Introduction

Carcinoembryonic antigen (CEA) is a tumor marker normally expressed in limited areas of human body. Overexpression of CEA contributes to improving the diagnosis of many diseases, including colorectal cancer [1], lung cancer [2], breast cancer [3], pancreatic cancer [4], and many other benign tumors [5]. Various immunological methods have been exploited for CEA determination, including electrochemical immunoassay [6, 7], photoelectrochemical immunoassay [8], fluoroimmunoassay [9], chemiluminescence immunoassay [10, 11], plasmonic immunosensor [12], and so on.

Electrochemical immunoassay is identified as the predominant analytical determination technology for clinical diagnosis due to its advantages such as cost effectiveness, extremely high sensitivity, fast response, and easy operation [13, 14]. Since an immunosensor is constructed based on high affinity of antigen and antibody, it is extremely important to increase antibody immobilization on the electrochemical platform surface. Hence, searching for a reliable and suitable material for effective antibody immobilization is still a crucial subject to explore.

Zeolitic imidazolate frameworks-8 (ZIF-8) as a representative of metal-organic framework (MOF) has a typical sodalite structure and a large specific surface area. Compared with other MOF materials, ZIF-8 has many advantages such as exceptional hydrothermal and chemical stability, a large pore volume, and a uniform pore structure [15,16,17]. These advantages enable ZIF-8 to provide a promising sensing layer for increasing antibody loading. In particular, an ultra-high surface area is one of the most important advantages of ZIF-8. This advantage makes ZIF-8 have considerable potential in serving as immobilization support nanocarriers in biomedical applications. For example, Dong et al. successfully synthesized heterogeneous composite ZnO@ ZIF-8 nanorods, using ZnO@ZIF-8/IL composite film as an immobilization matrix to construct an immunosensor for the determination of human IgG [18]. Wang et al. used zeolitic imidazolate framework-hydroquinone-bovine serum albumin (ZIF-8-HQ-BSA) as immunosensor probes to detect cytokeratin antigen 21-1 [19].

Ordered mesoporous carbon (OMC), a new type of carbon nanomaterials with ordered pores, has been applied widely in sensors due to its excellent properties such as high specific surface area, high thermal stability, flexible structure, electrical conductivity, and good biocompatibility [20,21,22]. OMC with excellent properties is especially desirable for efficient immobilization of nanomaterials and biomolecules. For example, Yang et al. used Au@CMK-3 nanocomposite as nanocarriers for highly dense immobilization of methylene blue and secondary antibodies for determination of prostate-specific antigen [23]. In addition, OMC is intrinsically a good conductor and exhibits outstanding advantages in electrical conductivity. The uniform and ordered porous structure of OMC provides a more advantageous route for electrolyte penetration and transportation.

In this work, an electrochemical immunosensor was constructed for CEA determination based on Au NPs modified ZIF-8 and ordered mesoporous carbon. Au NPs modified ZIF-8 nanoparticles (Au@ZIF-8) was successfully synthesized by reduction of HAuCl4 to improve antibody loading. Antibodies were immobilized on the surface of Au@ZIF-8 via Au-N bonds. Ordered mesoporous carbon with high specific surface area and good electrical conductivity was dropped on the surface of electrode to provide support matrix for Au@ZIF-8 and enhance the conductivity of an immunosensor. Electrochemical properties of biosensors were enhanced by the synergistic effect among Au NPs, ZIF-8, and OMC. The immunosensor shows a wide linear determination range with a low determination limit; therefore, it has extensive application perspective for CEA determination.

Materials and methods

Materials

CEA, anti-CEA antibodies, and alpha fetoprotein (AFP) were purchased from Abcam Co., England (www.abcam.com/). Zn(NO3)2·6H2O, methanol, immunoglobulin G (IgG), and human chorionic gonadotropin (HCG) were brought from Sigma-Aldrich Chemicals Co., USA (www.sigmaaldrich.com/). 2-Methylimidazole, chloroauric acid, potassium ferricyanide (K3Fe(CN)6), and potassium ferrocyanide (K4Fe(CN)6) were bought from Adamas Reagent Co., Ltd., China (www.adamas-beta.com/). OMC was purchased from Nanjing JCNANO Tech Co., Ltd., China (www.jcno.net/). Potassium chloride (KCl) was purchased from Shanghai Titan Scientific Co., Ltd., China (www.titansci.com/). Bovine serum albumin (BSA) was purchased from Yeasen Biotech Co., Ltd., China (www.yeasen.com/). All reagents used were of analytical grade and used without further purification. Deionized water was used for all experiments. The serum samples were provided by Shanghai Pudong Hospital. CEA and anti-CEA antibodies were dissolved with 0.01 mol L−1 phosphate buffer saline (PBS).

