Nonenzymatic Amperometric Aptamer Cytosensor for Ultrasensitive Detection of Circulating Tumor Cells and Dynamic Evaluation of Cell Surface N-Glycan Expression.

Dynamic assessment of glycan expression on the cell surface and accurate determination of circulating tumor cells are increasingly imperative for cancer diagnosis and therapeutics. Herein, a unique and versatile nonenzymatic sandwich-structured electrochemical cytosensor was developed. The cytosensor was constructed based on a cell-specific aptamer, the lectin-functionalized porous core-shell palladium gold nanoparticles (Pd@Au NPs). To establish the cytosensor, amine-modified-SYL3C aptamer was first attached to the surface of aminated Fe3O4@SiO2 nanoparticles (Fe3O4@SiO2-NH2 NPs) through cross-linked reaction via glutaraldehyde. Besides, in terms of noncovalent assembly of concanavalin A on Pd@Au NPs, a lectin-functionalized nanoprobe was established. This nanoprobe had the capabilities of both the specific carbohydrate recognition and the current signal amplification in view of the Pd@Au NPs as the electrocatalyst for the reduction of hydrogen peroxide (H2O2). Herein, we used MCF-7 cells as a model target, and the constructed cytosensor showed a low detection limit (down to three cells), a wide linear detection ranging from 100 to 1 × 106 cells mL-1. The established method sensitively realized the detection of the amount of cell and exact evaluation of glycan expression on cell surface, demonstrating great application prospects.

Introduction

Recent statistics reveals that cancer has turned into one of main causes of human death.[1] As related reports illustrated, cancer metastasis resulted in most of the cancer-relevant deaths.[2] When cancer cells move into the peripheral blood or lymphatic system from the original tumor position, they will turn into the circulating tumor cells (CTCs), being considered as an important element in cancer growth and metastasis[3] as well as becoming biomarkers for early cancer diagnosis.[4] Thus, it is important to detect the CTCs early toward effective treatment of cancers.[5] However, owing to the quantity of CTCs is greatly rare of the early-stage cancer patients, approximately ranging from 1–10 cells every milliliter, leading to the difficulty for their isolation and quantitative detection.[6,7] To make the detection of CTCs more sensitive, various developed approaches have been established containing fluorescence sensing and imaging,[8] X-ray radiography,[9] PCR,[10] flow cytometry,[11] and spiral computed tomography.[12] Whereas, a majority of establishing methods had some drawbacks, including time-consumption, high-cost in the experimental procedures, and sophisticated instruments. Therefore, it is significant to set up a sensitive and accurate approach to detect CTCs in clinical analysis.

All eukaryotic cell surfaces were covered by plentiful carbohydrates that are covalently associated with potential lipids and proteins.[13] There is no doubt that it is significant for carbohydrates on the cell surface which are generally considered as the surface biomarkers to regulate many cellular processes, containing cell–cell communication, differentiation, and signaling.[14] In terms of the cellular condition and status, it is a dynamic process for the expression of glycan on cell surface.[15] The N-linked glycosylation (N-glycan) is involved in the process to disturb the interactions of a biomolecule; therefore, they control many important physiological processes, for instance, the growth of the tumor,[16] and dynamic alters of N-glycans on epidermal growth factor were measured to be related to the expression of Mgat5 which effects cell activity and tumor metastasis.[17] Upon the processes of tumorigenesis, differentiation, and brain aging, the abnormal expression of N-glycan on the cell surface was also observed.[18] Hence, it is important for the dynamic analysis of cell surface N-glycans in the diagnosis and treatment of disease.

Currently, more and more methods have been applied for glycan analysis.[19−21] Whereas most of these approaches are inappropriate for glycan determination of living cells in situ as a result of their destructive procedures and complicated apparatus even though they comprehensively reveal elaborated structure information. Recently, electrochemical glycan biosensors[22] on account of lectins, a type of glycan-specific recognizing proteins,[23] have been rapidly used in glycan analysis to partially compensate the drawbacks with underlying utilizations for easy and dynamic evaluate of glycans on cell surfaces.

Concanavalin A (Con A) not only combines with mannose but also has greater interaction to the part of N-glycans.[24] Con A can be bounded with glycoproteins including high mannose, hybrid, and biantennary compound variety of N-glycans.[25] On this basis, N-glycosylated positions of glycoproteins were further mapped by Con A.[26,27] Therefore, based on the affinity between Con A and the trimannoside motif, the majority of N-glycans on the cell surface could be monitored and recognized.

