The ZMP nanocomposites were characterized by X-ray diffraction (XRD) and transmission electron microscopy (TEM). The XRD graph (Figure 1a) showed the characteristic peaks of both Fe₃O₄ (311, 2θ = 37.5°) and ZrO₂ (111, 2θ = 27.5°; 022, 2θ = 52°) crystals, which indicated that the ZrO₂ wrapped in a layer of Fe₃O₄ is in the crystalline state. X-ray fluorescence spectrum (XRF) analysis of the ZMPs showed Zr-K_β(17.8 keV), Zr-K_α(15.8 keV), Zr-L_β (2.1 keV), Zr-L_α (2.0 keV), Fe-K_β (7.1 KeV), and Fe-K_α (6.4 keV) peaks (data not shown), which indicated that elemental Fe and Zr exist in these magnetic particles. TEM (Figure 1b) showed that the ZMP structure was of the core (black)-shell (white) type, with Fe₃O₄ being the nucleus and ZrO₂ being the shell. When the ZMP surface was coated with a layer of antibody, these two components unite (Figure 1c).

As shown in (Figure 1d), the probe showed the appearance of a cross-linked “bead chain” because the DNA molecules that absorbed the ZMP were cross-linked with each other in solution. The ZMPs then wrapped around the DNA, forming cross-linked bead chains. XRF showed that the chain contained Zr-k_α (2.1 keV), Fe-k_α (6.4 keV), P-k_α peak (1.13 keV), and S-k peaks (2.3 keV) (data not shown). Because DNA is rich in phosphoric acid, the presence of P-k_α peak groups on the ZMP surface further confirmed that the ZMPs coated the DNA. We examined the ultraviolet spectrum after the reaction between the ZMPs and DNA and found that the function absorption peaks of DNA redshifted from 520 nm to 524 nm, implying that the DNA and ZMPs had interacted and this led to the function absorption redshift. Figure 2 shows that the DNA/(ZMPs-HRP-CEA Ab₂)_n magnetic probe has good superparamagnetism under the reaction of the 0.3 mTde external magnetic field.

Figure 3 shows the electrochemical characterization of differently modified electrodes in 1 mM Fe(CN)₆ ^3–/4–. GCE modified with CS showed a redox peak with a large peak-to-peak separation (curve a). When nano-Au was modified on the electrode, the anodic and cathodic current peaks increased, and the peak-to-peak separation reduced (curve b); an explanation for this could be that nano-Au accelerates electron transfer. When the Ab₁ adsorbed onto nano-Au (curve c), the current decreased, because the antibody may hinder electron transfer. When the electrode was incubated with BSA (curve d), the peaks declined further.

As shown in Figure 4, the GCE/CS-Au NPs/Ab₁ electrode exhibited no redox peak in the blank (PBS) between −0.3 V and 0.8 V. When 5 mmol/L H₂O₂ was added into the solution, the modified electrode (Figure 5b) showed a pair of redox peak (100 mV/s) between −0.48 V and −0.55 V. When the immunosensor was used to test CEA and the sandwich immune complex was formed (Figure 4c), the reduction current of the OPD oxidization complex increased obviously, and the oxidation current reduced. This is because the HRP enzyme on the surface of the probe catalyzes the reaction between H₂O₂ and OPD and increases the detection signal.

The concentration of HRP immobilized on the ZMPs is critical for the performance of the immunosensor because HRP catalyzes the reduction of OPD by H₂O₂. To obtain the optimum amount of HRP-labeled and tagged antibody on ZMPs, we compared the OD values of HRP-CEA Ab₂ labeled on ZMPs by using the enzyme-linked immunosorbent assay (ELISA) at 450 nm. We separately added 100, 200, 300, 400, and 500 μL of 12 μg/L HRP-CEA Ab₂ into 5 mL of 2 mg/mL ZMP NPs and then added 1 mmol/L H₂O₂ for detection. As seen from (Figure 5a), the OD values of HRP-CEA Ab₂ increased as its concentration increased. When 400 μL HRP-CEA Ab₂ (4.8 ng) was added, the OD values stabilized, indicating that the adsorption of HRP in ZMPs reaches saturation at 400 μL.

