2.1.

SPR Sensor

A miniaturized SPR sensor (SPRmicro, K-MAC, Daejeon, Korea) based on the Kretschmann type was used and its size was $45\mathrm{mm}\times 140\mathrm{mm}\times 130\mathrm{mm}$. Figure 1 shows a schematic diagram of the SPR sensor, which consists of a BK7 cylindrical prism, a sensor chip, a light source, and a detector array. As shown in this figure, a beam from a 780 nm light emitted diode first penetrates a bandpass filter ($780\mathrm{nm}\pm 5$) and reaches the sensor chip with an incident beam range from 64 to 71.4 deg through the prism. Subsequently, the reflected light is directed to a two-dimensional complementary metal oxide semiconductor (2-D-CMOS). A flowcell with two separate flow channels (a sample channel and a reference channel) excludes external environment variations such as temperature, since the SPR sensor does not have a thermostat. The reflectance from the two sensing channels was measured giving rise to the grayscale image in Fig. 2(a). The dark areas in the grayscale images denote a dip in the SPR reflectance curve because there is almost no reflected light. The upper and lower areas in the grayscale images signify the collected reflectance for both the reference and the sample channels. As the refractive index is changed by adsorbing the biomolecules on the sensor chip surface, the dark position of the sample channel is shifted, as shown in Fig. 2(b). In the intensity detection mode, the incident angle at the steepest slope in the SPR reflectance curve is fixed. The change in the reflectance was measured at this fixed angle, as shown in Fig. 2(c). The analyte signals were obtained by subtracting the reference signal from that of the sample signal. A peristaltic pump was used to flow the analytes or reference solution over the two sensing channels after air bubbles were removed with a degasser.

Fig. 1

Schematic diagram of a miniaturized surface plasmon resonance (SPR) sensor and a waveguide-coupled bimetallic (WcBiM) chip.

Fig. 2

(a) SPR grayscale image from two-dimensional complementary metal oxide semiconductor before the adsorption of the biomolecules. (b) SPR grayscale image after the adsorption of the biomolecules. The dark area of the sample channel is shifted. (c) Intensity detection mode. Blue and red solid lines represent before the reaction and after the reaction, respectively.

2.2.

Waveguide-Coupled Bimetallic Chip

The WcBiM stack, which consisted of a glass substrate (Eagle XG corning glass) and a waveguide (WG; $\mathrm{ZnS}\text{-}{\mathrm{SiO}}_2$) layer sandwiched between an Au layer and a silver (Ag) layer, is represented in the right part of Fig. 1. $\mathrm{ZnS}\text{-}{\mathrm{SiO}}_2$ was used as the waveguide layer due to the outstanding adhesion and smooth interface between Au and Ag metals. The WcBiM chip was fabricated on the corning glass using the radio frequency (RF) magnetron sputtering method with 5 mTorr (0.67 Pa) of working pressure in a pure Ar atmosphere. The RF power for the $\mathrm{ZnS}\text{-}{\mathrm{SiO}}_2$ and metallic Ag/Au was set to 80 and 20 W, respectively. As demonstrated in the previous work,²¹ an optimized multilayer stack of $\mathrm{Au}(31\mathrm{nm})/\mathrm{WG}(190\mathrm{nm})/\mathrm{Ag}(25\mathrm{nm})$ was selected using Fresnel equations and commercial optical thin-film software (SCI Film Wizard™, Carlsbad, CA, USA). To prove the high sensitivity of the WcBiM chip, its reflectance curve was compared with that of a conventional Au chip, which consists of $\mathrm{Au}(50\mathrm{nm})/\mathrm{Cr}(2\mathrm{nm})$.

2.3.