Electrochemical biosensors have aroused a widespread concern as a result of their overwhelming advantages, for example, low detection limit and cost, simple preparation, and quick detection.[28] To increase the high sensitivity of electrochemical biosensors, many efforts have been made via several signal amplification approaches.[29] Upon them, as a result of the excellent specificity and efficiency of enzymes, the amplified approach of enzyme catalysis has drawn considerable interests.[30] However, the use and selection of enzymes have several drawbacks, such as, poor stability, complex immobilization, and high sensitivity to pH and temperature.[31] To circumvent the above limitations, enzyme-free electrochemical cytosensor based on nanomaterials has recently been developed, which was as an effective strategy for selective and sensitive analysis.[32,33] It was reported that a variety of nanomaterials related to metal and metal oxide have been applied to set up effective electrochemical biosensor as electrocatalysts because of its possessed intrinsic peroxidase-like activity.[34] Noble-metal-related nanomaterials, particularly, Pd, Pt, and Au, have extensive applications owing to the excellent biocompatibility, outstanding catalytic ability, and high ratio of surface.[35] Lately, core–shell bimetallic nanocrystals have exhibited excellent catalytic property over the corresponding monometallic nanomaterials on account of the synergistic and electronic effect.[36] Their catalytic characteristics are attributed to shape and size, as well as structure and composition.[37] More and more attention has been drawn on the synthesis of bimetallic nanoparticles (NPs) upon highly porous/dendritic structures.[36] On the basis of their large surface area, the dendritic nanocrystals are in favor of excellent catalytic property.[38] In view of the above theory, core–shell palladium gold NPs as mimetic peroxidase were synthesized to detect cancer cells by electrochemical method.

Aptamers as a type of synthetic oligonucleotides have specific and high interactions to a series of small molecules, proteins, and cells.[39] Compared with antibodies, aptamers have considerable characteristics, for instance, low molecular weight, small size, tunable sequences, and programmable structure. In view of the above advantages, aptamers have become powerful identification ligand in medical diagnosis. Lately, functional aptamers have been employed for targeting a large amount of cancer cells, and aptamer-combined nanomaterials have been widely applied to improve molecular affinity.[40] By taking advantage of the high recognition and good biocompatibility, functional aptamer composites provided an outstanding approach to design electrochemical cytosensor in early medical diagnostics.

Owing to the superior characteristics and excellent biocompatibility of nanocrystals, more and more nanomaterials have been employed for the immobilization and electrochemical analysis of cells.[14] Lately, magnetic Fe3O4 NPs have attracted a considerable attention owing to their unique magnetic feature, high ratio of surface, low toxicity, and simple preparation.[41] Importantly, they can load more biomolecules.[42] However, naked Fe3O4 NPs are unstable in air, easily aggregating, and absent in the activating group to binding, which limited their further application in electrochemical biosensor. To overcome the above drawbacks, Fe3O4 NPs are coated by silica with a core–shell structure that can not only prevent Fe3O4 NP aggregation but also be functionalized easily to increase compatibility. On the basis of the above advantageous properties, Fe3O4@SiO2 nanostructures can be applied as the nanocarriers to immobilize aptamer in our electrochemical cytosensor.

Here, we develop a nonenzymatic sandwich-structured biocompatible amperometric nanobiosensor for ultrasensitive detection of CTCs and synchronous dynamical assessment of N-glycan expression on CTCs surface. The cytosensor was fabricated by taking advantage of Pd@Au NPs acting as a signal amplification probe and aminated Fe3O4@SiO2 NPs as nanocarriers. In the experiments, MCF-7 cells were used as model CTCs lines. Scheme illustrated the established process of the amperometric cytosensor. The cytosensor was fabricated through these steps: first, amine-functionalized Fe3O4@SiO2 nanostructures were cross-linked with amine-terminated cell-targeting aptamers, SYL3C–NH2,[43] via glutaraldehyde (GA) to build a biocompatible interface to capture CTCs. The aptamer successfully immobilizing on the nanocomposite interfaces was confirmed through UV–vis spectra, zeta (ζ) potential, and confocal fluorescence image characterizations. Subsequently, bovine serum albumin (BSA) was applied to block the nonspecific absorption sites. Afterward, Pd@Au NPs were conjugated with Con A through Au–N bond and electrostatic interactions, which acted as a glycan recognition unit as well as signal amplification probe. UV–vis absorption spectra were measured to manifest the successful conjugation between Pd@Au NPs and Con A. Finally, the sandwiched cytosensor was established by incubating Fe3O4@SiO2–NH2/aptamer/BSA with CTCs and Pd@Au-conjugated Con A successively. As expected, the established biosensor indicated an advantageous selectivity and sensitivity to assay CTCs. Furthermore, when the external stimuli of N-glycan inhibitor or released enzyme interfere, the developed cytosensor could effectively react with the variation of cell-surface glycans. Although in contrast with untreated cells, negligible change of the current signal was measured by the O-glycan inhibitor. Therefore, the above performance suggested the feasibility of assessing N-glycan expression on cell surface through the prepared electrochemical cytosensor. Thus, the proposed cytosensor possessed the ability for early diagnosis of cancer and opened an avenue for understanding the glycan function physiology of in the cellular events and clinical analysis of glycoprotein relevant diseases.

