Polystyrene 96-well microtiter plates (Costar, NY, USA) were used to perform the immunoreactions. CEA and anti-CEA antibody were purchased from R & D Systems (Minneapolis, MN, USA). Goat-anti-mouse IgG was purchased from Beijing Chengwen Company (Beijing, China). Bovine serum albumin (BSA), HAuCl₄·3H₂O (99.99%) was bought from Sigma-Aldrich (USA). Trisodium citrate, NaH₂PO₄, Na₂HPO₄, and NaCl are obtained from Beijing Chemical Reagents Company (Beijing, China). The luminol stock solution (2.5 × 10⁻² M) was prepared by dissolving luminol (obtained from Sigma-Aldrich, USA) in 0.1 M NaOH solution and stored in a dark place. The buffers used were as follows: (A) coating buffer, 0.05 M carbonate/bicarbonate buffer solution, pH 9.6; (B) blocking buffer, 1% (w/v, g·mL⁻¹) BSA in 0.01 M sodium phosphate buffered saline with 0.05% (v/v) Tween 20 (PBST-BSA), pH 7.4. The blocking buffer was stored at 4 °C and used within a week; (C) washing buffer, 0.01 M PBS with 0.05% (v/v) Tween 20, pH 7.4; In all the procedures, the water used was purified through an Olst ultrapure K8 apparatus (Olst, Ltd., China, resistivity >18 MΩ). All other reagents were of analytical reagent grade and used as purchased without further purification. The CL intensity was measured and recorded with an Ultra-weak luminescence analyzer and software BPCL-T15 (Institute of Biophysics, Academic Sinica, Beijing, China). Two peristaltic pumps were used to deliver all the chemicals. A six way injection valve fitted with a 100 μL sample loop was used for the injection of the sample solution. PTFE tubing (1.0 mm i.d.) was used to connect all components in the flow system. Transmission electron microscopy (TEM) was performed with a JEOL-100CX electronmicroscope (Jeol, Japan) under 80 kV accelerating voltage.

A solution of 15 nm AuNPs was synthesized according to a literature procedure [27] with slight modifications. Briefly, 5.0 mL of 1% trisodium citrate was quickly added to 100 mL of boiling 0.01% HAuCl₄. Solution under reflux was stirred for 10 min, during which the color changed to red. After being slowly cooled down to room temperature, the solution was centrifuged to remove impurities and ions, and then diluted to 100 mL. The AuNPs have an average diameter of ca. 15 nm.

The preparation of AuNPs-labeled goat-anti-mouse IgG was performed according to the modification in literature [28]. The pH of the AuNPs was adjusted to 8.0 with 0.1 M Na₂CO₃. Precisely, 1.0 mL of 550 μg·mL⁻¹ goat-anti-mouse IgG (10% more than the minimum amount, which was determined using a flocculation test) was added to 5 mL of pH-adjusted AuNPs, followed by incubation at room temperature for 30 min. Afterward, 5% BSA was added to a final concentration of 1% with stirring about 5 min. To remove the excess of antibody, the conjugates were centrifuged at 10,000 rpm for 10 min, and the soft sediment was resuspended in 0.01 M PBS containing 1% BSA. It can be used directly or stored in 0.01 M PBS buffer with 50% glycerol for several months at −20 °C.

The assay was performed in a polystyrene 96-well microtiter plate. Initially, 200 μL of serially diluted human CEA dilutions in coating buffer were coated on wells of the plate overnight at 4 °C. The unbound antigen was washed off three times with 300 μL of washing buffer and the uncoated active sites of polystyrene substrate were saturated with 300 μL of blocking buffer for 1 h at 37 °C, in which BSA was used as a blocking agent to prevent nonspecific adsorption of the antibody in the next step. Afterward, 200 μL of 5 μg·mL⁻¹ anti-CEA antibody was added into the wells and incubated for 2 h at 37 °C. The wells were washed three times with washing buffer followed by addition of 200 μL AuNPs-labeled goat-anti-mouse IgG. Finally the wells were washed thoroughly with washing buffer (three times) and pure water (three times). Then, the wells were immediately treated with 200 μL per well of a mixture of 0.01% HAuCl₄ and 0.4 mM NH₂OH·HCl in a dark for 2 min at 30 °C.

Metallic gold on enhanced plates were washed three times with 300 μL of pure water, dried, and dissolved with 200 μL of 2.0% HNO₃-3.4% HCl for three hours at room temperature to ensure that the Au nanoparticles were completely dissolved. Solutions were then transferred to 1.5 mL centrifuge tubes and 200 μL H₂O and 75 μL 0.1 M sodium tartrate (C₄H₄O₆Na₂) were added to each sample in turn. Then 1.0 M NaOH was added to the AuCl₄⁻ solutions to adjust pH. As shown in Figure 1, AuCl₄⁻ was reacted with the mixture of luminol and H₂O₂ in the flow cell to produce the CL signal. The CL signals were monitored with a photomultiplier tube adjacent to the flow cell.

