1. Introduction

Cancer is one of the leading causes of death worldwide [1]. Especially, breast cancer was the most common cancer in women with 627,000 deaths in 2018, and its rising incidence rates had been observed in many countries during the last few decades [2,3]. Early screening and effective postoperative evaluation of breast cancer can significantly reduce cancer mortality [4]. Although great progress has been made in the treatment and evaluation of breast cancer, mainstream imaging examination and tissue biopsy still bring high costs, radiation damage and psychological burden to patients [5–7]. Alternatively, many research results showed that cancer biomarkers have great potential in early diagnosis and monitoring of cancers [8]. Some cancer biomarkers have been used in routine cancer screening [9]. For example, the concentration of carcinoembryonic antigen (CEA) in the serum has been used as an important auxiliary indicator for the diagnosis of breast cancer, as well as for the monitoring of recurrence and metastasis after surgery [10]. Although achieving detection of CEA using traditional means such as enzyme-linked immunosorbent assay, fluoimmunoassay, electrochemiluminescence assay and electrochemical immunoassay [11–15], these methods usually have certain limitations in clinical application, such as high cost, complicated operation and susceptibility to interference. Therefore, it is of great significance to explore a cost-effective, rapid, sensitive and accuracy strategy for CEA detection in clinical practice.

Surface-enhanced Raman spectroscopy (SERS) based on an inelastic scattering process has been developed as a powerful sensing tool, and its immunity to interference from aqueous solvents provides a novel and effective strategy for biomedical applications, especially for the detection of various cancer biomarkers [16–18]. So far, SERS substrates based on Au and Ag nanostructures have been widely used for targeted detection of cancer biomarkers. For example, Wu et al. developed a core-satellite structured aptasensor for sensitive detection of target biomarkers. This Au@Ag core-Au satellites sandwich structure generated lots of “hotspot” around the metal nanoparticles, significantly enhancing the SERS signal of the probe [19]. Recently, Muhammad et al. used electrochemical deposition approach to prepare highly ordered Au nanoparticle (Au NP) arrays for label-free detection of biomarkers [20]. To further improve homogeneity and stability, Lin et al. combined nanoparticles encapsulating an internal standard molecular with surface molecular imprinting technology to produce a SERS sensor with spectral correction to achieve highly sensitive and specific detection of CEA [21]. This SERS substrate with uniformly distributed nanoparticles and internal standard calibration signals can effectively improve stability and reproducibility. Signal calibration process using internal standard molecules have shown vital influence for SERS-based quantitative assay, where diverse strategies with more convenient preparation should be explored for the increase of accuracy and reliability. Here, we report a novel SERS sensing platform with sandwich structure for the quantitative detection of CEA. The aptamer-functionalized SERS probes and the two-dimensional nanoparticle monolayers were combined via the target CEA to form a sandwich structure, which enhanced the Raman signal of probes in the “hotspot” region. Notably, the characteristic peak of the silicon wafer was employed to real-time correct the intensity of the Raman probe for stable and accurate concentration detection. Furthermore, this method was applied to the detection of real blood samples to evaluate its potential in clinical applications.

2. Materials and methods

2.1 Materials and instruments

Trisodium citrate and 4-Mercaptobenzoic acid (4MBA, 98%)were obtained from Shanghai Macklin Biochemical Co., Ltd. Bovine serum albumin (BSA), and hydrogen tetrachloroaurate hydrate (HAuCl₄·4H₂O) were purchased from Sinopharm Chemical Reagent Co., Ltd (Beijing, China).3-Mercaptopropionic acid (3-MPA), ethanol (≥99.7%), dichloromethane, polyvinyl pyrrolidone (PVP), n-hexane, 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide (EDC), N-hydroxysuccinimide (NHS), alpha-fetoprotein (AFP), carcinoembryonic antigen (CEA) and CEA antibody were all obtained directly from Fuzhou Bellman Biotechnology Co., Ltd. CEA aptamer was purchased from Sangon Biotech (Shanghai, China).Silicon wafers were purchased from Xuchen e-commerce firm. Unless otherwise specified, ultrapure water (resistivity of 18.2 MΩ·cm) purified by Milli-Q system was used throughout the work process. All glassware was soaked in chromic acid for 24 hours for cleaning, and then rinsed with pure water at least five times and dry in an oven. The detailed sequence information of CEA aptamer is as follows: 5’-SH-ATACCAGCTTATTCAATT-3’.

