Dual-Mode Sensing Platform for Cancer Antigen 15-3 Determination Based on a Silica Nanochannel Array Using Electrochemiluminescence and Electrochemistry

An electrochemiluminescence-electrochemistry (ECL-EC) dual-mode sensing platform based on a vertically-ordered mesoporous silica films (VMSF) modified electrode was designed here for the sensitive and selective determination of cancer antigen 15-3 (CA 15-3), a specific biomarker of breast cancer. VMSF was assembled through a rapid electrochemically assisted self-assembly (EASA) method and plays a crucial role in signal amplification via a strong electrostatic interaction with the positively charged bifunctional probe Ru(bpy)32+. To construct the biorecognition interface, epoxy functional silane was linked to the surface of VMSF for further covalent immobilization of the antibody. As a benefit from the specific combination of antigen and antibody, a non-conductive immunocomplex layer was formed in the presence of CA 15-3, leading to the hinderance of the mass and electron transfer of the probes. Based on this strategy, the dual-mode determination of CA 15-3 ranging from 0.1 mU/mL to 100 mU/mL with a LOD of 9 μU/mL for ECL mode, and 10 mU/mL to 200 U/mL with a LOD of 5.4 mU/mL for EC mode, was achieved. The proposed immunosensor was successfully employed for the detection of CA 15-3 in human serum without tedious pretreatment.


