface through oxidative self-polymerization of dopamine (DA) in the alkaline solution, which could not only stabilize the inner layer of MB, but also act as the coupling agent for the immobilization of recognitive anti-CA19-9 antibody through Michael addition or Schiff-based reactions. After blocking the non-specific sites with bovine serum albumin (BSA), the immunosensor, denoted as Anti-CA19-9/PDA/MB/ErGO-CNT/ITO, was obtained. When CA19-9 specifically interacted with anti-CA19-9 on the surface of the electrode, the electrochemical signal was reduced due to the poor electron transport efficiency of proteins, realizing the reagentless and label-free electrochemical determination of CA19-9.  Figure 2a shows the UV-Vis spectra of the supernatant of GO, CNT, and GO-CNT nanocomposite. As seen, CNT without dispersants has no obvious peaks, which arises from the precipitation of CNT after centrifugation. GO shows an obvious absorption peak at 228 nm and a shoulder peak at 300 nm, corresponding to the π→π* transitions of aromatic C-C bond and n→π* transition of C=O bond, respectively. For the GO-CNT nanocomposite, the absorption peak at 300 nm does not shift but that at 228 nm shifts bathochromically to 245 nm, which is due to the π-stacking interactions between the multiple aromatic regions of GO and the sidewalls of CNT [45]. FTIR was also employed to confirm the complex of GO and CNT. As shown in Figure 2b, a broad absorption band at 3445 cm −1 and two absorption bands at 1723 cm −1 and 1623 cm −1 are observed at the spectrum of GO, which is attributed to the hydroxyl, carbonyl, and carboxyl groups, respectively. By contrast, GO-CNT nanocomposite has similar absorption bands but not observed at the spectrum of CNT, suggesting the successful formation of GO-CNT nanocomposite. It is noteworthy that the presence of oxygen-containing groups in GO-CNT nanocomposite could provide abundant binding site for further immobilization of functional species.  Figure 2a shows the UV-Vis spectra of the supernatant of GO, CNT, and GO-CNT nanocomposite. As seen, CNT without dispersants has no obvious peaks, which arises from the precipitation of CNT after centrifugation. GO shows an obvious absorption peak at 228 nm and a shoulder peak at 300 nm, corresponding to the π→π* transitions of aromatic C-C bond and n→π* transition of C=O bond, respectively. For the GO-CNT nanocomposite, the absorption peak at 300 nm does not shift but that at 228 nm shifts bathochromically to 245 nm, which is due to the π-stacking interactions between the multiple aromatic regions of GO and the sidewalls of CNT [45]. FTIR was also employed to confirm the complex of GO and CNT. As shown in Figure 2b, a broad absorption band at 3445 cm −1 and two absorption bands at 1723 cm −1 and 1623 cm −1 are observed at the spectrum of GO, which is attributed to the hydroxyl, carbonyl, and carboxyl groups, respectively. By contrast, GO-CNT nanocomposite has similar absorption bands but not observed at the spectrum of CNT, suggesting the successful formation of GO-CNT nanocomposite. It is noteworthy that the presence of oxygen-containing groups in GO-CNT nanocomposite could provide abundant binding site for further immobilization of functional species.  Figure 3a shows the repetitive cyclic voltammetry (CV) curves of the bare ITO electrode in a potential range from 0 to −1.0 V. In this process, electrochemical reduction of  Figure 3a shows the repetitive cyclic voltammetry (CV) curves of the bare ITO electrode in a potential range from 0 to −1.0 V. In this process, electrochemical reduction of GO occurs and cathodic peak current appears at −1.0 V, eventually resulting in the electrodeposition of ErGO-CNT onto the ITO surface. It is noteworthy that above cathodic peak current gradually decreased with the continuous scanning, further suggesting the reduction of GO and electrodeposition of ErGO-CNT. In order to confirm the electrochemical reduction of GO, the XPS technique was used. Figure 3b,c shows the XPS spectra of GO-CNT/ITO and ErGO-CNT/ITO electrodes. As seen, four characteristic peaks located at 284.5, 286.3, 287.5, and 288.4 eV are ascribed to the C-C/C=C, C-O, C=O, and O-C=O, respectively. When ErGO-CNT was deposited on the ITO electrode, the intensities of characteristic peaks related to oxygen-containing groups of carbon decreased, indicating that the electrodeposition of GO-CNTs was accompanied by the reduction of oxygen-containing functional groups. As shown in Figure 4a,b, numerous wrinkles of GO and bundled structures of CNT are observed, indicating the successful modification of ErGO-CNT on the ITO surface.

