Graphite powder with a purity of 99.95% and Aβ_1–42 peptides were purchased from Sigma-Aldrich (St. Louis, MO, USA). Trifluoroacetic acid (TFA), magnesium chloride (MgCl₂), sodium nitrate (NaNO₃), sulfuric acid (H₂SO₄), hydrogen peroxide (H₂O₂), potassium permanganate (KMnO₄), acetonitrile (ACN), TFA, sodium chloride (NaCl), thiol poly(ethylene glycol) amine (SH-PEG-NH₂, purity ≥ 95%), N-hydroxysuccinimide (NHS, purity ≥ 98%), and 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide (EDC, purity ≥ 99%) were obtained from Merck (Darmstadt, Germany). High purity water (18.2 MΩ⋅cm) was filtered by a Milli-Q system (Millipore, Bedford, MA, USA). The other chemicals and solvents of analytical grade were obtained from Xilong Scientific Co., Ltd. (Guangzhou, China). The SH-SY5Y cell line was obtained from BeNa Culture Collection (Beijing, China). Dulbecco’s modified essential medium (DMEM), fetal bovine serum (FBS), penicillin, streptomycin, all-trans retinoic acid (ATRA), and brain-derived neurotrophic factor (BDNF) were purchased from Gibco (SanFrancisco, CA, USA). Human serum samples were provided by Linping Hospital of Traditional Chinese Medicine (Hangzhou, China).

The aptamer for Aβ_1–42 (5′-AGT CTA GGA TTC GGC GTG GGT TAA TTT TTT GCT GCC TGT GGT GTT GGG GCG GGT GCG-3′) was previously reported and synthesized by Sangon Biotechnology Inc. (Shanghai, China) (Chan et al., 2019). It was modified with a 5′-thiol modifier for GO modification and further labeled with fluorescein at the 3′-end for detecting and quantifying the modification on the GO surface.

GO was synthesized according to a modified Hummer’s method (Zhao et al., 2020; Song et al., 2021). In brief, 2 g of graphite powder and 1 g of NaNO₃ were dissolved in 40 mL H₂SO₄ in a round bottle flask in an ice bath, because the reaction could produce a slight exotherm. After cooling to room temperature (ca. 25°C), the ice bath was removed and the reaction was sustained for 30 min after the gradual addition of 6 g KMnO₄ under stirring. Then, 30% H₂O₂ solution (20 mL) was gradually added to remove excess KMnO₄ until the mixture turned brilliant yellow. Afterward, the synthesized GO mixture was neutralized using 5% HCl and ultrapure deionized water after centrifugation (10,000 g for 15 min at 4°C). The residual salt in the obtained solid material was removed by dialysis in deionized water. Finally, the synthesized GO was dried under vacuum at room temperature overnight and characterized by a scanning electron microscope (SEM, FEI Inspect F50, Thermo Fisher Scientific, MA, USA).

The procedure for fabricating and utilizing Apt-GO for detecting Aβ_1–42 under SADLI MS was displayed in Figure 1. According to the method described in Gulbakan et al. (2010) with slight modifications, the PEGylation of GO was performed (Gulbakan et al., 2010). In brief, the GO solution (1 mg/mL) was dispersed in PBS buffer (pH = 7.4) containing EDC (0.0022 M) and NHS (0.0015 M), followed by sonicating for 8 h to activate the carboxyl groups. Afterward, HS-PEG-NH₂ (0.1306 M) was added to the solution and the mixture was sonicated for 1 h at room temperature. By centrifugation at 10,000 g for 10 min four times, the excess ligand was washed away. The precipitate was redispersed in 100 μL of PBS (pH = 7.4). The aptamer solution was heated at 90°C for 30 min, followed by cooling at room temperature. The obtained PEGylated GO (100 μL) was incubated with 5′-SH-Aptamer in PBS (200 μL). Then, the mixed solution was centrifuged at 10,000 g for 10 min and washed several times to remove the excess unbound aptamers. The precipitate was dispersed in 300 μL ultrapure water.

The preparation of Aβ_1–42 was carried out by using SH-SY5Y neuroblastoma cells as AD model cells according to the previous study (Feng et al., 2021). The SH-SY5Y cell line was cultured in DMEM. Before experiments, the SH-SY5Y cells were grown for three generations and used at a low passage number (<13). In detail, the cells were cultured in a humidified incubator at 37°C with 5% CO₂. The culture medium was supplemented with 10% heat-inactivated FBS and 100 U/mL penicillin/streptomycin, and it was replaced three times per week. After reaching 70–80% confluence, the cells were differentiated with ATRA, DMEM, and BDNF in the medium. The treated cells with neuron-like phenotype were processed with Aβ_1–42 for 24 h and harvested as AD models.