Introduction
Cancer has become the second major cause of death worldwide [1]. According to the latest statistics from the International Agency for Research on Cancer (IARC), breast cancer occurs with the highest incidence and is also the major cause of cancer-related mortality in females [2], indicating that it is the most common and high-risk cancer globally. The diagnosis of breast cancer mainly relies on clinical screening or imaging analysis that often requires invasive means to obtain tissue samples and may bring sufferings to the patients [3]. The establishment of a noninvasive/minimally invasive, non-radioactive, simple, and rapid technique for early diagnosis remains a serious challenge and an unmet demand. Tumor markers, a kind of biological macromolecules existing in accessible body fluids (e.g., blood, serum, and urine) [4], provide a new standard for cancer diagnosis, health monitoring and personalized treatment. Cancer antigen 15-3 (CA 15-3) is an acidic glycoprotein with a molecular weight of 300-450 kDa that is often found in those who suffer from breast cancer [5,6]. The serum content of CA 15-3 in healthy people is usually less than 30 U/mL [7]. Abnormally elevated serum CA 15-3 levels indicate a higher risk of breast cancer. Simple but sensitive identification and quantitation of serum CA 15-3 level is critical for cancer screening, determination of disease progression, and treatment effect assessment.
Enzyme-linked immunosorbent assay (ELISA) [8], fluorescent immunoassay (FIA) [9], and radioimmunoassay (RIA) [10] have been reported for the quantitation of CA 15-3, but cumbersome instruments, trained operators and time-consuming preprocesses are always required. The electrochemical (EC) sensing technique has proven itself to be a powerful tool because of the fast response, simple operation, and ease of integration [11,12] compared with the traditional immunoassay, despite the hidden risk of inaccurate test results. To guarantee the accuracy and reliability of the diagnosis, dual-mode bioassays with different mechanisms and independent signal conversion have been developed. To date, a series of dual-mode biosensors which incorporate colorimetric-electrochemiluminescence (ECL) [13], ECL-electrochemical impedance spectroscopy (EIS) [14], fluorescence-colorimetry [15], and ECL-EC [16,17] have been reported. Among them, the ECL-EC dual-mode that combines the wider linear range and zero background signal of the former and the fast and easy operation characteristics of the latter with highly sensitivity, has attracted much attention. Generally, two different probes for ECL and EC are involved in the reported dual-mode detection strategy, which requires more laborious procedures. Tris(2,2 -bipyridyl) ruthenium (Ru(bpy) 3 2+ ), as an excellent bifunctional probe with both outstanding EC and ECL response, is a good candidate for constructing an ECL-EC dual-mode sensor, not only simplifying the manual work, but also reducing the analysis error effectively [17,18].
In the recent three decades, functional mesoporous materials possessing abundant unique mesostructures have exhibited tremendous potential in the fields of adsorption, separation, catalysis, and sensing [19][20][21]. Among them, the vertically-ordered mesoporous silica films (VMSF) made of numerous perpendicularly aligned nanochannels with a ultrasmall and uniform diameter of 2~3 nm has attracted much attention [22,23]. The deprotonation of plentiful silanols with pK a of~2 on the VMSF surface endows its negatively charged property [24,25]. All these characteristics endow VMSF with a good capability of prominent charge-based permselectivity, high molecular permeability, and outstanding antifouling ability [26][27][28]. Therefore, VMSF has been widely used in the construction of sensing platforms to serve certain functions (e.g., preconcentration through electrostatic interaction [29], hydrogen bonding [30] or hydrophobic extraction [31], specific recognition by tailoring recognitive molecules [32], and antifouling ability through size exclusion effect of nanopores [33]). As a monolayer of modification material with a nanoscale thickness, diverse VMSF-based sensors have been developed for the determination of biomolecules (DNA, antibodies, and antigens) [17,34], pharmaceutical molecules [35,36], small organic pollutants [30,37], and metal ions [38,39] in biofluids (blood, serum, urine, and sweat) and environmental samples. Su's group observed a two-orders-of-magnitude ECL intensity enhancement when equipping indium tin oxide (ITO) electrode with a layer of VMSF due to the strong electrostatic attraction between the negatively charged nanochannels and the cationic ECL luminophore Ru(bpy) 3 2+ and obtained a high ECL intensity even with a very low concentration [40]. On the other hand, the existence of silanols with excellent covalent reactivity towards silane coupling reagents provides the precondition to construct a recognition interface by introducing certain specific molecules to design a gated sensing scheme [41,42]. VMSF has become a fascinating material in the construction of sensing interfaces, demonstrating a great potential for direct and sensitive detection of targets in complex samples without cumbersome pretreatment processes including filtration, separation, and pre-enrichment.
In this work, an ECL-EC dual-mode immunosensor constructed on an ITO substrate composed of a layer of mesoporous silica membrane as the signal amplifier, and a monomolecular layer of anti-CA 15-3 antibody as the recognition element was designed for the sensitive and selective detection of the breast cancer biomarker CA 15-3. Compared with medical imaging and pathological biopsy applied to the clinical diagnosis of breast cancer, this method offers more hope for the early detection of breast cancer rapidly and conveniently. As shown in Scheme 1, VMSF was first assembled on ITO through the electrochemical assisted self-assembly (EASA) method. The construction of the immune recognition interface for further identifying and capturing targets specifically was achieved by immobilizing the antibodies with the assistance of (3-glycidyloxypropyl)trimethoxysilane (GPTMS) linked to the outer surface of VMSF beforehand. A layer of immunocomplex with poor conductivity and large size would form in the presence of the CA 15-3 antigen and hamper the mass transfer and electron transfer of the bifunctional probe Ru(bpy) 3 2+ , leading to decreased ECL-EC signal and the final signal-off detection of CA 15-3. The benefits of the immunosensor lie in the easy fabrication of the modification material which serves as both a signal amplifier to enrich Ru(bpy) 3 2+ and an anti-fouling filter, as well as in the generalizability of the detection strategy which is suitable for the quantitation of various targets by simply changing the immobilized bio-receptors. this method offers more hope for the early detection of breast cancer rapidly and conveniently. As shown in Scheme 1, VMSF was first assembled on ITO through the electrochemical assisted self-assembly (EASA) method. The construction of the immune recognition interface for further identifying and capturing targets specifically was achieved by immobilizing the antibodies with the assistance of (3-glycidyloxypropyl)trimethoxysilane (GPTMS) linked to the outer surface of VMSF beforehand. A layer of immunocomplex with poor conductivity and large size would form in the presence of the CA 15-3 antigen and hamper the mass transfer and electron transfer of the bifunctional probe Ru(bpy)3 2+ , leading to decreased ECL-EC signal and the final signal-off detection of CA 15-3. The benefits of the immunosensor lie in the easy fabrication of the modification material which serves as both a signal amplifier to enrich Ru(bpy)3 2+ and an anti-fouling filter, as well as in the generalizability of the detection strategy which is suitable for the quantitation of various targets by simply changing the immobilized bio-receptors. Scheme 1. Illustration of the preparation of the immunosensor and detection of CA 15-3.  were purchased from Sigma-Aldrich. Ethanol (99.8%), hydrochloric acid (HCl, 38%), and acetone were from Hangzhou Shuanglin Chemical Reagent Co. Ltd. Phosphate buffered saline (PBS, 0.01 M, pH = 7.4) was obtained by mixing NaH 2 PO 4 and Na 2 HPO 4 in a certain ratio. All chemicals and reagents were used as received without any further purification. All the aqueous solutions used here were prepared using ultrapure water (18.2 MΩ cm) from Milli-Q Systems (Millipore Inc., Massachusetts, America). Serum of healthy men was obtained from the Center of Disease Control and Prevention (Hangzhou, China) for real sample analysis.