Apparatus

Transmission electron microscope (TEM) images were collected on a JEOL JEM-2100F field emission TEM. Scanning electron microscope (SEM) images were acquired using a Hitachi S-4800 field-emission SEM. The Fourier transform infrared spectroscopy spectra were collected on a PerkinElmer Spectrum 100 Fourier transform infrared spectrometer (FTIR). Raman spectroscopy spectra were measured with a Renishaw inVia Reflex Raman spectrometer. X-Ray diffraction (XRD) patterns were measured with a Rigaku D/max-2600PC X-ray diffractometer. X-Ray photoelectron spectroscopy (XPS) images were recorded with a Thermo Fisher ESCALAB 250Xi X-ray photoelectron spectrometer.

Electrochemical measurements

Electrochemical experiments were performed with a CHI660E electrochemical workstation (Chenhua Instrument Shanghai Co., Ltd., China, www. chinstr.com) with a conventional three-electrode system. The modified glassy carbon electrode (GCE, 3 mm in diameter) acted as the working electrode. Glassy carbon electrode can be reused with high repeatability by a simple grinding treatment after measurement. A saturated calomel electrode (SCE) was used as the reference electrode while a platinum plate served as the auxiliary electrode. Cyclic voltammetry (CV) was performed at a scan rate of 50 mV s−1. Electrochemical impedance spectroscopy (EIS) was measured from 100 kHz to 1 Hz with signal amplitude of 5 mV. Differential pulse voltammetry was performed from 0 to 0.6 V with a pulse amplitude of 50 mV and width of 50 ms. Electrochemical experiments were carried out in 0.1 M phosphate buffer that contained 5 mM Fe(CN)63−/4− and 0.1 M KCl.

Synthesis of ZIF-8

ZIF-8 was synthesized using procedures previously reported in the literature [24]. Forty milliliters of Zn(NO3)2·6H2O methanol solution (0.1 mol L−1) was quickly poured into 40 mL 2-methylimidazole methanol solution (0.1 mol L−1) and stirred for 1 h. Milky white solution was collected, centrifuged, and repeatedly washed with methanol for several times to collect white precipitate. The white precipitate was dried in a vacuum oven under 50 °C to collect ZIF-8 powder.

Synthesis of Au@ZIF-8 nanoparticles

One hundred milligrams of ZIF-8 powder was dispersed in methanol solution with ultrasonication for 30 min. A total of 426 μL chloroauric acid methanol solution (0.1 g mL−1) was added into the uniformly dispersed ZIF-8 methanol solution. A total of 2.5 mL NaBH4 methanol solution (0.01 M) was then added into the above mixture and stirred vigorously. The solution gradually turns purple under stirring. After 1 h, the mixture was centrifuged to collect a light purple precipitate. The resulting precipitate was separated from the solution and washed several times with methanol, and dried in a vacuum oven under 50 °C to collect Au@ZIF-8 nanoparticles.

Fabrication of immunosensor

Before modification, GCE was polished to a mirror with a 0.05 μm Al2O3 power slurry, then cleaned by sonication in deionized water, ethanol, and acetone for 3 min. GCE was then dried at room temperature. A total of 0.25 mg ordered mesoporous carbon was dispersed in 1 mL methanol solution by sonication to obtain OMC methanol solution. And 3 μL OMC methanol solution was dropped on the GCE surface. Then, 3 μL Au@ZIF-8 methanol solution (2 mg mL−1) was cast on the modified GCE surface and allowed to dry at room temperature. Afterwards, 10 μL anti-CEA antibodies was immobilized onto the modified GCE surface for at least 12 h at 4 °C. Unbound antibodies were washed away with 0.01 M PBS. Electrodes were incubated in 1% BSA for 1 h to block the remaining active sites and then washed by PBS. Three microliters of various CEA concentrations was dropped on the modified electrode and incubated at 37 °C for 40 min. After being washed by PBS to remove unbound antigen, the electrodes were introduced into an electrochemical test cell and measured by DPV. The calibration plot was obtained by plotting current responses of different CEA concentrations versus the corresponding CEA concentrations. The fabrication scheme of an immunosensor is shown in Fig. 1.