Chemicals and Reagents

For surface modification of the sensor chip, APTES in ethanol was utilized. EDC and sulfo-NHS were used to activate the carboxyl group of protein A. All chemical materials and proteins, including IgG, their relevant antibodies, phosphate-buffered saline (PBS, pH 7.4), protein A, and bovine serum albumin (BSA), were supplied from Sigma-Aldrich (St. Louis, Missouri). Diluted $1\times \mathrm{PBS}$ was used as the diluent of all proteins. SAM and the conventional Au chip were purchased from K-MAC.

2.4.

Linkage Layers

Protein A and SAM were used as the linkage layers to immobilize the anti-IgG, EDC-NHS-activated protein A. For SAM formation, 1 mM of NHS-SAM solution was prepared by dissolving 2 mg of NHS-SAM purchased from K-MAC in 2.5 ml of ethanol. A SAM was produced on the surface of the bare WcBiM chip by submerging it in the SAM solution for 12 h. The SAM-formed WcBiM chip was then installed in the SPR sensor. The formation of protein A on the WcBiM chip is relatively simple compared with the formation of SAM and the EDC-NHS-activated protein A on the chip. A total of $200\mu \text{g}/\mathrm{ml}$ of protein A solution was injected into the sample channel of the SPR sensor, giving rise to the formation of protein A on the bare chip. For EDC-NHS-activated protein A formation, surface modification of the WcBiM chip was carried out to make the surface reactable, as depicted in Figs. 3(a) and 3(b). First, a hydroxyl group ($─\mathrm{OH}$) was formed on the bare WcBiM chip using an oxygen plasma device (Femto Science, Hwaseong-si, Gyeonggi, Korea), as shown in Fig. 3(a). The power was set to 60 W for 1 min, and the working pressure was 0.1 Torr. Then, the WcBiM chip was immediately dipped in 2% APTES in ethanol for 1 h, resulting in an amino group (${\mathrm{NH}}_2$)-functionalized surface, as shown in Fig. 3(b). A total of $200\mu \text{g}/\mathrm{ml}$ of protein A (980 μl) was reacted with 20 μl of a catalyst agent, EDC (4 mg) and sulfo-NHS (11 mg), and diluted in PBS for 1 h, resulting in an activated carboxyl group of the protein A. Then the ${\mathrm{NH}}_2$-functionalized WcBiM chip was installed in the SPR sensor, and the EDC-NHS-activated protein A was injected into the sample channel of the SPR sensor. The EDC-NHS-activated protein A was captured on the ${\mathrm{NH}}_2$-functionalized WcBiM chip, as shown in Fig. 3(c). In the whole experiment, $150\mu \text{g}/\mathrm{ml}$ of anti-IgG was used. A total of $100\mu \text{g}/\mathrm{ml}$ of BSA was flowed to minimize nonspecific binding before antibody-antigen interactions. The volume of all proteins was 200 μl. They were injected into the fluidic channel at a flow rate of $20\mu \text{l}/\min$. The experiment was repeated three times.

Fig. 3

(a) Hydroxyl group (OH) formed on the surface of sensor chip by using oxygen plasma. (b) Amino group (${\mathrm{NH}}_2$) formed on OH-terminal surface by immersing the chip into 2% of 3-aminopropyltriethoxysilane (APTES) in the ethanol. (c) EDC-NHS-activated protein A captured on the ${\mathrm{NH}}_2$-functionalized surface.

3.1.