Apparatus and Electrodes
An MPI-E II ECL analytical system (Xi'an Remex Analytical Instrument Ltd., Xi'an, China) was used to perform ECL measurements. EC measurements, including cyclic voltammetry (CV), differential pulse voltammetry (DPV), and electrochemical impedance spectroscopy (EIS) were performed on an Autolab electrochemical workstation (PGSTAT302N, Metrohm, Switzerland). A traditional three-electrode system was adopted. Bare or modified ITO, platinum wire or platinum plate, and Ag/AgCl electrode (saturated with KCl solution) were used as the working electrode, counter electrode and reference electrode, respectively. Zhuhai Kaivo Optoelectronic Technology provided ITO glasses (<17 Ω/square, thickness: 100 ± 20 nm), which were cleaned ultrasonically with 1 M NaOH, acetone, ethanol and deionized water, respectively. The geometric surface area of the electrode was fixed to 1 cm × 0.5 cm.

Preparation of VMSF/ITO Electrode
According to the reported electrochemically assisted self-assembly (EASA) method, an easy and rapid modification of VMSF on the ITO electrode was achieved [43]. A sol for film electrodeposition was prepared as follows: first mix 20 mL of ethanol and 20 mL of NaNO 3 (0.1 M, adjusted to pH = 2.6 by HCl) together, then add 1.585 g of CTAB and 3050 µL of TEOS. After their complete dissolution, the solution was stirred for 2.5 h at room temperature for hydrolyzation. The assembly of VMSF was conducted under a galvanostatic condition (j = −0.70 mA/cm 2 ) for 10 s after immersing the ITO in the hydrolyzed sol, using an Ag/AgCl (saturated KCl) as reference electrode and a platinum plate (2 cm × 4 cm) as counter electrode. The as-prepared electrode was rapidly removed from the sol and washed with copious deionized water, dried with N 2 stream, and further treated overnight at 120 • C. The obtained electrodes were denoted SM@VMSF/ITO, which retained surfactant micelles (SM) templates in the nanochannels. The templates can be conveniently removed using 0.1 M HCl-ethanol by stirring for 5 min and the obtained electrode modified with open nanochannels was termed VMSF/ITO.

Characterization of VMSF/ITO Electrode
The morphology of VMSF was characterized with a JEM-2100 TEM with an acceleration voltage of 100 kV. The TEM specimen was prepared by dispersing the VMSF peeled off from the VMSF/ITO in ethanol through ultrasonication, and finally dropped onto the copper grids for investigation. The thickness of VMSF was characterized using a SU8010 SEM with an acceleration voltage of 5 kV. Before observation, a fresh cross section was obtained by breaking up the VMSF/ITO and spraying it with gold.
The integrity and permeability of VMSF were investigated with the CV technique using Fe(CN) 6 3− and Ru(NH 3 ) 6 3+ , two types of standard redox probes with different charges.