Characterization of ErGO-CNT/ITO Electrode
To examine the electrode performance of obtained ErGO-CNT/ITO, electrochemically active surface area (ECSA) of bare ITO and ErGO-CNT/ITO electrodes were compared. ECSA of bare ITO can be accurately calculated by using the Randles-Sevcik equation in the presence of reversible probe (K 3 [Fe(CN) 6 ]) [46].
where A, C, n, D, and ν represent the effective surface area, the bulk concentration of K 3 [Fe(CN 6 )], the number of electrons transferred (n = 1), the diffusion coefficient of K 3 [Fe(CN 6 )] (6.67 × 10 −6 cm 2 s −1 ), and the scan rate, respectively. Figure 5 gives the linear relationship (I p -ν 1/2 ) between the redox peak currents of 0.5 mM K 3 [Fe(CN) 6 ] obtained from these two electrodes (bare ITO and ErGO-CNT/ITO) and the square root of scan rate. According to Equation 1, the ECSA of the bare ITO and ErGO-CNT/ITO were calculated to be 0.36 and 0.45 cm 2 . Despite a slight increase in ECSA, ErGO-CNT/ITO exhibits remarkable enrichment ability towards MB ( Figure S1) due to the electrostatic attraction and π-π effect, showing the great potential for the construction of reagentless electrochemical sensors.  Figure 3a shows the repetitive cyclic voltammetry (CV) curves of the bare ITO electrode in a potential range from 0 to −1.0 V. In this process, electrochemical reduction of GO occurs and cathodic peak current appears at −1.0 V, eventually resulting in the electrodeposition of ErGO-CNT onto the ITO surface. It is noteworthy that above cathodic peak current gradually decreased with the continuous scanning, further suggesting the reduction of GO and electrodeposition of ErGO-CNT. In order to confirm the electrochemical reduction of GO, the XPS technique was used. Figure 3b,c shows the XPS spectra of GO-CNT/ITO and ErGO-CNT/ITO electrodes. As seen, four characteristic peaks located at 284.5, 286.3, 287.5, and 288.4 eV are ascribed to the C-C/C=C, C-O, C=O, and O-C=O, respectively. When ErGO-CNT was deposited on the ITO electrode, the intensities of characteristic peaks related to oxygen-containing groups of carbon decreased, indicating that the electrodeposition of GO-CNTs was accompanied by the reduction of oxygen-containing functional groups. As shown in Figure 4a

Characterization of ErGO-CNT/ITO Electrode
where A, C, n, D, and ν represent the effective surface area, the bulk concentration of K3[Fe(CN6)], the number of electrons transferred (n = 1), the diffusion coefficient of K3[Fe(CN6)] (6.67 × 10 −6 cm 2 s −1 ), and the scan rate, respectively. Figure 5 gives the linear relationship (Ip-ν 1/2 ) between the redox peak currents of 0.5 mM K3[Fe(CN)6] obtained from these two electrodes (bare ITO and ErGO-CNT/ITO) and the square root of scan rate. According to Equation 1, the ECSA of the bare ITO and ErGO-CNT/ITO were calculated to be 0.36 and 0.45 cm 2 . Despite a slight increase in ECSA, ErGO-CNT/ITO exhibits remarkable enrichment ability towards MB ( Figure S1) due to the electrostatic attraction and ππ effect, showing the great potential for the construction of reagentless electrochemical sensors.