Scheme 1

Schematic Illustration of the Fabrication Process of Proposed Electrochemical Cytosensor

Results and Discussion

Characterization of Fe3O4@SiO2 Nanostructures

The transmission electron microscopy (TEM) image is depicted in Figure A,B, from which the black Fe3O4 was surrounded by light-gray SiO2 shell, demonstrating the Fe3O4@SiO2 was synthesized successfully. The scanning electron microscopy (SEM) image (Figure C) shows a large number of uniform spherical Fe3O4@SiO2 NPs with smooth surfaces. It can be seen from Figure D that the average size of the composite was estimated to be 389 nm.

Figure 1

(A,B) TEM and (C) SEM images of Fe3O4@SiO2 nanocomposites (D) particle size distributions of Fe3O4@SiO2 nanocomposites.

The magnetic properties of the as-prepared Fe3O4 and Fe3O4@SiO2 NPs were explored by a VSM at 294 K. Figure S1 depicts the typical magnetization curves of Fe3O4 and Fe3O4@SiO2 NPs with the saturation magnetization (Ms) at 14 kOe of 59.8 and 49.9 emu g–1, respectively. The 16% decrease could be attributed to increase the diamagnetic SiO2 shell. Even though the Ms value of Fe3O4@SiO2 NPs was inferior to Fe3O4, it also had an outstanding magnetic response and the ability to separate easily from the solution under low magnetic field. As shown in the inset, upon putting a magnet next to the bottle, Fe3O4@ SiO2 nanostructures were quickly attracted to the side of the vial, making the solution clear. Although the magnet was taken away, Fe3O4@SiO2 NPs could be redispersed in the solution again via ultrasound and shaking. The above results suggest Fe3O4@SiO2 NPs can be applied in the area of biological separation and detection.

To confirm the successful amine modification of Fe3O4@SiO2 NPs, Fourier transform infrared (FT-IR) spectra of the as-prepared nanocomposites were also measured. As depicted in Figure S2, in the case of Fe3O4@SiO2 NPs, the symmetric and asymmetric stretching vibrations of the absorption bands of Si–O–Si in the silica shell was centered on 795 and 1088 cm–1, respectively. The bands at 957 and 575 cm–1 were assigned to the stretching vibrations of Si–OH and the Fe–O, respectively. After amino modification by APTES, the characteristic band of the C–H stretching vibration of the propyl group at 2917 and 2849 cm–1 appeared. These results clearly indicated that amine was successfully modified.

Characterization of Aptamer-Functionalized Fe3O4@SiO2 Nanocomposites

The construction of the immobilizing aptamer (Fe3O4@SiO2–NH2/aptamer) was characterized by UV–vis spectra. It can be seen from Figure A that Fe3O4@SiO2–NH2 NPs exhibited no obvious characteristic absorption peaks in UV–vis region (curve b). However, pure aptamer appeared at an absorbance peak at 265 nm (curve a). After incubation of the aptamer with Fe3O4@SiO2–NH2 nanostructures, an obvious absorption peak appeared at 263 nm. These results demonstrated that aptamers have been successfully immobilized on the Fe3O4@SiO2–NH2 nanostructures.

Figure 2

Characterization of aptamer-functionalized Fe3O4@SiO2–NH2 nanostructures (A) UV–vis spectra of aptamer (a), Fe3O4@SiO2–NH2 NPs (b), and Fe3O4@SiO2–NH2/aptamer conjugate (c) (B) zeta (ζ) potential results of Fe3O4@SiO2–NH2 NPs (a), aptamer (b), and Fe3O4@SiO2–NH2/aptamer conjugate (c). Error bar = standard deviation (n = 3). (C) Confocal fluorescence image of Fe3O4@SiO2–NH2/FITC labeled-aptamer.