Fabrication of Immunosensor towards CA 15-3
For the immunosensor fabrication, the construction of a biorecognition interface is of great significance. To immobilize the antibody on the outer surface of the VMSF/ITO electrode, the bifunctional reagent GPTMS with both epoxy and silyl groups was selected as the crosslinker [34]. Typically, the SM@VMSF/ITO was immersed in GPTMS (2.26 mM in ethanol) for 60 min, then the GPTMS was covalently linked to the outer surface instead of the inner channels of the VMSF through silane reaction. The electrode was then treated with 0.1 M HCl-ethanol to remove the SM templates. The electrode modified with epoxy-functionalized nanochannels was named O-VMSF/ITO. Anti-CA 15-3 antibodies (anti-CA 15-3) were further immobilized by covering the O-VMSF/ITO with 50 µL of anti-CA 15-3 solution (10 µg/mL in PBS) for 90 min at 37 • C via the ring opening reaction between active epoxy groups on the VMSF and amino residues of the anti-CA 15-3. After rinsing with PBS, it was incubated with s BSA solution (1%, w/w) for 90 min to block the nonspecific binding sites. The immunosensor termed as anti-CA 15-3/O-VMSF/ITO was obtained and was stored at 4 • C for further tests. The procedure of the immunosensor fabrication is illustrated in Scheme 1.

Dual-Mode Tests
Before the ECL and EC tests, the immunosensor was incubated with different concentrations of CA 15-3 (50 µL in PBS) for 45 min at 37 • C and was noted as CA 15-3/anti-CA 15-3/O-VMSF/ITO, followed by thoroughly washing with PBS. For ECL measurement, the CA 15-3/anti-CA 15-3/O-VMSF/ITO was placed in PBS containing 10 µM Ru(bpy) 3 2+ and 3 mM TPA. The ECL process was triggered by a continuous CV procedure with the potential range of 0 to 1.25 V at 0.1 V/s, which was recorded simultaneously along with the ECL signals. The voltage of the photomultiplier tube (PMT) was set at 400 V. For the EC measurement, the CA 15-3/anti-CA 15-3/O-VMSF/ITO was immersed in PBS including 10 µM Ru(bpy) 3 2+ and let to stand for 30 min to preconcentration before recording the DPV curves.
Real sample analysis was conducted on human serum from healthy adults using the standard addition method without complex preprocessing. The serum was spiked with a known amount of CA 15-3, followed by 50-times dilution with PBS, and was analyzed by the established methods.