Stabilization Effect of PDA
MB with a reversible redox and low-cost properties is an organic dye that belongs to the phenothiazine family, which has been widely used as an excellent electrochemical redox probe in the field of electrochemical sensors [47]. Cationic MB was immobilized onto the ErGO-CNT/ITO surface via electrostatic and π-π effects under stirring for 60 min ( Figure S3), which did not change the surface morphology

Stabilization Effect of PDA
MB with a reversible redox and low-cost properties is an organic dye that belongs to the phenothiazine family, which has been widely used as an excellent electrochemical redox probe in the field of electrochemical sensors [47]. Cationic MB was immobilized onto the ErGO-CNT/ITO surface via electrostatic and π-π effects under stirring for 60 min (Figure S3), which did not change the surface morphology of ErGO-CNT/ITO (Figure 4b,c). After self-polymerization of PDA, smooth surface was presented at the PDA/MB/ErGO-CNT/ITO (Figure 4c,d). After self-polymerization of PDA, smooth surface was presented at the PDA/M), indicating the formation of uniform PDA film. Considering the protective effect of PDA in previous reports [48,49], the stability of PDA/MB/ErGO-CNT/ITO was investigated. Figure 6 compares the 20 consecutive CV scans of the prepared MB/ErGO-CNT/ITO electrode and PDA/MB/ErGO-CNT/ITO electrodes. As displayed, the MB/ErGO-CNT/ITO electrode has a pair of well-defined redox peaks, corresponding to the signal of MB and suggesting the successful immobilization of MB. However, the redox peak current signals of MB at the MB/ErGO-CNT/ITO decreased with an increase of scanning cycles while those at the PDA/MB/ErGO-CNT/ITO hardly changed. This is because the presence of PDA is able to prevent the inner MB from falling off the electrode surface and effectively stabilize the immobilization of MB. Moreover, the effect of pH of the supporting electrolyte on the PDA/MB/ErGO-CNT/ITO was studied. As shown in Figure S2, the anodic peak current increases with the decrease of pH value and the anodic current peak shifts positively, which is attributed to the proton-involved electrochemical reaction of MB [50,51]. However, considering the stable activity of the protein under neutral conditions and 91% of signal at the pH of 4.0, pH = 7.4 was chosen as the optimal experimental condition.

Fabrication of Immunosensor
The feasibility of the immunosensor construction was verified by the electrochemical method. Figure 7a is the CV curves of each electrode in 0.1 M KCl solution containing 2.5 mM Fe(CN)6 3-/4-. As shown, Fe(CN)6 3-/4-displays a pair of reversible redox peaks at the PDA/MB/ErGO-CNT/ITO electrode. Then, recognitive antibody (anti-CA19-9) can be immobilized onto the PDA/MB/ErGO-CNT/ITO electrode surface based on the interaction between the dopaquinone structure of PDA and the -SH or -NH2 groups of anti-CA19-9. After blocking the non-specific sites with BSA, CA19-9 was detected by Anti-CA19-9/PDA/MB/ErGO-CNT/ITO electrode. Owning to the non-conductive steric hindrance layer of proteins, both anti-CA19-9 and CA19-9 hinder the electron exchange of electrochemical probes on the electrode surface, leading to the decreased redox peak signals and larger peak-to-peak difference. Figure 7b shows the EIS responses of each electrode to 2.5 mM Fe(CN)6 3-/4-. Each curve consists of a semicircle in the high frequency region and a linear part in the low frequency region, representing an electron transfer-limited process and a diffusion-limited process, respectively. As seen, the charge transfer resistance was