SPR Reflectance Curve

To investigate the SPR response characteristics of the WcBiM SPR sensor using the EDC-NHS-activated protein A, protein A, and SAM, the SPR reflectance curves were first obtained for the WcBiM and conventional Au chips. Figure 4 shows the recorded grayscale images for the WcBiM and Au chips and their corresponding SPR reflectance curves. As shown in Figs. 4(a) and 4(b), dark areas in the grayscale images corresponding to a dip in the SPR reflectance curve were observed [Figs. 4(c) and 4(d)] since there was almost no reflected light. The upper and lower areas in the grayscale images signify the collected reflectance for the sample and the reference channels. The images show that the dark area in the grayscale image for the WcBiM chip is much narrower than that in the Au chip. As depicted in Figs. 4(c) and 4(d), numeric data for the SPR reflectance curve were acquired by fitting the reflectance profile of the grayscale image. The resonance angles with the reflectance were 64.79 deg with 4.01% for the WcBiM chip and 65.57 deg with 3.85% for the Au chip. The FWHM of the WcBiM and Au chips was 1.47 and 2.23 deg, respectively. Thus, the width of the SPR reflectance curve of the WcBiM chip was narrower than that of the Au chip. The differential values with respect to the incident angle were calculated to find the optimized incident angles where the higher resolution can be obtained in the intensity detection mode. The highest absolute values of the derivative were $158.92\%/\mathrm{deg}$ at 64.13 deg for the WcBiM chip and $107.84\%/\mathrm{deg}$ at 64.98 deg for the Au chip. In other words, the gradient in the SPR curve of the WcBiM chip was steeper than that of the Au chip. From these results, it was expected that the WcBiM chip would have higher sensitivity in the intensity detection mode because it leads to a larger reflectance change.

Fig. 4

Grayscale images for the (a) WcBiM chip and (b) conventional Au (gold) chip. Numeric data for the SPR reflectance curve acquired by fitting the intensity profile of the grayscale image for the (c) WcBiM chip and (d) conventional Au (gold) chip.

3.2.

IgG Detection

To verify the sensitivity of the WcBiM SPR sensor using the EDC-NHS-activated protein A, a specific binding test, was carried out during the antigen–antibody interaction. The antigen-antibody interaction led to a shift in the SPR reflectance curve, including the resonance angle. The reflectance increment was measured at the fixed incident angle with the steepest slope in the SPR reflectance curve. For the detection of IgG, the surface of the WcBiM chip was modified with anti-IgG using the EDC-NHS-activated protein A, protein A, and SAM. In this experiment, SAM was formed on the sensor chip before it was installed in the SPR sensor, and the EDC-NHS-activated protein A and protein A ($200\mu \text{g}/\mathrm{ml}$) were injected into the sample channel after the bare SPR chip was installed in the sensor. To immobilize anti-IgG, anti-IgG solution ($150\mu \text{g}/\mathrm{ml}$) was flowed across the WcBiM chip, which was coated with three-linkage layers. BSA with a concentration of $100\mu \text{g}/\mathrm{ml}$ was injected to block any reactable sites and prevent nonspecific interactions. Finally, IgG was introduced at increasing concentrations to study the sensitivity and the linearity of SPR responses. The concentrations used were 50, 100, and $150\mathrm{ng}/\mathrm{ml}$.

The typical output responses from the two channels of the SPR system are shown in Fig. 5(a). The SPR response of the sample channel was increased by binding the proteins on the WcBiM chip after the sequence injection of the proteins. The reference channel was shown to be relatively constant compared to the sample channel. To exclude external environmental factors, the reference signal was subtracted from the sample signal, as shown in Fig. 5(b). The change in the reflectance following the injection of the EDC-NHS-activated protein A and protein A was, on average, 6.5111% and 11.8186%, respectively. The reflectance increment due to the anti-IgG and BSA using the EDC-NHS-activated protein A was, on average, 5.1293% and 0.8983%, respectively. The mean values when using protein A were 9.0293% and 1.1914%. In the case of SAM, they were 7.7069% and 0.8292% as shown in Fig. 5. The change in the reflectance with anti-IgG was highest when using protein A, as shown in Fig. 6. The vertical lines show the error bars. The error bar using protein A is the largest among the three-linkage layers due to the protein being randomly adsorbed on the sensor chip.

Fig. 5

(a) Signals from two channels: sequence injection of anti-IgG, BSA, and IgG (50, 100, and $150\mathrm{ng}/\mathrm{ml}$) at the sample channel and injection of PBS solutions at the reference channel. (b) The actual meaningful signals due to biomolecular interaction acquired by taking off the reference signal from the sample signal.