Transmission electron microscope (TEM) images were obtained using a Tecnai-G20 (FEI, USA), and scanning electron microscope (SEM) images were obtained from MIRA LMS (TESCAN, Czech Republic). UV-Vis absorption spectra were recorded with a Lambda 950 spectrophotometer (PerkinElmer, US). SERS spectra were obtained using confocal Raman microscope (Horiba, Japan) with a 785 nm excitation laser was used to measure the sample. The size distribution and the zeta potential of SERS tags were analyzed by dynamic light scattering (DLS) with a Malvern Zetasizer NanoZ instrument following dispersion in water.

2.2 Synthesis of Au NPs

The stable Au NPs were prepared according to the method reported by Frens [22]. Briefly, 1.5 mL of sodium citrate aqueous solution (1% by weight) was quickly added to 100 mL of boiling and stirring chloroauric acid solution (0.01% by weight). Keep boiling and stirring for 15 min until the color of the solution turns to wine red, and then cool the synthesized Au NPs solution to room temperature.

2.3 Preparation of SERS probes

CEA aptamer, 4MBA, BSA and the obtained AuNPs solution were used to assemble SERS probes. First, Au NPs were washed twice in ultrapure water by centrifugation (12,000 rpm for 10 min). Then, 40µL CEA aptamer (10uM) and 40uL 4MBA (10µM) were added to the Au NPs solution that was re-suspended in 2 mL of ultrapure water, followed by being stored overnight at room temperature after shaking. After 12 hours, the solution was washed twice more to remove the unbound material. 500µL BSA (3%wt) was added to the above solution and incubate for one hour to make nanoprobes. The final solution was centrifuged at 8000 rpm for 10 min and re-suspended in ultrapure water. These SERS probes were stored at 4°C for further use.

2.4 Assembly of two-dimensional nanoparticle monolayers

The assembly of the two-dimensional nanoparticle monolayers is based on the reported method of three-phase interface self-assembly [23]. Typically, 2 mL Au nanoparticle solution synthesized above was the basis for making SERS substrate. After centrifugation, the most of the supernatant was removed and the precipitate was dispersed into PVP solution (1wt%, dissolve in ethanol). After the second round centrifugation, the precipitate was re-suspended in ethanol.1 mL dichloromethane and 1.8 mL water were mixed with 200µL PVP stabilized nanoparticle solution in a 5 mL centrifuge tube that was then oscillated evenly and left standing until the bubbles disappear. Subsequently, 200µL n-hexane was added along the tube wall to form a closely packed nanoparticle monolayer. After the upper layer of n-hexane was removed, the nanoparticle monolayer was transferred to silicon wafers.

2.5 Assembly of the SERS sensing platform

The above two-dimensional SERS substrate was immersed in 3-mercaptopropionic acid solution (10µM) for 12 hours. Then, the processed substrate was rinsed thoroughly with ultrapure water and reacted with the mixed solution of EDC (10µM) and NHS (10µM) for two hours to activate the carboxyl group. Subsequently, it was incubated with CEA antibody (0.05 mg/mL) for 2 hours to complete the functionalization. The functionalized two-dimensional substrate reacted with CEA for two hours and then washed with ultrapure water. The SERS probes and substrate were incubated at room temperature The SERS probes and substrates were incubated for 2 hours at room temperature and washed again with ultrapure water. The sandwich structure of SERS sensing platform for detecting CEA was naturally air-dried at room temperature and reserved for further use.