Figure B depicts the variations of zeta potential of aptamer coupled with Fe3O4@SiO2–NH2 NPs. The average zeta potentials of the Fe3O4@SiO2–NH2, aptamer, and Fe3O4@SiO2–NH2/aptamer were measured to be +53.3, −19.7, and +28.8 mV, respectively. The prepared Fe3O4@SiO2–NH2 microspheres are positively charged because of the presence of the high amine content in the shells. Upon the aptamer functionalized on the surface of the microspheres, the results have shown a decrease in the value of zeta potential from 53.3 to 28.8 mV. This apparently reveals that Fe3O4@SiO2–NH2/aptamer conjugation carries less positive charge owing to the attachment of negatively charged ssDNA strands in comparison with normal Fe3O4@SiO2–NH2 NPs.

Meanwhile, confocal fluorescence image was also obtained to demonstrate the conjugation of aptamer to Fe3O4@SiO2 NPs surface. Upon the excitation at 488 nm, the fluorescein isothiocyanate (FITC)-labeled aptamer was employed to conjugate with the NPs to allow them to generate green fluorescence. As shown in Figure C, the bright-green fluorescence dots were observed. These results indicated that the aptamer has successfully conjugated with the Fe3O4@SiO2 nanocomposites.

Characterization of Pd@Au NPs and Pd@Au-Con A Bioconjugates

It can be seen from Figure A–C that the TEM image of the prepared Pd@Au NPs showed well-defined dendritic nanocomposites with a solid core and highly branched subunits and were homogeneous in morphology and size. The SEM image (Figure D) of the NPs showed that Pd@Au NPs had a kind of spherical nanostructures and uniformed dispersion. A magnified SEM image (the inset of Figure D) showed that the NPs form core–shell structure clearly. To further characterize Pd@Au NPs, energy-dispersive spectroscopy (EDS) of Pd@Au NPs is shown in Figure E, which demonstrated that the NPs were composed of Pd and Au elements, demonstrating the formation of a Pd@Au nanostructure. Figure F showed the X-ray diffraction (XRD) of Pd@Au NPs, which was used to investigate the crystal structure of samples. As shown in Figure F, four diffraction peaks appearing at 39.21, 45.59, 66.33, and 79.78 were fairly matched with Au (reference code: 00-004-0784) and Pd (reference code: 00-005-0681), which corresponded to the (111), (200), (220), and (311) planes of the face-centered cubic Pd@Au NPs, respectively.[44]

Figure 3

(A) Low-magnification and (B,C) high-magnification TEM image of Pd@Au NPs. (D) SEM image of Pd@Au NPs; Inset: the magnified SEM image. (E) EDS image and (F) XRD pattern of the Pd@Au NPs.

The UV–vis absorption spectrum is shown in Figure S3. The curve a represented that the Pd@Au NPs had no apparent absorption. During the conjugation of Con A, an adsorption peak at 280 nm (curve c), which belonged to the characteristic absorption of Con A (curve b), appeared to demonstrate the successful synthesization of the Pd@Au-Con A bio-composites.

Characterization of Pd@Au NP-Modified Electrode

Before evaluating Pd@Au NPs as labels for cytosensor, we explored the performance of Pd@Au NPs toward the detection of H2O2 as a consequence of the sensitivity of the cytosensor that comes from Pd@Au NPs toward the reduction of H2O2. To compare the results, three types of NP-modified electrodes (Pd, Au, and Pd@Au NPs) were monitored in the same concentration of H2O2. As shown in Figure S4C, the largest reduction current was measured with the Pd@Au NP-modified electrode. Meanwhile, the reduction current of the Pd@Au NPs electrode is even higher than the reduction current of the Pd NP (Figure S4A) and Au NP (Figure S4B) electrodes together, demonstrating that the synergetic effect existed in the Pd@Au NPs. It manifests better catalytic performance of Pd@Au NPs compared with that of individual Pd NPs and Au NPs of the current study.

To further investigate the catalytic performance of the obtained NPs upon H2O2 reduction, cyclic voltammetry (CV) was performed on the Pd@Au NPs at different scan rates in the presence of 1 mM H2O2 in deoxygenated PBS. As depicted in Figure S5, the reduction peak potentials of H2O2 at about −0.4 V gradually shifts to the negative position and the reduction peak currents enhance rapidly with the increase of the scanning rate. In addition, the cathodic currents at −0.4 V were linearly related to the square-root of the scan speed (inset of Figure S5), manifesting that the electrochemical process toward H2O2 of the Pd@Au/GCE is a diffusion-confined process with an irreversible H2O2 electroreduction.