Morphology and Permeability of VMSF/ITO
As expected, the VMSF was assembled on the surface of the ITO electrode due to polycondensation of TEOS under the catalysis of hydroxide ions generated from the electroreduction of protons/water and nitrate ions, where CTAB served as the template, and was proved by TEM and SEM images (Figure 1a,b). VMSF presents a 2D structure of intact film with a thickness of 90 nm and is free of any cracks. A too thin VMSF would lead to an inadequate enrichment of Ru(bpy) 3 2+ , and would finally fail to achieve a high sensitivity. With the prolongation of electrodeposition time, a much thicker VMSF with more silica aggregates byproducts on its surface would be obtained, which would restrict the diffusion of Ru(bpy) 3 2+ species across the film to the underlying electrode surface. According to the many reported works [23,24,[28][29][30], a procedure of electrodeposition for 10 s at j = −0.70 mA/cm 2 was selected here to obtain a VMSF with an appropriate thickness. This not only achieves an effective electrostatic attraction to Ru(bpy) 3 2+ , but also guarantees an efficient and rapid response. The film contains numerous nanochannels that are highly ordered and hexagonally packed and whose diameter is 2~3 nm.
Considering that VMSF plays the role of the modification material that is on the electrolyte/electrode interface, it is crucial to determine how VMSF affects the mass transfer. The permeability is investigated by comparing the electrochemical behaviors of Ru(NH 3 ) 6 3+ and Fe(CN) 6 3−/4− on ITO, SM@VMSF/ITO and VMSF/ITO. As shown in Figure 1c,d, SM@VMSF/ITO exhibited no peak current but only a non-faradic current towards the Ru(NH 3 ) 6 3+ probe and a rather large charge transfer resistance (R ct ) represented by the equivalent diameter of Nyquist curve obtained using the EIS technique, a powerful tool to study interfacial property [44]. This is due to the hydrophobic SM templates inside the nanochannels that block the mass transfer of charged probes (Ru(NH 3 ) 6 3+ and Fe(CN) 6 3−/4− ) and finally hinder the electron exchange. The nanochannels open after the removal of SM, which facilitates the diffusion of particles, and thus VMSF/ITO shows a pair of well-defined redox peaks to Ru(NH 3 ) 6 3+ and a much smaller R ct . In addition, VMSF/ITO exhibits a relatively higher peak current towards Ru(NH 3 ) 6 3+ than that of ITO, which derives from the deprotonation of abundant silanol with a pK a of~2 in the nanochannel walls that can create a strong electrostatic attraction to positively charged species [25]. Considering that VMSF plays the role of the modification material that is on the electrolyte/electrode interface, it is crucial to determine how VMSF affects the mass transfer. The permeability is investigated by comparing the electrochemical behaviors of Ru(NH3)6 3+ and Fe(CN)6 3−/4− on ITO, SM@VMSF/ITO and VMSF/ITO. As shown in Figure  1c,d, SM@VMSF/ITO exhibited no peak current but only a non-faradic current towards the Ru(NH3)6 3+ probe and a rather large charge transfer resistance (Rct) represented by the equivalent diameter of Nyquist curve obtained using the EIS technique, a powerful tool to study interfacial property [44]. This is due to the hydrophobic SM templates inside the nanochannels that block the mass transfer of charged probes (Ru(NH3)6 3+ and Fe(CN)6 3−/4− ) and finally hinder the electron exchange. The nanochannels open after the removal of SM, which facilitates the diffusion of particles, and thus VMSF/ITO shows a pair of well-defined redox peaks to Ru(NH3)6 3+ and a much smaller Rct. In addition, VMSF/ITO exhibits a relatively higher peak current towards Ru(NH3)6 3+ than that of ITO, which derives from the deprotonation of abundant silanol with a pKa of ~2 in the nanochannel walls that can create a strong electrostatic attraction to positively charged species [25].
The excellent capability of VMSF to enrich cations was further verified by Ru(bpy)3 2+ , a bifunctional probe with outstanding ECL and EC activities. The ECL process was triggered by a continuous CV procedure; the mechanism is as follows: The excellent capability of VMSF to enrich cations was further verified by Ru(bpy) 3 2+ , a bifunctional probe with outstanding ECL and EC activities. The ECL process was triggered by a continuous CV procedure; the mechanism is as follows: During the positive scanning of potential, TPA and Ru(bpy) 3 2+ were electrooxidized to TPA · radical and Ru(bpy) 3 3+ . The redox reaction between TPA · radical and Ru(bpy) 3

3+
led to the generation of excited Ru(bpy) 3 2+ * that emitted light (λ max ≈ 620 nm) when turning to the ground state, which was recorded by the photomultiplier. As shown in Figure 2, the ECL and DPV signals produced by 10 µM Ru(bpy) 3 2+ on ITO were very weak while those obtained on VMSF/ITO were remarkable enhanced thanks to the electrostatic preconcentration of the nanochannel layer in the positively charged probe. VMSF/ITO tends to provide an adequate basic signal and exhibits enormous potential in the subsequent immunosensor fabrication.
led to the generation of excited Ru(bpy)3 2+ * that emitted light (λmax ≈ 620 nm) when turning to the ground state, which was recorded by the photomultiplier. As shown in Figure 2, the ECL and DPV signals produced by 10 μM Ru(bpy)3 2+ on ITO were very weak while those obtained on VMSF/ITO were remarkable enhanced thanks to the electrostatic preconcentration of the nanochannel layer in the positively charged probe. VMSF/ITO tends to provide an adequate basic signal and exhibits enormous potential in the subsequent immunosensor fabrication.