During the detection of Aβ_1–42, GO played the role of matrix in SALDI-MS analysis. GO (1 mg) was dispersed in ultrapure water (1 mL) under sonication for 20 min. The above GO slurry (2 μL) was successively pipetted on the stainless steel target plate and dried for SALDI-MS analysis. Based on the reflection mode employing delayed extraction, the MS analysis was performed on 5800 Proteomics Analyzer (Applied Biosystems, Framingham, MA, USA) coupled with a pulsed Nd:YAG laser (1 kHz, 355 nm wavelength). In this study, the delay time, acceleration voltage, and repetition rate were optimized to 200 ns, 20 kV, and 200 Hz, respectively. For the SALDI-MS analysis, the number of laser shots was 200 per analysis.

For selective enhancement of Aβ_1–42, the Apt-GO slurry was centrifuged at 4,000 g for 15 min and the precipitate was redispersed in the binding buffer (20 mM Tris-HCl, 140 mM NaCl, and 2 mM MgCl₂, pH = 7.5). Then, the Aβ_1–42 standards and disrupted AD cells model solution were added. The mixture was incubated at 37°C for 30 min and centrifugated at 4,000 g for 15 min. Afterward, the precipitate was redispersed in 0.1% TFA solution for SALDI-TOF/MS analysis as described above.

The MS data were processed by the Data Explore Software called Analyst (AB Sciex, USA). Statistical analysis was carried out by Microsoft office software, IBM SPSS Statistics (version 21.0), and the analysis of variance (ANOVA) followed by Duncan’s new multiple range tests. All experiments above were performed at least in triplicate.

The morphology and surface of graphite and GO were characterized by a scanning electron microscope (SEM, FEI Inspect F50, Thermo Fisher Scientific, MA, USA). As seen in Figure 2A, the interlayer space of graphite was small and the surface was relatively smooth. After chemical modification (Figure 2B), the rough surface and larger lamellar structure were formed with many wrinkles due to the presence of oxygen-containing functional groups. It was reported that the surface of GO contains many polar moieties, mainly including hydroxyl epoxy and carboxy groups, which ensures that the GO has more polar and hydrophilic characteristics (Du et al., 2020).

Typically, PEGylation of GO nanoparticles is achieved by a ligand exchange process using HS-PEG-NH₂ molecules as a spacer unit. In this study, the bifunctional PEG linker carriers, –NH₂ groups, bind to the carboxyl groups on the surface of GO nanoparticles, and the –SH groups attach to the SH-functionalized aptamers by forming a disulfide bond (Xia et al., 2012). According to Fourier transform infrared (FTIR) spectrum depicted in Figure 2C, the characteristic peaks of GO nanoparticles at 1,720 and 1,610 cm^–1 were carboxyl (C = O) stretching and vibration of carbon skeleton (C = C), respectively. The characteristic peak of the symmetric tensile vibration of the epoxy group (C-O) was evident at 1,050 cm^–1. The result was in line with the previous reports (Du et al., 2020), which further demonstrated that GO nanoparticles were fabricated successfully. In addition, the FTIR spectrum of aptamer-modified GO was characterized. The N-H bending and C-N stretching vibration peaks were observed in the aptamer-modified GO nanoparticles at 1,535 cm^–1, indicating the conjugation of HS-PEG-NH₂ with GO. This is attributed to the exposure to amine groups of GO resulting in nucleophilic attacks (Caicedo et al., 2020). The characteristic peak of a disulfide (S-S) bond was observed at 540 cm^–1, representing the cross-linking of SH-functionalized aptamers and the thiol groups in the HS-PEG-NH₂ molecules. The above result verified that the conjunction between aptamers and GO nanoparticles using HS-PEG-NH₂ molecules was successful.

The aptamer binding capacity of the Apt-GO nanoparticles was evaluated by comparing the UV adsorption value of the supernatant before and after immobilization at the wavelength of 260 nm. The binding capacity was calculated to 36.1 nmol/mg (RSD = 11.5%, n = 3), much higher than that of GO (50 pmol/mg). The higher immobilized amount should be attributed to the ultra-high specific surface area of PEGylation of GO nanoparticles.

To evaluate the selectivity of the Apt-GO sensor, a schematic analysis of the GO-based aptamer-assisted MS analysis was carried out. As seen in Figure 3, four replication experiments were designed. Group A was used to prove the function of GO as the matrix for ionization on SALDI MS analysis. The GO solution was spotted on the SALDI plate and dried under room temperature; then, the analyte was spotted on the GO matrix and dried for laser desorption ionization. Group B was used to analyze the analyte extracted with GO and ionized on GO. The GO was added to the analyte solution for extraction. Afterward, the mixture was washed rigorously, and the supernatant was removed. The obtained GO pellets were spotted on the SALDI plate for ionization analysis. Group C was used to analyze the analyte extracted with non-covalent-aptamer bound GO. The mixture of the aptamer and GO solution was added to the analyte for selective enrichment. Group D was designed to evaluate the selectivity of the Apt-GO by covalent bonding. The aptamer-modified GO was added into the analyte for specific enrichment. The following steps were the same as in the above experiments.