The effect of the electrolyte pH in the ECI is examined from two aspects: First, it directly influences the activity of HRP. Second, it affects the peak potential of the electrode reaction. (Figure 5b) shows the magnitude of the catalytic current from modified electrodes with electrolyte of different pH values. When the pH value was 7.0, the catalytic current was the maximum. This indicates that when the pH is 7.0, the HRP activity in the modified electrode membrane is the highest. These findings are consistent with the original nature of HRP [17]. Thus, the progress of the enrichment of ZMPs with HRP does not change the pH value at which the maximum catalytic reactions of HRP occur. On mapping the reduction and oxidation peak potentials against the pH values, we obtained two lines between pH 4 and 8. Their slopes were −59.87 and −51.72 mV/pH unit. The peak potential increased with pH, indicating that protons participate in the electrode reaction of OPD oxidation. The ratio of the participating protons and the electrons transferred in the electrode reaction was 1:1. Since the process of the electrode response for OPD was a double-electron process, the number of protons participating in the electrode reaction is 2 [18], which is the same as the number of electrons gained/lost in the redoxidation of OPD.

The effect of the incubation temperature (Figure 5c) on the catalytic current was also studied. The immunosensor was found to possess a good current response signal at room temperature. Thus, all the experiments were conducted at room temperature. The incubation time of CEA Ab₂ with functionalized DNA/ZMPs-Au was also optimized (Figure 5d), and the results showed that the maximum time required was 30 min. Thus, the optimum experimental conditions for detection were found to be pH 7.0, room temperature, and an incubation time of 30 min.

As shown in Figure 6, the response current increased with the increase in the CEA concentration. When the concentration ranged from 0.008 to 200 ng/mL, the linear correlation coefficient was 0.9943. The detection limit (DL) was 5 pg/mL (3δ). The average level of CEA in healthy humans is 20 ng/mL, so our ECI can easily be used for CEA detection in lab samples without any need for dilution.

Next, we compared our sensor with other sandwich-type sensors. As shown in Table 1, most sandwich immunosensors have a low DL and high detection sensitivity, and their DL is 100 times lower than that of traditional ELISAs.

The immunosensors prepared at different times and in different batches were used to test 15 and 25 ng/mL CEA four times. The coefficient of variation obtained was 2.3% and 2.2%, respectively, indicating the good precision of the immunosensors. The proposed sensors were stored in PBS (pH 6.5; 4 °C) for 45 days, and they did not show any obvious changes (signal changes <5%), which shows that these sensors have good storage stability. The electrode surface can be updated by placing the electrode under the traction of a magnetic field (0.3 mT at the bottom of the electrolytic cell, with the magnetic field lines oriented perpendicular to the direction of the surface of the electrode in solution) and then mixing the solution in order to free the magnetic nanoprobes from the electrode surface. Updated electrodes can be used again for sandwich detection. Such electrodes were used to test 15 and 100 ng/mL CEA samples, and the measured values were 15.7 and 99.3 ng/mL, respectively, with relative standard deviation values (n = 3) of 3.1% and 3.2%, respectively. These findings suggest that the ECIs have good preparation repeatability.

We studied the effects of major interfering agents in blood serum on the immunosensor. When the concentration of CEA was 5 ng/mL, the difference in the sensor signal deviated only 5% from the normal with the following interfering agents: 10 times the normal level of CEA and hepatitis B virus; 200 times the normal level of BSA, glucose, and uric acid; and 800 times the normal levels of Na⁺, Fe²⁺, Fe³⁺, Zn²⁺, and Ca²⁺. This suggests that the sensor effectively resists disturbances caused by the main interfering agents in human serum.

As shown in Table 2, the level of CEA in human serum was determined using our ECI. The results were consistent with those obtained using ELISA, and the recoveries were between 95% and 107%, indicating that our method is suitable for detecting CEA in serum. The DL of the ECI can reach 5 pg/mL, while that of ELISA can only reach a maximum of 0.1 ng/mL. Thus, our method is 100-fold more sensitive than ELISA. Therefore, it is fit for detecting minor changes in the CEA concentration in serum, which is in turn useful for the early diagnosis of cancer.