Characterization of the Immunosensor Construction
The CV and EIS techniques were employed to monitor the interface features during the bioelectrode's stepwise assembly. As shown in Figure 3a, VMSF/ITO showed two well-defined redox peaks, which were decreased after the modification of GPTMS, indicating the presence of epoxy group entities. The peak currents were further reduced with the immobilization of antibodies and the capture of CA 15-3 antigens, suggesting that an insulating protein layer was formed on the electrode surface, creating an obstacle to both mass and electron transfer. Figure 3b shows the EIS results fitted according to the Randles equivalent circuit. As shown by the inset, the Randles equivalent circuit contained a solution resistance (Rs), a double-layer capacitance (Cdl), Warburg impedance (Zw), and an apparent charge transfer resistance (Rct) that is equal to the equivalent diameter of the semicircle in the high-frequency region [45]. The Rct of VMSF/ITO was the smallest (141 Ω) and

Characterization of the Immunosensor Construction
The CV and EIS techniques were employed to monitor the interface features during the bioelectrode's stepwise assembly. As shown in Figure 3a, VMSF/ITO showed two welldefined redox peaks, which were decreased after the modification of GPTMS, indicating the presence of epoxy group entities. The peak currents were further reduced with the immobilization of antibodies and the capture of CA 15-3 antigens, suggesting that an insulating protein layer was formed on the electrode surface, creating an obstacle to both mass and electron transfer. Figure 3b shows the EIS results fitted according to the Randles equivalent circuit. As shown by the inset, the Randles equivalent circuit contained a solution resistance (R s ), a double-layer capacitance (C dl ), Warburg impedance (Z w ), and an apparent charge transfer resistance (R ct ) that is equal to the equivalent diameter of the semicircle in the high-frequency region [45]. The

Dual-Mode Tests of CA 15-3
The specific recognition and combination between the anti-CA 15-3 antibody and the CA 15-3 antigen formed an antibody-antigen immunocomplex. In analogy with the aforementioned CV and EIS results, Figure 4a shows decreased ECL responses with the stepby-step assembling of the bioelectrode, confirming the feasibility of the ECL analysis. The