The performance of the Apt-GO was characterized by SALDI MS. Figure 4A shows the MS analysis of Aβ_1–42 in the AD cell model analyzed on GO served as the matrix without aptamer modification and any extraction step. Under the positive ionization mode, the SALDI-MS of the Aβ_1–42 could be detected in m/z of 4,515.7 with an average signal-to-noise (S/N) value of 15 among huge background ions. The analysis result of the AD cell model containing Aβ_1–42 enriched and ionized with GO nanoparticles without modification is displayed in Figure 4B. The peak intensity and S/N value were enhanced slightly, and the background ions were reduced. This was mainly attributed to the unique morphology of GO nanoparticles that could facilitate the diffusion of Aβ_1–42, which was beneficial for the extraction of Aβ_1–42 from complex biological samples. In addition, it was found that the π-π and hydrophobic interactions existed between GO nanoparticles and Aβ_1–42 peptide, which could further stabilize the cross-linking between GO nanoparticles and Aβ_1–42 by salt bridges and H-bonds (Sun et al., 2016; Jin et al., 2019). As seen in Figure 4C, the peak intensity of Aβ_1–42 was low and the S/N value of 17 was obtained. This result could be attributed to non-covalent-aptamer bound GO. During the washing step, the aptamer-captured parts of Aβ_1–42 were released from the surface of GO due to the absence of chemical functionalization. However, the background of MS was relatively clear to some extent, which could be ascribed to the physical adsorption capacity of GO (Sun et al., 2008). In this study, the MS signal was generated from the surface of GO. After the washing step, the peak signal of Aβ_1–42 decreased because of the removal of the aptamer-captured Aβ_1–42. On the contrary, the peak intensity and S/N value of Aβ_1–42 were both enhanced significantly (p < 0.05) due to the covalent modification between GO and the aptamer by chemical conjugation (Figure 4D). Figure 4E showed the peak of Aβ-42 standard spiked in human serum sample, which was used for comparison. This result demonstrated that the performance of Apt-GO for selective enhancement of Aβ_1–42 was efficient, and chemical conjugation of the aptamer was critical for analyte capture. This result was in line with the reported study for GO-based fluorescence sensors. It revealed the restoration of fluorescence that had previously been quenched by GO after the aptamer left the surface upon target addition (Wang et al., 2010).

It is necessary to evaluate the recognition reproducibility of Apt-GO for capturing target protein. In this study, one batch of Apt-GO was fabricated and stored at 4°C every week. A total of five batches of the sensors were fabricated within 5 weeks. All obtained sensors were incubated with Aβ_1–42 at the same time after the final batch was fabricated. According to the SALDI-MS result, the Aβ_1–42 could be captured by Apt-GO fabricated in different weeks with similar peak intensities and S/N values (RSD, 15.2%). This result was attributed to the stable covalent bond between GO and aptamers by the π-π and hydrophobic interactions, which exhibited enhanced resistance to nucleases and protected aptamer from enzymatic digestion effectively (Tang et al., 2012).

The reusability of Apt-GO for target proteins is essential in practical applications. It reveals that the conformation between aptamer and target protein is formed and maintained by the intra-molecular forces, mainly including hydrogen bonding and hydrophobic and van der Waals interactions. Once this conformation changes, the binding affinity between the aptamer and its target will disappear (Ocsoy et al., 2013). In theory, aptamers undergo conformational changes upon thermal denaturation. In this study, the Aβ_1–42 was released from Apt-GO by heating. The Apt-GO could be reused after centrifugation and washing procedures. The reusability of Apt-GO for Aβ_1–42 was carried out at 65°C, and the result revealed that the enrichment efficiency was nearly 70% loss after three extraction cycles (Figure 5). This result was mainly ascribed to the destroyed conformation of aptamers and broken chemical conjugation between GO and aptamers caused by thermal denaturation. This result indicated that the enrichment performance of recycled Apt-GO would degenerate significantly (p < 0.05).

To further investigate the feasibility and practical applicability of Apt-GO in the measurement of Aβ_1–42, human blank serum samples with different concentrations of Aβ_1–42 were prepared (0.1, 1, and 10 μM). Under the same condition, the prepared samples were incubated with Apt-GO independently. As seen in Table 1, the recoveries of samples were in the range of 95.10–101.28%, indicating superior detection performance, dependability, and accuracy in real samples. Although the data for the three samples were limited, the work demonstrated that it was possible to distinguish patients from healthy people by improving the sensitivity of the biosensor for AD biomarkers.