A schematic representation of the detection principle of this noncompetitive CL immunoassay is shown in Scheme 1. The human CEA analyte is first immobilized on a 96-well polystyrene microtiter plate. The mouse-anti-CEA as primary antibody is then captured by the CEA and sandwiched by a goat-anti-mouse IgG secondary antibody labeled with colloidal gold. After that, we used a method for enlargement of colloidal Au nanoparticles called “seeding”, based on the colloidal Au surface-catalyzed reduction of Au³⁺ by NH₂·OH. While NH₂·OH is thermodynamically capable of reducing Au³⁺ to bulk metal, the reaction is dramatically accelerated by Au surfaces. As a result, no new particle nucleation occurs in solution, and all added Au³⁺ goes into production of larger particles. In this case, Au NPs are specifically enlarged by HAuCl₄ and NH₂OH·HCl for 2 min [9]. Next, the AuNPs are dissolved in a 2.0% HNO₃-3.4% HCl solution and the gold ions (AuCl₄⁻) were stripped out from the solid polystyrene substrate surface. The concentration of CEA was quantified based upon the concentration of dissolved AuCl₄⁻, which was quantified by the CL intensity. A comparison of analytical performance before and after gold amplification was also investigated and the detailed optimization of the gold amplified FI-CL immunoassay is reported in the following sections.

The AuNPs used in this work before and after enlargement with HAuCl₄ and NH₂OH·HCl are shown in Figure 2A,B. To detect the enlarged colloidal gold label by CL, the gold was dissolved to form AuCl₄⁻, which can catalyze the luminol-H₂O₂ CL reaction. As the aqua regia (16.0% HNO₃-27% HCl) resulted in a high CL background, dilution of the aqua regia was used to dissolve the colloidal gold. Therefore, the effect of the dissolution reagent concentration on CL intensity was investigated. The signal/noise ratio increased with increasing dissolution reagent concentration and reached a maximum value at 1:7 dilution of aqua regia (2.0% HNO₃-3.4% HCl), as shown in Figure 3. It is possible that AuNPs could not be completely dissolved when the concentration of the dissolution reagent was lower than the 1:7 dilution of aqua regia. The signal/noise ratio decreased gradually when it exceeded 1:7 and it is likely that the strong ionic strength in the solution induces an increase in the CL background intensity when a much higher concentration of dissolution reagent is used. Thus, the 1:7 dilution of aqua regia was used as the optimum dissolution reagent for the CL assay.

The catalytic activity of AuCl₄⁻ on the luminol-H₂O₂ CL reaction is greatly influenced by the pH of the AuCl₄⁻ solution. Therefore, the effect of AuCl₄⁻ pH on CL intensity was investigated by using 1.0 M NaOH and 0.1 M C₄H₄O₆Na₂ to adjust the acidity of the solution after dissolving the AuNPs with 1:7 (v/v) aqua regia. As shown in Figure 4, the signal/noise ratio increased with increasing pH of sample solution because the blank CL intensity decreased with increasing sample pH below 4.0, reaching the highest value at pH 4.0. However, when the pH of the sample was higher than 4.0, the signal/noise ratio decreased. The cause is that the stability of AuCl₄⁻ has been destroyed. Therefore, pH 4.0 was chosen as the optimum pH value for the CL measurement.

To optimize the proposed CL assay, the effects of luminol and H₂O₂ pH on the CL intensities were studied. The optimization of the pH value of luminol was investigated over the pH range 10.0–13.0. The result indicates that the signal/noise ratio had a maximum at pH 12.5. The effects of luminol and H₂O₂ concentration on the CL intensity of the sample and blank were also examined. This indicated that the CL intensity increased as the luminol and H₂O₂ concentrations were increased but the CL intensity of the blank became very high when the concentration of luminol exceeded 1.0 × 10⁻⁴ M and led to poor reproducibility. Considering the CL intensity and the consumption of the reagents, 7.5 × 10⁻⁵ M and 0.01 M H₂O₂ were chosen for subsequent work (data not shown).

Under the optimum conditions described above, the relationship between CEA concentration and CL intensity was investigated. The results showed that the CL intensities of luminol-H₂O₂-AuCl₄⁻ increased with the increase of concentration of the CEA ranging from 10 pg·mL⁻¹ to 1 μg·mL⁻¹ (Figure 5A). The linear range for CEA was 100 pg·mL⁻¹ to 1 μg·mL⁻¹ with the equation of lg[y] = 3.8077 + 0.1215l g[x] (y is the CL intensity; x is the concentration of CEA, ng·mL⁻¹; n = 3, R = 0.9892; Figure 5B). The relative standard deviations for 10 pg·mL⁻¹ and 1 μg·mL⁻¹ of CEA were 0.44% and 4.7%, respectively (n = 3). The detection limit was 20 pg·mL⁻¹. Table 1 shows the comparison between the proposed CL method and general immunoassay formats for determination of CEA. It can be seen that the proposed CLIA is competitive with or better than other immunoassay formats and has the advantage of simple instrumentation.