2.6. Collection of clinical blood samples

Serum samples were collected from 10 healthy individuals and 10 breast cancer patients from the First Hospital of Fujian Medical University in a male to female ratio of 1:4, 6 cases aged >45 years and 4 cases aged <45 years. After a 12-hours fast, 1 ml of whole blood was collected between 7 am and 8 am. Each sample was centrifuged at 1000 rpm for 10 minutes to separate the serum, and the supernatant was collected and stored at −80°C for storage. Informed consent was obtained from all participants.

2.7 Measurement of the SERS spectrum

After wavelength calibration of the Raman spectrometer with a single crystal silicon 520 cm⁻¹ Raman peak, the samples were measured using a confocal Raman spectrometer (Horiba, Japan) with a 785 nm laser (laser power 5.8 mW) and a 50x objective (laser size 3um) in the range 600-1800cm⁻¹. All demonstrated spectra are the average of three random spectra for each sample, and each spectrum was integrated for 10 seconds.

3. Results and discussion

3.1 Mechanism and characterization of the sensing platform

Figure 1 illustrates the typical scheme we designed to capture and detect CEA. Briefly, CEA aptamer-modified 4MBA@Au nanoparticles served as SERS probes, while CEA antibody-modified gold nanoparticle monolayers served as immune substrates. When an immune reaction occurred, the immune-modified gold nanoparticle monolayers captured the free CEA, which further bound to the aptamer-functionalized nanoparticles, thus immobilizing the SERS probe with signal molecules on the nanoparticle monolayers and forming a sandwich complex of SERS nanoprobe/target protein/AuNPs monolayers. This sandwich structure was the key part in this SERS sensing platform. The SERS intensity of 4MBA 1583 cm⁻¹was found to be correlated with the concentration of CEA, due to that higher concentration of CEA led to more SERS probes being immobilized on the monolayers. Therefore, the concentration of CEA could be quantified by detecting the intensity from the characteristic Raman peaks of Raman probe molecules.

 

Fig. 1. Assembly schematic of SERS sensing platform based on sandwich structure: fabrication process of aptamer-functionalized SERS probes and immune-modified nanoparticle monolayers and specific capture of target CEA.

Download Full Size | PDF

3.2 Characterization of the SERS sensing platform

Figure 2(a) indicated that the average diameter of the gold nanoparticles used to assemble the sandwich structure is 40 nm, which was further confirmed by the transmission electron microscope (TEM) in Fig.2e. At the same time, the UV–vis spectrophotometer scanned Au NPs, Au@4MBA&Apt@BSA NPs and Au@4MBA&Apt@BSA@CEANPs in the wavelength range of 400-700 nm (Fig.2b). Results showed that AuNPs had a maximum absorption peak at 522 nm. Compared with Au NPs, the diameter of Au@4MBA&Apt@BSA NPs increased to 99 nm, and the corresponding peak position of the maximum absorption peak red-shifted from 522 nm to 533 nm.Meanwhile, the successful modification changed the zeta potential of the nanoparticles. The Au NPs that originally had a negative charged rose from −21.1 mV to −14.5 mV. According to previous studies, the characteristic vibrational modes observed in the SERS spectra of the 4-MBA, including the ν (CC) ring respiration mode (∼1070 and 1575 cm⁻¹), and other less intense modes, including δ (CH) (1132 and 1173 cm⁻¹) and νs (COO-) (1375 cm⁻¹) [24–26]. In addition, after adding 4MBA and aptamer to gold nanoparticles, strong characteristic peaks of Raman reporters could be observed (Fig.2d), and the BSA used in the blocking site had little effect on the SERS signal of the signal molecule. The above characterization proves the successful synthesis of SERS probe. During the assembly process of the SERS sensing platform, the hydrodynamic diameter of the SERS probe with the addition of CEA increased to 135 nm (Fig.2a). The corresponding UV absorption spectrum peak position and the zeta potential were respectively found to have about 9 nm red shift (Fig.2b) and a change of 3.2 mV (Fig.2c), which well proved the success of CEA capture. This phenomenon was attributed to the fact that the Zeta-potential is a model quantity determined by measuring the electrophoretic mobility in suspension, the value of which depends on the properties of the nanomaterial, the solution conditions and the theoretical model applied. When the SERS probe captures CEA, it changes the surface charge of SERS probe [27].