Electrocatalytic Behaviors of Pd@Au NP-Con A as a Label

Because the high sensitivity of Pd@Au NPs toward the detection of H2O2 has been demonstrated; the cytosensor for detecting 1 × 104 cells mL–1 MCF-7 cells using Pd@Au-Con A as labels were established and characterized (Figure S6). The sensitivity of the enzyme-free cytosensors was influenced directly by the catalytic property of the signal labels. The electrocatalytic performances of different signal labels including Au NPs-Con A (a), Pd NPs-Con A (b), and Pd@Au NPs-Con A (c) were investigated to detect MCF-7 cells through the reduction of H2O2. As expected, the catalytic currents are the highest for Pd@Au NPs-Con A (Figure S6, curve c), whereas the currents are much weaker signals detected for Au NPs-Con A (Figure S6, curve a) and Pd NPs-Con A (Figure S6, curve b), respectively. The inset is the response of the cytosensors to 400 cells mL–1 MCF-7 cells using Pd@Au-Con A as a label.

Electrochemical Investigation of the Cytosensor

CV is used to certify each step in the process of the cytosensor, which was carried out in 5 mM Fe(CN)64–/3– containing 0.1 M KCl. As depicted in Figure A, for the bare GCE, a pair of reversible redox peaks were measured (curve a) when Fe3O4@SiO2–NH2/aptamer was coupled onto the above electrode; the peak current obviously kept falling and the gap between redox peak potentials became larger (curve b) as a result of the poor conductivity of aptamer. Subsequently, Fe3O4@SiO2–NH2/aptamer was blocked with BSA. The peak current was further declined after the following to capture the target cells (curve d). This decreasing response may be because of the electronically inert character of BSA and cell. Finally, the above biocomposite was incubated with the Pd@Au-Con A. The peak currents had a slight decrease (curve e) as a result of the bad conductivity of the blocking behaviors of the lectins.

Figure 4

CVs (A) and electrochemical impedance spectra (EIS) (B) of (a) bare GCE, (b) aptamer/Fe3O4@SiO2–NH2/GCE, (c) BSA/aptamer/Fe3O4@SiO2–NH2/GCE, (d) MCF-7 cell/BSA/aptamer/Fe3O4@SiO2–NH2/GCE, and (e) Pd@Au-Con A/MCF-7 cell/BSA/aptamer/Fe3O4@SiO2–NH2/GCE in a solution of 0.1 M KCl solution containing 5 mM [Fe(CN)6]4–/3–, respectively. Scan rate: 100 mV s–1. Inset: the equivalent circuit applied to fit the impedance data. Rs, solution resistance; CPE, constant phase-angle element; Ret, electron-transfer resistance; and Zw, Warburg diffusion resistance.

The EIS was also conducted to confirm the assembly processes of cytosensor construction step by step. The Nyquist plot consists of a straight linear part and a semicircular part. The electron-transfer kinetics of the redox probe at the electrode surface was regulated by the electron-transfer resistance (Ret), which is equal to the semicircle diameter.[45] As shown in Figure B, curve a represented bare GCE and it was nearly a straight line. After sequential modification of Fe3O4@SiO2–NH2/aptamer (curve b), blocking of BSA (curve c), capturing of cells (curve d), and coupling with Pd@Au-Con A nanoprobes (curve e), the resistance increased successively. The results were consistent with CV. In conclusion, the cytosensor was established successfully.

Optimization of the Experimental Paraments

To obtain the best response for the detection of cancer cells, the analytical parameters of the experiment including the concentration of aptamer, the amount of probe, and incubation time for target cells and signal probe have been optimized. With the increase of aptamer concentration, the current increased obviously and then kept a constant level at concentrations above 20 μM (Figure S7A), demonstrating that the aptamer on the surface of Fe3O4@SiO2 NPs was enough to capture the cells. With a prolonging cell incubation time, the current increased step by step and reached an identical value after 90 min (Figure S7B), which indicated the saturated capture of MCF-7 cells at the BSA/aptamer/Fe3O4@SiO2–NH2/GCE surface. Moreover, the current was also connected with the volume of nanoprobe solution and trended the largest value at 60 μL (Figure S7B). The current value was rapidly increased as the incubation time increased and reached a maximum for 60 min, which suggested the saturation identified between Con A and glycan (Figure S7D). Therefore, the best parameters to detect cells were 20 μM for aptamer, 60 μL for volume of signal probe, and 90 and 60 min for incubation time of cell and probe, respectively.