Dual-Mode Tests of CA 15-3
The specific recognition and combination between the anti-CA 15-3 antibody and the CA 15-3 antigen formed an antibody-antigen immunocomplex. In analogy with the aforementioned CV and EIS results, Figure 4a shows decreased ECL responses with the step-by-step assembling of the bioelectrode, confirming the feasibility of the ECL analysis. The inset suggested an outstanding stability of the biosensor under continuous CV scanning with a relative standard deviation (RSD) of less than 2%. The I ECL -time curves recorded on bioelectrodes incubated with a series of different amounts of CA 15-3 are shown in Figure 4b. With the increase in CA 15-3, more immunocomplex was formed, which perturbed the mass transfer of Ru(bpy) 3 2+ and restrained the electron transfer at the electrode. A linear relationship for CA 15-3 was obtained from 0.1 mU/mL to 100 U/mL using the equation I ECL = −997.0 logC CA15-3 + 2537 (R 2 = 0.991). LOD is defined by the International Union of Pure and Applied Chemistry (IUPAC) as the concentration (c L ) that can be detected with a reasonable certainty for a given analytical procedure. It can be derived from the smallest measure (x L ), which is calculated with the equation x L = x b1 ± ks b1 , where x b1 is the mean of the blank measures, s b1 is the standard deviation of the blank measures, and k is a numerical factor commonly recommended as 3. Here, + and − are applied to signal-on and signal-off sensors, respectively [46]. Thus, the LOD was calculated to be 9 µU/mL according to the calibration curve equation. Compared with the other reported methods listed in Table 1, this proposed ECL immunosensor exhibits a relatively wider linear range and an exceedingly lower LOD for CA 15-3 determination without elaborate and complex labor during the construction process. Diverse sophisticated nanomaterials and sandwich strategies are widely used in other related works listed in Table 1. They require a complex preparation process, several steps of incubation, and consume more high-cost biological reagents. Our proposed sensor is simple in preparation, does not need multiple incubations for detection, and is much more economic.  In addition, considering the excellent electroactivity of Ru(bpy)3 2+ , CA 15-3 was quantified using the DPV method. Similar to the ECL mode, the feasibility of the EC mode was verified by the gradually reduced DPV response during the bioelectrode fabrication as shown in Figure 5a. Upon the increase in CA 15-3, a group of peak currents which decreased linearly with the logarithm of the antigen concentration from 10 mU/mL to 200 U/mL were obtained (Figure 5b), and the calibration curve was represented in Figure 5c using the equation I = −0.82 logCCA15-3 + 4.60 (R 2 = 0.992). Similar to the ECL mode, the LOD was calculated to be 5.4 mU/mL (S/N = 3). Although the EC method was inferior in LOD compared with the ECL mode, a higher detection upper limit was achieved. Although the LOD of our proposal is not as low as 10 −5 U/mL for the sandwich strategy using Ab2/HRP-MBs and Ab1/AuE, our immunosensor is free of complex labeling operations or several steps of incubation. Dual-mode tests that combine the merits of ECL and EC strategies succeed to broaden the working linear range while achieving an ultrasensitive LOD. The total cost of a single fabricated immunosensor is approximately 0.366 USD, which is much lower in cost compared to ELISA which costs around 1.31 USD for a single sample test. Furthermore, the materials involved here are all commercially available, indicating its potential for industrialization, as well as the possibility of application in microfluidic Point- In addition, considering the excellent electroactivity of Ru(bpy) 3 2+ , CA 15-3 was quantified using the DPV method. Similar to the ECL mode, the feasibility of the EC mode was verified by the gradually reduced DPV response during the bioelectrode fabrication as shown in Figure 5a. Upon the increase in CA 15-3, a group of peak currents which decreased linearly with the logarithm of the antigen concentration from 10 mU/mL to 200 U/mL were obtained (Figure 5b), and the calibration curve was represented in Figure 5c using the equation I = −0.82 logC CA15-3 + 4.60 (R 2 = 0.992). Similar to the ECL mode, the LOD was calculated to be 5.4 mU/mL (S/N = 3). Although the EC method was inferior in LOD compared with the ECL mode, a higher detection upper limit was achieved. Although the LOD of our proposal is not as low as 10 −5 U/mL for the sandwich strategy using Ab 2 /HRP-MBs and Ab 1 /AuE, our immunosensor is free of complex labeling operations or several steps of incubation. Dual-mode tests that combine the merits of ECL and EC strategies succeed to broaden the working linear range while achieving an ultrasensitive LOD. The total cost of a single fabricated immunosensor is approximately 0.366 USD, which is much lower in cost compared to ELISA which costs around 1.31 USD for a single sample test. Furthermore, the materials involved here are all commercially available, indicating its potential for industrialization, as well as the possibility of application in microfluidic Point-of-Care devices [47,48]. However, the immunosensor is susceptible to matrix effects, and a dilution is required before testing bio-samples to guarantee the accuracy in clinical application. In addition, an alkali condition may lead to the hydrolysis of VMSF as the proposed immunosensor is applicative in neutral medium to keep its mechanical stability. The current clinical determination of CA 15-3 is mainly based on a chemiluminescence immunoassay. This method not only involves labeling technology, but also requires two antibodies, as well as several steps of incubation and washing, which is high-cost and time-consuming. Our proposed method involves only one step of incubation, which meets the needs of rapid clinical determination of CA 15-3 at low cost.