 

Fig. 2. Size distribution (a), UV-vis spectra (b) and zeta potentials (c) obtained for Au NPs (I), Au@4MBA&Apt@BSA NPs (II) and Au@4MBA&Apt@BSA@CEA NPs (III). (d) SERS spectra of SERS sensor platform assembly process. (e)TEM image of Au NPs. (f) SEM image of self-assembled two-dimensional substrate, insert: SERS mapping for two-dimensional substrate.

Download Full Size | PDF

Figure 2(f) shows a scanning electron microscope (SEM) image of a self-assembled two-dimensional substrate. Au NPs were evenly distributed on the surface of the silicon wafer to form a dense monolayer. The Raman intensity mapping in the inset further demonstrated the high homogeneity of the two-dimensional substrate over an area of 200µm.After immune-modification of the two-dimensional substrate, some Raman signals appeared on the substrate (Fig2.d), which might be caused by the modified 3-MPA, CEA antibody and the activation process. In particular, the “background noise” generated by these immune modifications had almost no effect on the Raman signal of the characteristic peak of the signal molecule 4MBA at 1583 cm⁻¹. High-performance SERS probes and uniformly distributed two-dimensional substrate provided the foundation for the outstanding performance of the SERS sensing platform for CEA detection.

3.3 Standard curves and specificity for quantitative assay

To evaluate the quantitative detection performance of the SERS sensing platform, various concentrations of CEA (1 ng/mL-10µg/mL) were prepared to obtain characteristic SERS spectra in standard solutions. In the presence of CEA, nanoparticles with Raman report molecules were assembled onto the substrate and the resulting sandwich structure further enhanced the SERS signal intensity. Fig.3a and Fig.3b show the SERS spectra obtained in the concentration range of 1 ng/mL-10µg/mL and each SERS spectrum with intensity calibration was the average of three spectra randomly collected from the sandwich structure platform. The result indicated that the SERS spectral intensity of peak at 1583 cm⁻¹ gradually decreased associated with decreasing CEA concentration. It has obviously observed that the intensity of 4MBA at 1583 cm⁻¹ was strongly related to CEA concentration. In this work, the typical peak at 1583cm⁻¹ of the signal molecule and the peak at 940 cm⁻¹ of the silicon substrate were selected to carry out the ratiometric analysis (Fig.3c). Correspondingly, the I_1583cm-1/I_940cm-1 is utilized as the quantitative value and the calibration curve is y = 0.79(lgC)²−0.541 lgC + 0.383, with a correlation coefficient square (R²) of 0.9796 (y is the I_1583cm⁻¹/I_940cm⁻¹, and C represents the corresponding CEA concentration). The minimum detection limit of the SERS platform is 0.258 ng/mL. Although the detection limits are not particularly outstanding, the SERS platform as a CEA assay has meet the requirement of clinical testing [28–32]. In addition, we plotted the SERS spectra for 1583 cm⁻¹ absolute intensity (Fig. 3(d)). Although this spectrum also shows a similar trend to the method of ratiometric analysis, the regression curve with intensity calibration has a lower linear correlation and higher standard deviation (Fig. 3(e)). In order to evaluate the specificity of the proposed SERS sensor platform, two common proteins (AFP and BSA) solutions were added to the sensor platform respectively. As shown in Fig. 3(f), the blank solution has extremely low Raman background signal at 1583 cm⁻¹, while the presence of CEA makes the detection platform produce a significant Raman signal. The sandwich structure showed almost no significant response to other non-target proteins compared to the CEA solution, indicating that this SERS sensing platform has excellent specificity for CEA.