Analysis and Detection of the Cytosensor

Upon the best detection parameters, the cytosensor was conducted by using a chronoamperometric technique in deoxygenated PBS. As displayed in Figure A, the catalytic currents increased obviously with the increase of cell amounts, indicating that the MCF-7 cells are attached onto the interface of the biosensor. The H2O2 catalytic response was connected with the concentrations of captured cells. The catalytic current value has a linear relationship with the logarithmic value of cell concentration ranging from 100 to 1 × 106 cells mL–1 with the limit of detection (LOD) down to 30 cells mL–1 (Figure B). The regression equation was expressed as: I (μA) = −17.9 × log C cell + 19.97, where I is the catalytic current of H2O2. In view of the volume of MCF-7 cell suspension being 100 μL, the established method could realize the LOD down to three cells, which was superior to those previous reports and is listed in Table . It manifested that the developed electrochemical biosensor to detect MCF-7 cells exhibited a wide linear range and excellent sensitivity. The high sensitivity could be attributed to the combination of the signal enhancement by the Pd@Au-Con A nanoprobe catalyst of the reduction of H2O2; the advantage of the large surface of Fe3O4@SiO2 NPs improves the ability to capture cells and enhances the sensitivity of the cytosensor.

Figure 5

(A) Chronoamperometric responses of the cytosensor to different concentrations of MCF-7 cells in PBS solution from a to f: 100, 400, 1 × 103, 1 × 104, 1 × 105, and 1 × 106 cells mL–1. (B) Linear relationship between the current and the logarithm of the MCF-7 cell concentration. Three measurements were conducted for each data point.

Table 1

Comparison of the Developed Cytosensor with Other Nanomaterial-Based Cytosensors

sensor fabricationa	cells	methodb	linear range (cell mL–1)	LODc (cell mL–1)	refs
FA–PEI–CNT	Hela	EIS	2.4 × 102 to 2.4 × 105	90	(46)
AuNPs–gelatin	HL-60	EIS	1 × 104 to 1 × 107	5 × 103	(47)
SiO2@QDs–ConA	HL-60	ASV	1 × 103 to 1 × 107	1 × 103	(48)
Ru–DNA–AuNPs	Ramons	ECL	5 × 102 to 1 × 105	300	(49)
CNS–APBA	MCF-7	DPV	5 × 102 to 1 × 106	25	(50)
CNTs@PDA–FA	Hela	EIS	5 × 102 to 5 × 106	500	(51)
MWCNTs	HL-60	DPV	2.7 × 102 to 2.7 × 107	90	(52)
M-Pd@Au-ConA	MCF-7	amperometry	1 × 102 to 1 × 106	30	this work

FA, folate; PEI, polyetherimide; CNTs, carbon nanotubes; AuNPs, gold nanoparticles; QDs, quantum dots; ConA, concanavalin A; CNS, carbon nanospheres; APBA, 3-aminophenylboronic acid; PDA, polydopamine; and Pd@Au, porous core–shell palladium gold nanoparticles.

EIS, electrochemical impedance spectra; ASV, anodic stripping voltammetry; ECL, electrochemiluminescence; and DPV, differential pulse voltammetry.

LOD, limit of detection.

Specificity, Reproducibility, and Stability of the Cytosensor

The specificity of the established cytosensor was assessed through the chronoamperometric method. The current response signals of the cytosensor were compared with the EpCAM low and negative tumor cells, containing the MCF-10A, MB-MDA-231 cell, and HEK-293T.[53] According to the prediction, if the cell membrane lacks EpCAM, it will inhibit the SYLC3-aptamer-coupled sensor surface. When the SYLC3-aptamer-coupled sensor was incubated with the control cell of equal concentration 1 × 104 cells mL–1, ignorable or minimum responses were observed, demonstrating that the proposed sensor can effectively recognize the MCF-7 cells with good specificity and binding affinity from the mixing cells (Figure S8A). To evaluate the possibility of the proposed method to detect MCF-7 cells in multiple cell suspensions, the same method was conducted in mixed cell solution at same concentrations of MCF-7, MCF-10A, MB-MDA-231, and HEK-293T cells. As shown in the Figure S8A, even though the current response in mixed cell suspensions revealed a slight deviation, it was nearly same as the MCF-7 cell alone, indicating that the established sensor is feasible to detect MCF-7 cell in complicated mixtures. The possibility of the outstanding selectivity is the SYLC3–aptamer possessing a strong binding ability to EpCAM-high expression of MCF-7 cells, whereas not to the EpCAM negative cells (HEK-293T), and EpCAM-low expression of cancer cells (MCF-10A and MB-MDA-231).

To investigate the reproducibility of the cytosensor, five measurements were investigated identically in PBS containing the same concentration (1 × 104 cells mL–1) of MCF-7 cells (Figure S8B). The relative standard deviation was 5.8%, indicating that the presented cytosensor exhibited good reproducibility.