Fig. 1
figure 1

Schematic illustration of fabrication of the immunosensor for CEA determination

Full size image

Real sample determination

Blood samples of various human subjects were obtained from Shanghai Pudong Hospital. To collect a serum sample, the blood sample was centrifuged at 4000 rpm for 15 min to remove some proteins and the clear supernatant was collected for further use. Serum samples were stored at 4 °C before analysis. To determine CEA concentrations in the serum sample, serum samples (6 μL) were incubated with anti-CEA antibodies at 37 °C for 30 min. Unbound CEA was washed away with 0.01 M PBS. Then, DPV was performed in 0.1 M PBS that contained 5 mM Fe(CN)63−/4− and 0.1 M KCl from 0 to 0.6 V to record current signals of real samples. The current signals were compared with the working curve to calculate CEA concentrations in serum.

Results and discussion

Choice of materials

To obtain excellent immunosensor performance, ZIF-8 was selected for antibody immobilizing. The unique sodalite structure of ZIF-8 can provide a large specific surface area to increase antibody loading. Au NPs were attached on ZIF-8 to ensure antibodies were being anchored on ZIF-8 surface through Au-N bonds. Au NPs can also promote electron transfer and improve the biocompatibility of biosensors. OMC was used to improve electrochemical signals and increase the effective surface area of GCE due to its excellent electrical conductivity and high specific surface area.

Material characterization

The morphologies and compositions of ZIF-8 and Au@ZIF-8 were observed via various techniques. TEM images of ZIF-8 (Fig. 2a) and Au@ZIF-8 (Fig. 2b) display that the pristine ZIF-8 crystals have a uniform hexagonal morphology with a particle size of about 200 nm. Compared with ZIF-8, the decoration of Au NPs on ZIF-8 nanoparticles did not change the shapes of ZIF-8 crystals significantly. Au NPs are uniformly dispersed on ZIF-8 without obvious agglomeration. SEM images of ZIF-8 (Fig. 2c) and Au@ZIF-8 (Fig. 2d) also suggest that the presence of Au did not affect the morphology and size of ZIF-8. FTIR spectrum and Raman spectroscopy of ZIF-8, SEM image of OMC, and elemental mapping images of Au@ZIF-8 are described in detail in the Electronic Supporting Material (ESM) (Fig. S1, Fig. S2 and Fig. S3).

Fig. 2
figure 2

a TEM image of ZIF-8; b TEM image of Au@ZIF-8; c SEM image of ZIF-8; d SEM image of Au@ZIF-8

Full size image

TEM and SEM observations were further validated by XRD. In Fig. 3a, the diffraction spectrum of ZIF-8 is completely consistent with the simulation. No diffraction peaks from impurities are observed in the XRD patterns. The unchanged XRD patterns between ZIF-8 and Au@ZIF-8 nanoparticles reveal that the crystal structure of ZIF-8 nanoparticles was unaffected by Au NP decoration. In accordance with the analysis of the TEM and SEM results mentioned above, the XPS spectra of Au/ZIF-8 further confirm the existence of Au NPs on the surface of ZIF-8. XPS spectra show that Au 4f 7/2 and Au 4f 5/2 appear at 84.1 eV and 87.6 eV, respectively (Fig. 3c). This means that the Au element exists in zero valence [25]. Figure 3 d is an XPS diagram of the Zn element with two binding energies at 1021.8 eV and 1044.8 eV, which correspond to Zn2p 3/2 and Zn2p 1/2, respectively [26, 27]. The binding energy of N 1s is about 399 eV (Fig. 3e), which is assigned to C–N bonds existing in a 5-membered imidazole ring [28].

Fig. 3
figure 3

a XRD patterns of Au@ZIF-8, ZIF-8, and situated ZIF-8; b XPS survey spectra of Au@ZIF-8; c Au 4f spectrum; d Zn 2p spectrum; e N 1s spectrum; f O 1s spectrum

Full size image

Characteristics of different electrodes

The electron transfer kinetics for the Au NPs@ZIF-8/OMC/GCE along with Au NPs@ZIF-8/GCE, bare GCE, ZIF-8/GCE, anti-CEA/Au NPs@ZIF-8/OMC/GCE, and CEA/anti-CEA/Au NPs@ZIF-8/OMC/GCE were investigated using CV in 0.1 M KCl that contained 5 mM Fe(CN)63−/4− at a scan rate of 50 mV s−1. In Fig. 4a, the peak current intensity of Au NPs@ZIF-8/OMC/GCE is larger than that of Au NPs@ZIF-8/GCE. This indicates that OMC enhanced the electrochemical performance of GCE significantly. The peak current obviously decreased after ZIF-8 was dropped on an electrode. This was ascribed to the nonconducting properties of ZIF-8. Compared with ZIF-8/GCE, an increase in the redox peak current for Au NPs@ZIF-8/GCE was observed due to the good conductivity of Au NPs. After anti-CEA was deposited on Au NPs@ZIF-8/OMC modified electrode, a significant decrease in peak current intensity was clearly observed. This suggests that the anti-CEA layer was successfully immobilized on the surface of a modified electrode. Critical reduction of peak current is observed upon binding CEA to anti-CEA antibody. This is attributed to the formation of an immunocomplex layer by CEA and anti-CEA combination. The cyclic voltammogram of Au NPs@ZIF-8/OMC/GCE modified GCE represents quasi-reversible redox peak currents at 0.307 (Epa) and 0.234 V (Epc). The peak-to-peak separation is found to be around 73 mV and the ratio of the anodic to cathodic peak currents almost close to unity 1.