OPD and H₂O₂ were purchased from Shanghai Crystal Pure Reagent Co. Ltd in China. We made the Fe₃O₄/ZrO₂ magnetic particles by ourselves. CEA monoclonal antibody solution (Ab₁) and CEA ELISA kits were obtained from Biocell (Zhengzhou, China); the latter contain standard CEA solution and HRP-labeled CEA monoclonal antibody (HRP-Ab₂). 1 mg/mL Nano gold colloid (5 nm) Calf thymus DNA and BSA were purchased from Aldrich (Sigma Co. Ltd, USA). All other reagents were of analytical grade. Double-distilled water was used for all experiments.

Cyclic voltammetric measurements were performed on a CHI 660a electrochemistry workstation (Shanghai CH Instruments, China) with a three-electrode system composed of a platinum wire auxiliary electrode, a saturated calomel reference electrode (SCE), and a bare or modified GCE/CS/nano-Au electrode as the working electrode. Scanning electron micrographs were taken with a scanning electron microscope (SEM; S-4800; Hitachi, Tokyo, Japan). Further, we used an S2 RANGER X-ray fluorescence spectrometer (Bruker, Germany), ST-360 microplate reader (Shanghai Science Biotechnology), NdFeB magnet (Hangzhou Magnet Equipment Ltd., China), and an ultra-pure water meter (Millipore, USA).

First, colloidal nano ZrO₂ and nano Fe₃O₄ particles were synthesized as reported previously [29,30]. Next, 1 g nano Fe₃O₄ particles were added into 100 mL of 10 mg/mL colloidal nano ZrO₂, and the mixture was stirred for 6 h at room temperature. The resulting particles were separated by applying an external magnetic field. The maximum size of the ZMPs was 18.6 nm (93.2% of the total number of particles), and the minimum was 45.83 nm (6.8% of the total number of particles). Second, HRP-Ab₂ was added to 5 mL of 2 mg/mL ZMP solution, the mixture was stirred constantly for 12 h, and the unabsorbed Ab₂ was eliminated by magnetic separation. The final mixture was washed repeatedly. Subsequently, HRP was used to tagged unreacted and nonspecific sites on the ZMPs. Magnetic cross-linked nanochains (DNA/(ZMPs-HRP-CEA Ab2)_n) were obtained from a mixture of 10 mg/mL calf thymus DNA and ZMPs-HRP-CEA Ab₂ that was gently stirred in a magnetic field for 6 h and then separated magnetically. The obtained probes were resuspended in PBS of pH 7.0 containing 0.2% BSA. The formation of this probe was monitored by TEM.

First, 10 μL of 0.5% CS ethanol was deposited onto the polished GCE surface. The GCE/CS obtained when the ethanol was volatilized completely was washed with double-distilled water. Second, 50 μL 1 mg/mL nano-Au (5 nm) was electrochemically deposited onto the CS membrane surface; Finally, GCE/CS/nano-Au was obtained, and the electrode was scanned 10 times with the potential increasing from −0.8 to 0.8 V in 0.1 M NaOH solution at 100 mV/s. Then, this electrode was soaked in 0.1 M PBS (pH 7.0) containing 1 mg/mL Ab1 for 12 h, and the GCE/CS/nano-Au/Ab1 (hereafter referred to as GCE/Ab1) immunoelectrode was obtained. The electrode was incubated with 3 mg/mL BSA for 30 min to block the unreacted and nonspecific sites (Scheme 1).

Scheme 1 shows the detection principle of the immunosensor. The procedure is based on the typical procedure used in sandwich-type immunoreactions. When the immunoreaction between GCE/Ab₁ and the CEA antigen was completed, DNA/(ZMPs-HRP-Ab₂)_n probes were captured through the “sandwich” mode to form sandwich-type immune complexes (Ab₁/CEA/(DNA/(ZMPs-HRP-Ab₂)_n). The response current of this immunosensor was found to be positively correlated with CEA in a certain concentration range. Therefore, it could be used for the quantification of CEA, because the HRP on the surface of the immune complex can catalyze the reaction between H₂O₂ and OPD.