Fabrication of Immunosensor
The feasibility of the immunosensor construction was verified by the electrochemical method. Figure 7a is the CV curves of each electrode in 0.1 M KCl solution containing 2.5 mM Fe(CN) 6 3-/4-. As shown, Fe(CN) 6 3-/4displays a pair of reversible redox peaks at the PDA/MB/ErGO-CNT/ITO electrode. Then, recognitive antibody (anti-CA19-9) can be immobilized onto the PDA/MB/ErGO-CNT/ITO electrode surface based on the interaction between the dopaquinone structure of PDA and the -SH or -NH 2 groups of anti-CA19-9. After blocking the non-specific sites with BSA, CA19-9 was detected by Anti-CA19-9/PDA/MB/ErGO-CNT/ITO electrode. Owning to the non-conductive steric hindrance layer of proteins, both anti-CA19-9 and CA19-9 hinder the electron exchange of electrochemical probes on the electrode surface, leading to the decreased redox peak signals and larger peak-to-peak difference. Figure 7b shows the EIS responses of each electrode to 2.5 mM Fe(CN) 6 3-/4-. Each curve consists of a semicircle in the high frequency region and a linear part in the low frequency region, representing an electron transfer-limited process and a diffusion-limited process, respectively. As seen, the charge transfer resistance was increased after each modification, which was consistent with the current variation of CV responses shown in Figure 7a. At the same time, we used the inner probe strategy to characterize the electrode construction process. It can be seen from Figure 7c,d, the CV and differential pulse voltammetry (DPV) signals of MB gradually decreased with the stepwise modification. All above results demonstrate the successful construction of the immunosensor.
mM Fe(CN)6 3-/4-. As shown, Fe(CN)6 3-/4-displays a pair of reversible redox peaks at the PDA/MB/ErGO-CNT/ITO electrode. Then, recognitive antibody (anti-CA19-9) can be immobilized onto the PDA/MB/ErGO-CNT/ITO electrode surface based on the interaction between the dopaquinone structure of PDA and the -SH or -NH2 groups of anti-CA19-9. After blocking the non-specific sites with BSA, CA19-9 was detected by Anti-CA19-9/PDA/MB/ErGO-CNT/ITO electrode. Owning to the non-conductive steric hindrance layer of proteins, both anti-CA19-9 and CA19-9 hinder the electron exchange of electrochemical probes on the electrode surface, leading to the decreased redox peak signals and larger peak-to-peak difference. Figure 7b shows the EIS responses of each electrode to 2.5 mM Fe(CN)6 3-/4-. Each curve consists of a semicircle in the high frequency region and a linear part in the low frequency region, representing an electron transfer-limited process and a diffusion-limited process, respectively. As seen, the charge transfer resistance was increased after each modification, which was consistent with the current variation of CV responses shown in Figure 7a. At the same time, we used the inner probe strategy to characterize the electrode construction process. It can be seen from Figure 7c,d, the CV and differential pulse voltammetry (DPV) signals of MB gradually decreased with the stepwise modification. All above results demonstrate the successful construction of the immunosensor.

Electrochemical Determination of CA19-9
CA19-9 was detected by DPV based on the decrease of the electrochemical signal of MB after CA19-9 binding, and the results were shown in Figure 8. As shown, after incubating with different concentrations of CA19-9, anodic peak currents decrease with the increasing CA19-9 concentration. This is because the formed antigen-antibody complex hinders the access of electrolyte anions to the electrode surface and charge compensation could not occur. The good linear relationship between anodic peak currents and the logarithm of CA19-9 concentration could be found from 0.1 mU/mL to 100 U/mL (Figure 8b), yielding a regressive equation of IEC = -0.321 lgCCA19-9 + 4.27 (R 2 = 0.990) and the limit of detection (LOD) of 0.54 nU/mL (S/N = 3). Analytical performance of our sensor with some other reported methods is provided in Table 1. As compared, our proposed sensor has a wider linear range and a rather lower LOD.

Electrochemical Determination of CA19-9
CA19-9 was detected by DPV based on the decrease of the electrochemical signal of MB after CA19-9 binding, and the results were shown in Figure 8. As shown, after incubating with different concentrations of CA19-9, anodic peak currents decrease with the increasing CA19-9 concentration. This is because the formed antigen-antibody complex hinders the access of electrolyte anions to the electrode surface and charge compensation could not occur. The good linear relationship between anodic peak currents and the logarithm of CA19-9 concentration could be found from 0.1 mU/mL to 100 U/mL (Figure 8b), yielding a regressive equation of I EC = -0.321 lgC CA19-9 + 4.27 (R 2 = 0.990) and the limit of detection (LOD) of 0.54 nU/mL (S/N = 3). Analytical performance of our sensor with some other reported methods is provided in Table 1. As compared, our proposed sensor has a wider linear range and a rather lower LOD.
MB after CA19-9 binding, and the results were shown in Figure 8. As shown, after incubating with different concentrations of CA19-9, anodic peak currents decrease with the increasing CA19-9 concentration. This is because the formed antigen-antibody complex hinders the access of electrolyte anions to the electrode surface and charge compensation could not occur. The good linear relationship between anodic peak currents and the logarithm of CA19-9 concentration could be found from 0.1 mU/mL to 100 U/mL (Figure 8b), yielding a regressive equation of IEC = -0.321 lgCCA19-9 + 4.27 (R 2 = 0.990) and the limit of detection (LOD) of 0.54 nU/mL (S/N = 3). Analytical performance of our sensor with some other reported methods is provided in Table 1. As compared, our proposed sensor has a wider linear range and a rather lower LOD.