Fig. 4
figure 4

a Cyclic voltammetry curves of Au@ZIF-8/OMC/GCE (curve a), bare GCE (curve b), Au@ZIF-8/GCE (curve c), ZIF-8/GCE (curve d), anti-CEA/Au@ZIF-8/OMC/GCE (curve e), CEA/anti-CEA/Au@ZIF-8/OMC/GCE (curve f); b EIS of Au@ZIF-8/OMC/GCE (curve a), bare GCE (curve b), Au@ZIF-8/GCE (curve c), ZIF-8/GCE (curve d), anti-CEA/Au@ZIF-8/OMC/GCE (curve e), CEA/anti-CEA/Au@ZIF-8/OMC/GCE (curve f); c DPV current change (ΔIDPV) determine 100 ng mL−1 CEA using electrodes modified by ZIF-8 (1), Au@ZIF-8 (2), OMC (3), ZIF-8/OMC (4), Au@ZIF-8/OMC (5) in 0.1 M KCl that contained 5 mM [Fe(CN)6]3−/4− (n = 3)

Full size image

ElS spectra of electrodes before and after each step of modifications are shown in Fig. 4b. Au@ZIF-8 modified GCE presented smaller electron transfer resistance than ZIF-8 modified GCE. Due to the slow electron transfer kinetics of proteins, immobilization of antibody on modified electrode and combination between CEA and anti-CEA antibody lead to the increase of electron transfer resistance. This is in agreement with CV results. To further indicate the enhanced electrochemical sensing performance of Au NPs@ZIF-8/OMC, different types of modified electrodes were measured in the existence of 100 ng mL−1 CEA using DPV (Fig. 4c). The results show that the change of DPV peak current (ΔIDPV) from Au NPs@ZIF-8 is significantly higher than that of ZIF-8, and the peak current change value of Au NPs@ZIF-8/OMC is higher than that of Au NPs@ZIF-8 modified electrode. This is because the presence of Au NPs provides a large number of binding sites for antibody and the appearance of OMC promotes electronic transfer, thus improving electrochemical responses.

Optimization of experimental conditions

The following parameters were optimized: (a) concentration ratio of Au@ZIF-8 to OMC; (b) concentration of anti-CEA; (c) reaction time; (d) incubation temperature; (e) sample pH value. Respective text and figures on optimizations are given in the Electronic Supporting Material. In short, the following experimental conditions were found to give the best results: (a) optimal concentration ratio 8:1; (b) optimal loading 20 μg mL −1; (c) reaction time 30 min; (d) incubation temperature 35 °C; (e) best sample pH value 7.4.

Analytical performance of the immunosensor

The response studies of this immunosensor were investigated using DPV. DPV results of different CEA concentrations were measured under optimized conditions. Figure 5a exhibits the sensing performance of anti-CEA/Au NPs@ZIF-8/OMC/GCE for a series of different CEA concentrations ranging from 5 pg mL−1 to 400 ng mL−1. The peak current decreases with increasing CEA concentrations, which is ascribed to the insulating layer formed by antigen antibody binding. It is found that ΔIDPV is linearly related to the logarithm of CEA concentration in the analyte concentration range of 5 pg mL−1 to 400 ng mL−1 (Fig. 5b). The linear relationship between ΔIDPV and logarithmic value of CEA concentration demonstrates that the current response was the direct result of antigen-antibody binding. The linear regression equation is ΔI = 0.6032 log(CCEA) + 1.5053 with correlation coefficient of 0.9986 and low limit of determination (LOD) of 1.3 pg mL−1 (S/N = 3). The sensitivity is calculated as 37.96 μA·log(ng mL−1)·cm−2. Generally, the CEA level in healthy people is less than 5 ng mL−1. The average CEA concentration of 5.0 ng mL−1 in human serum is considered a cutoff value to distinguish abnormal from normal levels. Therefore, this immunosensor can completely meet the needs of clinical diagnosis toward CEA.