Stability plays an important role in the application of cytosensor. The stability was explored at 1 × 106 cells mL–1 of MCF-7 cells and each reading denotes the mean value of five measurements. It can be seen from Figure S8, the current response of the established cytosensor decreased to about 89.3% of its original value after 7 days. These negligible decreases of the current responses demonstrate an excellent stability of the cytosensor.

In conclusion, the proposed electrochemical cytosensor possessed a prominent level of selectivity, reproducibility, and stability, which manifest that the prepared cytosensor was suitable for quantitative determination of CTCs.

Dynamic Evaluation N-Glycan Expression on the Cell Surface

The established cytosensor was applied to evaluate the dynamic change of N-glycan expression on the cell surface upon treatment of drugs. Primarily, TM and BG were selected as the inhibitors of N-glycan and O-glycan expression on cell surface under the treatment for a successive 48 h (Figure A,B). Compared with control experiment, a gradual decreasing current signal can be measured upon the TM-treated cells. The chronoamperometric current signal alteration was 41.4% after 48 h of treatment. It is because of the fact that the TM restrained the first step in the biosynthesis of the N-glycosylation; therefore, the Con A hardly conjugates with the target cells effectively, which dramatically suppressed the capability for cell capture.[54] To confirm the observed change, confocal scanning microscopy was conducted to assay the treating MCF-7 cells in the stain with FITC-combined Con A (Figure S9). The cells showed weaker fluorescence intensity upon the incubation with TM for 48 h, and the change of fluorescence was calculated to be 38.2%. However, the amount of BG was much higher than that of TM, nearly no change of chronoamperometric current of BG-treated cells was observed compared to untreated cells. It is because that BG could block the O-glycosylation but does not interfere N-glycosylation.[55] Upon different inhibitors between TM and BG, the electrochemical signals confirm that most of the N-glycans on the cell surface could be identified and assessed by the proposed cytosensor.

Figure 6

(A) N-Glycan expression of MCF-7 cell surface without or with treatment of TM (10 μg mL–1), BG (500 μM), and PNGase F (2000 NEB units) for 48 h. (B) N-glycan expression on the MCF-7 cells surface treated: (a) without and (b) with TM (10 μg mL–1) at different times. The current was obtained at −0.4 V (vs SCE) in PBS (0.01 M, pH 7.4) containing 5 mM H2O2. CV scan rate: 100 mV s–1. The MCF-7 cell concentration: 1 × 105 cells mL–1.

Meanwhile, PNGase F, as an amidase, was applied to cleave N-linked proteins.[56] It can be seen from Figure A, the chronoamperometric signal of PNGase F incubated cells dramatically decreased to 39.5% by contrast with untreated cells. Moreover, the decreased level was indicated by flow cytometry (Figure S10), and the average value diminished to 43.1% after enzymatic release. Therefore, the proposed electrochemical method not only provided direct and sensitive in situ profile for determining the dynamic change of N-glycans expression influenced by inhibitors but also could roughly screen the inhibitors of N-glycan.

Conclusions

Herein, a novel nonenzymatic amperometric nanobiosensor has been fabricated for the ultrasensitive determination of CTCs and synchronous evaluation of N-glycan dynamical expression on CTC surface. With the assistance of Fe3O4@SiO2 nanocarriers, SYL3C aptamer and Pd@Au-Con A nanoprobes, and Fe3O4@SiO2/aptamer/cell/nanoprobes sandwich-type architecture were established on a GCE. This cytosensor provides a wide detection linear ranging from 100 to 1 × 106 cells mL–1 with a detection limit as low as 30 cells mL–1. In contrast with the previously reported cytosensors, the biosensor exhibits excellent capability for cancer cell detection with high sensitivity and good repeatability. Moreover, the developed cytosensor could effectively monitor the variation of cell surface glycans interfered by the external treatment. According to these excellent new properties of our cytosensor, it may have a great potential for clinical applications in the early diagnosis of cancer as well as contributing to reveal N-glycan-related biological processes and diseases.

Experimental Section

Cell Culture and Treatment

The MCF-7, MCF-10A, HEK-293T, and MB-MDA-231 cell lines were obtained from KeyGen Biotech. Co. Ltd. (Nanjing, China). The cells were grown in Roswell Park Memorial Institute 1640 medium (high glucose, KeyGen Biotech Corp., Ltd., Nanjing) added with 10% fetal calf serum (KeyGen Biotech Corp., Ltd., Nanjing) in a humidified incubator (5% CO2, 37 °C). At the logarithmic growth phase, the cells were collected by “digestion–centrifugation” procedure. Then, the cells were purified with sterile PBS twice. In the external stimuli study, TM-, BG-, and PNGase F-treated MCF-7 cells were obtained in a culture medium by adding 10 μg mL–1 (approximately 12 μM) TM, 500 μM BG, and PNGase F (2000 NEB units) by incubating 1 × 105 cells mL–1, respectively.