Selectivity, Reproducibility, and Stability of the Constructed Immunosensor
The selectivity of the constructed immunosensor was investigated. As shown in Figure 9a, except for CA19-9, the other four antigens including AFP, CA125, S100, and PSA did not change the peak current responses at the anti-CA19-9/PDA/MB/ErGO-CNT/ITO sensor. Even though all of tumor markers were mixed with CA19-9, the peak current of the electrode was nearly similar with that of a single CA19-9, proving the good selectivity of the developed immunosensor. The inter-electrode reproductivity and storage stability of the constructed immunosensor were also evaluated. The reproducibility of this immunosensor was investigated by preparing five electrodes in the same batch and the obtained RSD value of five different electrodes was 3.6% (Figure 9b). After 5 days of storage in a refrigerator (4°C), the response of the immunosensor towards CA19-9 retained about 94% of the initial signal ( Figure 9c). All these results show that the proposed immunosensor has good selectivity, reproducibility, and stability, making a good candidate for real sample analysis.
of the developed immunosensor. The inter-electrode reproductivity and storage stability of the constructed immunosensor were also evaluated. The reproducibility of this immunosensor was investigated by preparing five electrodes in the same batch and the obtained RSD value of five different electrodes was 3.6% (Figure 9b). After 5 days of storage in a refrigerator (4 ℃), the response of the immunosensor towards CA19-9 retained about 94% of the initial signal (Figure 9c). All these results show that the proposed immunosensor has good selectivity, reproducibility, and stability, making a good candidate for real sample analysis.

Determination of CA19-9 in Human Serum
In order to evaluate the practical application potential of our immunosensor, 50-time diluted human serum with 0.01 M PBS (pH = 7.4) was employed as the analytical samples. Table 2 shows the results of human serum samples with artificially spiking in a known (2) 1 U/mL CA125; (3) 1 ng/mL S100; (4) 1 ng/mL PSA; (5) 100 mU/mL CA19-9. Reproducibility (b,c) stability of immunosensor. The concentration of CA19-9 is 1 mU/mL. Error bars represent the standard deviations of three measurements.

Determination of CA19-9 in Human Serum
In order to evaluate the practical application potential of our immunosensor, 50-time diluted human serum with 0.01 M PBS (pH = 7.4) was employed as the analytical samples. Table 2 shows the results of human serum samples with artificially spiking in a known concentration of CA19-9, exhibiting satisfactory recoveries (between 95.0 and 110%) and small RSDs (between 0.36 and 1.3%, n = 3). Moreover, the human serum sample spiked with 10 U/mL was determined by our sensor and enzyme-linked immuno-sorbent assay (ELISA) and the obtained results of both methods were comparable, showing the reliability and accuracy of our immunosensor.

Experiments and Instrumentations
Scanning electron microscopy (SEM) was conducted on a field-emission scanning electron microscope (S-4800, Hitachi, Tokyo, Japan). Cyclic voltammetry (CV) and differential pulse voltammetry (DPV) measurements were performed on an Autolab PGSTAT302N electrochemical workstation (Metrohm, Herisau, Switzerland). Above electrochemical measurements were performed with a three-electrode system containing a bare or modified ITO as the working electrode, Ag/AgCl (saturated KCl solution) as the reference electrode, and a Pt wire electrode as the counter electrode. DPV measurement parameters include step (0.005 V), modulation amplitude (0.05 V), modulation time (0.05 s), and interval time (0.2 s). A PHI5300 electron spectrometer (PE Ltd., Waltham, MA, USA) was used to conduct X-Ray photoelectron spectroscopy (XPS) analysis (250 W, 14 kV, Mg Kα radiation). Ultraviolet-Vis (UV-Vis) absorption spectra were obtained from a UV-Vis spectrometer (UV-2450; Shimadzu, Kyoto, Japan). Fourier transform infrared spectroscopy (FTIR) data were recorded on a Vertex 70 spectrometer (Bruker, Woodlands, TX, USA) through KBr tablet method.