Fig. 5
figure 5

a DPV response curves of the immunosensor toward different concentrations of CEA (0.005, 0.05, 0.5, 5, 50, 100, 200, 400 ng mL−1) under optimal conditions. b The responding calibration plot (n = 3)

Full size image

In order to further demonstrate the superior performance of this immunosensor, the analytical performance of this method is compared with that of previously reported methods (Table 1). According to this comparison, the determination limit of this method is lower than that of the method reported previously. The linear range of this method is much wider than most of the immunosensors. The excellent performance of this sensor can be attributed to the combination of Au@ZIF-8 and OMC. Firstly, the large surface area of ZIF-8 can effectively increase antibody immobilization. Secondly, OMC can enhance the conductivity of biosensor by high electron transfer ability. Although a limitation of this technology is that it is a single-analyte assay, this biosensor exhibits unique advantages including rapid response, straightforward fabrication, and use of lesser chemicals and reagents. These advantages make it highly suitable for personalized POCT (point-of-care testing) and decentralization of healthcare management. Compared with other methods, this proposed sensor is expected to build small instruments to realize POCT in future.

Table 1 Comparison of different methods for CEA determination
Full size table

Specificity, stability, and reproducibility and regeneration of this immunosensor

The selectivity of this immunosensor was studied through comparing the electrochemical responses of 10 ng mL−1 CEA with and without interference. Some proteins in human plasma including 1 μg mL−1 HCG, 100 ng mL−1 AFP, 100 ng mL−1 IgG, and 100 ng mL−1 BSA were applied as typical interfering species [36,37,38]. As shown in Fig. 6a, DPV current responses with and without interferents show insignificant variations. This clearly demonstrates that no interaction occurred between the antibodies and interfering species. Thus, this immunosensor has high specificity. To investigate the stability of the immunosensor, electrodes were modified in the same procedure and stored in the PBS at 4 °C. Current responses were recorded by DPV measurements. As shown in Table S1, the current responses of the immunosensor remained 96% after 1 week, 94.8% of the initial response after 10 days. This suggests that the activity of anti-CEA antibody was effectively maintained and the stability of such sensor was also satisfactory. Reproducibility of the immunosensor was further analyzed by preparing a set of modified electrodes independently fabricated within the same batch for determining 5 ng mL−1 CEA. In Fig. 6b, the relative standard deviation (RSD) of 3.2% from all of the six electrodes indicates acceptable reproducibility of this immunosensor.

Fig. 6
figure 6

a Specificity of this immunoassay for 10 ng mL−1 CEA + 100 ng mL−1 AFP (1), 10 ng mL−1 CEA + 1 μg mL−1 HCG (2), 10 ng mL−1 CEA + 100 ng mL−1 BSA (3), 10 ng mL−1 CEA + 100 ng mL−1 IgG (4), 10 ng mL−1 CEA (5), interference without CEA (6) (n = 3). b Current response of six different electrodes treated in the same way for reproducibility study

Full size image

Real sample analysis

To evaluate the feasibility of this immunosensor in real biological samples, a recovery test with different CEA concentrations in diluted serum samples was carried out. The conditions of the antigen-antibody interaction are consistent with the previous experiments. Several human serum samples diluted 100-fold by PBS to minimize nonspecific adsorption were analyzed in the presence of 10, 50, and 100 ng mL−1 CEA. In Table 2, the recovery is found to be in the range of 101–103% and the corresponding RSD is in the range of 4.15–8.29%, respectively. In addition, CEA concentrations of 4 clinical serum samples were detected using this method and the results are compared with reference data from a Siemens Advia Centaur XP chemiluminescence immunoassay analyzer. In Table S2, the relative error between two methods was in the range of 2.82–5.31%. These acceptable results indicated that this immunosensor had significant potential for determination of CEA in clinical diagnostics.

Table 2 Determination of CEA in real samples (n = 3)
Full size table

Conclusion

In this work, an electrochemical immunoassay for CEA determination was constructed by using Au NPs modified ZIF-8 and OMC. Au NPs@ZIF-8 nanoparticles with a large surface area were used to increase antibody loading. OMC can enhance the conductivity of biosensors due to high electron transfer ability. The synergistic effect between Au NPs@ZIF-8 and OMC can amplify an electrochemical response effectively. This biosensor exhibited wide linear range, low determination limit, high specificity, satisfactory stability, and acceptable reproducibility. In addition, the results collected from this method were in acceptable agreement with data determined by the commercial chemiluminescence immunoassay analyzer. Future work should focus on the simultaneous determination of multiple analytes in serum.