Specificity, Selectivity and Stability of the Immunosensor
The proposed immunosensor was expected to be highly specific and selective for CA 15-3 detection due to the antigen-antibody recognition. Therefore, the bioelectrode was tested with different possible clinical cancer biomarkers such as the carcinoembryonic an-

Specificity, Selectivity and Stability of the Immunosensor
The proposed immunosensor was expected to be highly specific and selective for CA 15-3 detection due to the antigen-antibody recognition. Therefore, the bioelectrode was tested with different possible clinical cancer biomarkers such as the carcinoembryonic antigen (CEA) and the prostate specific antigen (PSA) (the concentrations of interfering proteins are both 500 ng/mL). As illustrated in Figure 6, the signal fluctuation caused by individual interfering proteins was less than 7%; an evident decrease in signal was observed when incubated with CA 15-3 or the mixture of all proteins. These results indicated the high specificity and outstanding selectivity of the fabricated immunosensor. The stability of the anti-CA 15-3/O-VMSF/ITO was evaluated by comparing the signal after several days of storage at 4 • C with the initial signal on the first day. Anti-CA 15-3/O-VMSF/ITO retained more than 95% of the initial signal after a 15-day storage at 4 • C, suggesting a satisfactory stability.
osensors 2023, 13, x FOR PEER REVIEW 11 of 14 retained more than 95% of the initial signal after a 15-day storage at 4 °C, suggesting a satisfactory stability.

Real Sample Analysis
To validate the practical applicability of the established method, the biosensor was used to determine the CA 15-3 spiked in healthy human serum samples. A series of known amount of CA 15-3 was added to the samples and then they were diluted with PBS 50 times. After the incubation with anti-CA 15-3/O-VMSF/ITO, the CA 15-3/anti-CA 15-3/O-VMSF/ITO was obtained for ECL and EC testing. The concentration of CA 15-3 could be calculated by substituting the obtained ECL and EC response into the original linear equation. The results are listed in Table 2. Satisfactory recoveries ranging from 95.4% to 104.0% and an RSD below 2.9% were obtained using the ECL-EC dual-mode tests, which implied the potential for clinical application.  . Relative ratio of ECL intensity (a) and DPV current (b) before (I 0 ) and after (I) incubation with buffer (Control), CEA (500 ng/mL), PSA (500 ng/mL), CA 15-3 (25 U/mL), or a protein mixture (25 U/mL CA 15-3 + 500 ng/mL other proteins). The error bars represent the standard deviation of three measurements under the same condition.

Real Sample Analysis
To validate the practical applicability of the established method, the biosensor was used to determine the CA 15-3 spiked in healthy human serum samples. A series of known amount of CA 15-3 was added to the samples and then they were diluted with PBS 50 times. After the incubation with anti-CA 15-3/O-VMSF/ITO, the CA 15-3/anti-CA 15-3/O-VMSF/ITO was obtained for ECL and EC testing. The concentration of CA 15-3 could be calculated by substituting the obtained ECL and EC response into the original linear equation. The results are listed in Table 2. Satisfactory recoveries ranging from 95.4% to 104.0% and an RSD below 2.9% were obtained using the ECL-EC dual-mode tests, which implied the potential for clinical application.

Conclusions
In summary, an ECL-EC dual-mode immunosensor with a simple fabrication strategy was designed for CA 15-3 determination with an efficient biosensing performance. The VMSF equipped on the electrode surface displays a strong electrostatic attraction to the positively charged bifunctional probes, and provides a large surface area with good biocompatibility for the biorecognition of elements binding with the aid of the crosslinker GPTMS. Integrating the merits of both the sensitivity of electrochemiluminescence and the efficiency of electrochemistry, the immunosensor demonstrated a satisfactory performance with a linear working range of more than 7 orders of magnitude and a LOD of 9 µU/mL. Furthermore, VMSF exhibits good resistance to foulants thanks to the ultrasmall diameter of nanochannels. Real sample analysis in human serum was performed, expecting to provide a new method for clinical CA 15-3 diagnosis.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author.

Conflicts of Interest:
The authors declare no conflict of interest.