Preparation of the MB/ErGO-CNT/ITO
GO was prepared by a modified Hummers method [55]. The fabrication process of GO-CNT dispersion is as the following: 1 mg/mL GO and 0.5 mg/mL CNT was mixed into PBS (0.2 M pH = 6.5) solution and sonicated for 1 h. The unstable CNT was removed by centrifugation at 8000 rpm for 10 min. To deposit the ErGO-CNT on the ITO, the clean ITO electrode (geometric area = 0.5 cm 2 ) was soaked into the above GO-CNT dispersion and applied 30 successive CVs scanning from 0 to -1.0 V at the scan rate of 50 mV/s. In this process, electrochemical reduction of GO and electrodeposition of their nanocomposite to the ITO electrode surface simultaneously occur. The obtained electrode was rinsed with water lightly, dried at 60 • C and applied a constant current of -350 mA for 10 s in 0.2 M PBS (pH = 6.5) to further reduce GO. ErGO-CNT supported by ITO electrode was finally prepared, termed as ErGO-CNT/ITO. MB/ErGO-CNT/ITO was obtained by immersing the ErGO-CNT/ITO electrode into a 0.01 M PBS (pH = 7.4) solution containing 1 mM MB under stirring for 60 min.

Fabrication of Immunosensor
In order to immobilize antibodies and encapsulate MB, the MB/ErGO-CNT/ITO electrode was first immersed in DA solution (1 mg/mL in 0.1 M PBS, pH = 8.5) for 1 h and self-polymerization of the DA was induced to obtain the polydopamine (PDA)/MB/ErGO-CNT/ITO electrode. After removing the residual DA, 50 µL of 100 µg/mL anti-CA19-9 in 0.01 M PBS (pH 7.4) was dropped on the above electrode surface and incubated at 4 • C overnight. Loosely bounded anti-CA19-9 antibody was carefully washed off and then 50 µL of 1 wt% BSA in 0.01 M PBS (pH 7.4) was used to block nonspecific binding sites at 37 • C for 1 h. Finally, 50 µL of various concentrations of CA19-9 was dropped onto the immunosensor surface and incubated at 37 • C for 0.5 h. After being rinsed with PBS, the DPV curves were recorded to realize the quantitative detection of CA19-9.

Conclusions
In this work, a simple and sensitive label-free electrochemical immunosensor is fabricated for reagentless determination of CA19-9 through ErGO-CNT nanocomposite and surface-immobilized MB probe. ErGO-CNT could be prepared by a simple and controllable electrodeposited method. The carboxyl group of CNT and the π-conjugated structure of ErGO-CNT provide binding sites for the cationic MB probe, which could be further stabilized by PDA. With the linkage of PDA, anti-CA19-9 antibody covalently binding to the MB/ErGO-CNT/ITO endows the electrode recognized capacity. Combining with the good conductivity of ErGO-CNT and stabilization of PDA, determination of CA19-9 was achieved by recording the reduction of redox signals of MB after incubation with CA19-9. Furthermore, the fabricated immunosensor exhibits convenient construction steps, high sensitivity, good selectivity, and has been successfully applied to the analysis in human serum, providing an easy and efficient strategy for sensitive determination of CA19-9.
Supplementary Materials: The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/molecules27206778/s1, Figure S1: CV curves of bare ITO and ErGO-CNT/ITO electrodes after soaking into MB solution; Figure S2: Effect of pH of supporting electrolyte on the PDA/MB/GO-CNT/ITO electrode; Figure

Data Availability Statement:
The data presented in this study are available on request from the corresponding author.