Fig. 6

Mean value of change in the reflectance for the injection of anti-IgG. Vertical lines represent the error bar in triplicate.

To compare the detection capabilities of the EDC-NHS-activated protein A, protein A, and SAM in the WcBiM sensor, the SPR response at IgG concentrations of 50, 100, and $150\mathrm{ng}/\mathrm{ml}$ was analyzed. The results of repeat experiments performed in triplicate showed that the mean reflectance change for IgG at the concentrations of 50, 100, and $150\mathrm{ng}/\mathrm{ml}$ using the EDC-NHS-activated protein A was 0.8405%, 1.6309%, and 2.6744%, respectively. In case of protein A, the average change in the reflectance at each concentration was 0.1987%, 0.5121%, and 0.8465%. The reflectance increments for SAM were 0.3988%, 0.7570%, and 1.3778%. These results demonstrate that the change in the reflectance was highest in the WcBiM SPR sensor chip using the EDC-NHS-activated protein A, even though the output signal for anti-IgG using protein A had the highest increment. Therefore, the binding of IgG does not seem to depend on the SPR response to anti-IgG. In addition, the EDC-NHS-activated protein A led to directional immobilization of anti-IgG, resulting in an increase in the binding of IgG. The increased binding capability is primarily due to the oriented formation of the protein A molecules due to the activated carboxyl group. The activated carboxyl group of protein A covalently interacted with the ${\mathrm{NH}}_2$-functionalized sensor chip, as shown in Fig. 3(c). Even though carboxyl group of the amino acid in the five Fc-binding domains of protein A reacts with the ${\mathrm{NH}}_2$-functionalized sensor chip, the other four homologous Fc-binding domains are still available to bind to Fc region of anti-IgG, resulting in Fab region of anti-IgG facing away from the sensor chip. Thus, it is expected that site-directed immobilization would enhance the ability of the sensor to detect IgG.

To confirm the observations and results, the calibration curves for each IgG concentration were obtained. The average change in the reflectance of IgG at the concentrations of 50, 100, and $150\mathrm{ng}/\mathrm{ml}$ are plotted in Fig. 7. The symbols depicted in Fig. 7 are the mean values for triplicate measurements, and the solid line joining them represents the best linear fitting. The error bars represent the standard deviation. As shown, the WcBiM SPR sensor using the EDC-NHS-activated protein A has the highest sensitivity among the three-linkage layers since the slope is the highest among the three-linkage layers. The sensitivity, defined as the change in the reflectance per unit change in the concentration of IgG, was 0.0185 [$\%/(\mathrm{ng}/\mathrm{ml})$] for the EDC-NHS-activated protein A, 0.0065 [$\%/(\mathrm{ng}/\mathrm{ml})$] for the protein A, and 0.0101 [$\%/(\mathrm{ng}/\mathrm{ml})$] for the SAM. The correlation coefficient for the three-linkage layers was almost equal at approximately 99%. To evaluate the limit of detection (LOD), the lowest detectable concentration of IgG was acquired by calculating Eq. (1) as follows.²²

Fig. 7

Calibration curve of IgG with 50, 100, and $150\mathrm{ng}/\mathrm{ml}$. Symbols and the error bars represent the mean values and the standard deviation for triplicate, respectively.

The standard deviation for the EDC-NHS-activated protein A, protein A, and SAM was 0.0239%, 0.0170%, and 0.0219%, respectively. The calculated values for the EDC-NHS-activated protein A, protein A, and SAM were $4.27\mathrm{ng}/\mathrm{ml}$, $12.83\mathrm{ng}/\mathrm{ml}$, and $8.24\mathrm{ng}/\mathrm{ml}$, respectively. Based on the analysis of the LOD, the WcBiM sensor system using the EDC-NHS-activated protein A had higher sensitivity than the systems using protein A or SAM. These observations suggest that the WcBiM sensor with the EDC-NHS-activated protein A can detect very low levels of biomolecules.