 

Fig. 3. (a) Raman spectra with different concentrations of CEA from 1 ng/mL−10µg/mL obtained on CEA SERS sensing planform. (b) The enlarged portion at the dotted line in (a). (c) CEA detection logarithmic concentration curves generated by ratio of the peaks at 940 cm⁻¹and 1583 cm⁻¹ (I₁₅₈₃/I₉₄₀). (d)Average Raman spectra of different concentrations of CEA without intensity correction. (e)CEA detection logarithmic concentration curves generated by plotting the average intensity at 1583 cm⁻¹. (f)Selectivity of the detection of CEA (1 ng/mL), with the blank control and the interfering substances of BSA and AFP presented at the same concentration.

Download Full Size | PDF

3.4 Clinical samples analysis

Due to the excellent performance of the sensing platform, we used it to explore the effectiveness of CEA detection in real serum samples. A total of 20 serum samples were collected from 10 healthy individuals and 10 breast cancer patients. Figure 4(a) shows the average Raman spectra obtained from the Raman sensing platform treated with serum samples. The standard deviations (SD), overlying as shaded color fill in the graph, for each group demonstrated the good spectral reproducibility within each group. Compared with the normal group, the Raman intensity of the characteristic peak (1583 cm⁻¹) in the average SERS spectrum of breast cancer patients changed significantly. This is further verified by the intensity ratio distribution of the Raman spectrum at 1583 cm⁻¹ in Fig. 4(b). The difference of Raman spectral intensity at 1583 cm⁻¹ may be caused by the relatively high concentration of CEA in the patients’ serum. In patients with tumors, serum containing high CEA concentrations (>10 ng/mL) can be diagnosed as positive [28,31,33–35]. Similarly, we compared the detection results obtained with and without the correction procedure as shown in Fig4.c and d. As we expected, the ratiometric analysis of Raman spectrum at 1583 cm⁻¹yields even better result that the normal and patient groups can be well distinguished with 100% accuracy. This result indicates that the sandwich structure of the SERS sensing platform is effective for detecting the tumor marker CEA in human serum.

 

Fig. 4. (a) Comparison of the mean SERS spectra for the normal serum versus that of the breast cancer with 940 cm⁻¹ intensity correction. (b) SERS intensity ratio (I_1583cm⁻¹/I_940cm⁻¹) in 20 serum samples obtained on CEA SERS sensing planform. (c) Distribution of spectral peak intensities at I_1583cm⁻¹/I_940cm⁻¹ in healthy humans(n = 10) and breast cancer patients (n = 10). (d) Distribution of spectral peak intensities at 1583cm⁻¹ in healthy humans(n = 10) and breast cancer patients (n = 10). *P value < 0.05. **P value < 0.01. ***P value < 0.001.

Download Full Size | PDF

4. Conclusion

In this work, we designed a SERS sensing platform with a sandwich structure that can rapidly and sensitively detect the tumor marker CEA. Gold nanoparticles with Raman reporter molecule 4MBA were synthesized as SERS probes, while gold nanoparticles with unmodified signal molecules self-assembled to form nanoparticle monolayers, both of which served as components of the SERS sensing platform. When an immune reaction occurred, the target protein CEA immobilized the free SERS probe on the uniform nanoparticle monolayers to form a sandwich structure, so that the Raman signal intensity on the SERS probe was greatly enhanced. In addition, we used the Raman peak of silicon wafers to calibrate the signal of the Raman reporter, which can effectively decrease the signal fluctuation. The Raman signal intensity of the characteristic peaks on the SERS sensing platform was detected to achieve CEA quantitative detection. The limit of detection (LOD) was 0.258 ng/mL, which fully meet the clinical detection requirement for CEA. Using this sensing platform, the real serum samples from tumor patients can be well identified from the healthy subjects with 100% discriminative accuracy with 100% discriminative accuracy. This assay CEA platform is easy to operate, specific, rapid and performs well in the detection of clinical serum samples, making it promising to become an alternative method for biomarkers detection in clinical practices.