Synthesis of Aptamer-Functionalized Aminated Fe3O4@SiO2 NPs

The Fe3O4 NPs were prepared through a one-pot reaction by a modification, as described in the previous protocol.[57] To prevent aggregation of Fe3O4 and improve the availability of combination with aptamer, it was modified by silanization and amination before combination. These processes were conducted in terms of the previous method without any modification.[58]

The aptamer-functionalized aminated Fe3O4@SiO2 NPs were synthesized through cross-linking by GA. Briefly, 5 mg of aminated Fe3O4@SiO2 NPs was suspended in 7.5 mL of PBS (0.01 M, pH 7.4). Subsequently, GA solution (2.5 mL, 2.5 wt %) was added to the above solutions. The suspension was mechanically agitated for 6 h at 25 °C. Then, the obtained product was purified with PBS to get rid of the redundant GA. The purified sample was resuspended in 10 mL of PBS. Then, 1 mL of the as-prepared aminated Fe3O4@SiO2 NPs (0.5 mg mL–1) was mixed with the amine-modified aptamer (50 μL, 20 μM) and incubated for 12 h with gentle shaking. Then, the prepared composite was purified three times with PBS and separated by an external magnetic force. Next, the nonspecific binding sites were blocked by the addition of BSA solution (100 μL, 1 wt %) to the above solution for 1 h. The prepared Fe3O4@SiO2–NH2/aptamer/BSA conjugate was purified three times with PBS under the external magnet and ultimately redispersed in PBS (0.5 mL) and stored in the refrigerator for the next study.

Synthetization of Pd@Au-Con A Nanoprobes

The porous Pd@Au NPs were synthesized by following a reported method[36] with a corresponding modification. In brief, 2 mL of Na2PdCl4 solution (10 mM), 0.5 mL of HAuCl4 aqueous solution (10 mM), and HDPC (0.1 g) were diluted in distilled water (25 mL) under stirring. Subsequently, AA solution (1.5 mL, 0.1 M) was added into the mixture rapidly. In addition, the above solution was maintained at 35 °C for 3 h. After that, the black products were purified under “centrifugation-wash-centrifugation” steps with distilled water and ethanol several times. Moreover, the resulting samples were dried for 24 h at 60 °C. Finally, the obtained porous Pd@Au NPs (0.02 g) were resuspended in 4 mL of ultrapure water.

The Pd@Au-Con A bioconjugates were prepared by 500 μL of the as-obtained Pd@Au NPs solution, and 100 μL of Con A solution (2 mM) was mixed into 10.0 mL of ultrapure water under slight mechanical stirring overnight at 4 °C. Subsequently, the supernatant was discarded under centrifugation, and the precipitate was purified with ultrapure water several times and resuspended with 1.0 mL of PBS containing 1 mM Mn2+ and Ca2+ to remain the binding affinity of Con A and stored at 4 °C.

Fabrication of the Cytosensor

Fe3O4@SiO2–NH2/aptamer/BSA conjugate suspension (100 μL; 1.0 mg mL–1) was mixed with 100 μL of MCF-7 cell suspension at various concentrations and incubated at 37 °C for 90 min. Under the external magnet force, the mixture was washed with a binding buffer to get rid of the free cells. Subsequently, the Fe3O4@SiO2–NH2/aptamer/BSA and MCF-7 cell complex (Fe3O4@SiO2–NH2/aptamer/BSA/MCF-7 cells) was further mixed with 60 μL of signal probe, incubated at 37 °C for 60 min to combine with the cells by specific identification, and purified with PBS several times to remove the excess Pd@Au-Con A complexes after magnetic separation. In the end, the cytosensor was resuspended in 200 μL of PBS and stored at 4 °C for usage.

Electrochemical Assay of the Sandwich-type Cytosensor

Generally, the GCE was polished carefully by 0.3 and 0.05 μm of Al2O3 powders before surface modification. After washing thoroughly, the pretreated GCE was coated with 6 μL of sandwich-architecture cytosensors with various concentrations of cells. For amperometric determination of the cytosensor, −0.4 V was chosen as the detection potential. After the current was stable under stirring, 10 μL of (5 M) H2O2 was added into PBS under gentle stirring, and the current change was recorded. The CV and EIS were recorded in 5 mM [Fe (CN)6]3–/4– containing 0.1 M KCl.