Huan
Li
a,
Zhijuan
Cao
a,
Yuhao
Zhang
b,
Choiwan
Lau
a and
Jianzhong
Lu
*a
aSchool of Pharmacy, Fudan University, 826 Zhangheng Road, Shanghai, 201203, China. E-mail: jzlu@shmu.edu.cn; Fax: +86 21-51980058
bZhongshan Hospital, Fudan University, 138 Yixueyuan Road, Shanghai, 200032, China
First published on 7th July 2010
A new concept is proposed in this article to detect multiple cancer markers in a single sample. Quantum dot (QD) fluorescence (FL) labels were successfully combined with enzyme chemiluminescence (CL) labels for simultaneous detection of three cancer markers in human serum using just a common 96-well plate reader and with equal detection limits for the three markers. As a proof-of-concept, herein we coupled one QD FL label with two enzyme CL labels for hybrid multiplexed detection of lung cancer markers as exemplified by neuron-specific enolase (NSE), carcinoembryonic antigen (CEA) and cytokeratin fragment (Cyfra21-1). A homogeneous “sandwich-type” detection strategy was employed herein, where the bead–antibody mixture first reacts with NSE, CEA and Cyfra21-1 to initiate three immunoreactions in a single tube; and then the formed conjugates sandwiches with biotin, digoxin and fluoresceinisothiocyanate (FITC)-modified detection antibodies, and further reacts with a mixture of streptavidin QD, anti-FITC horseradish peroxidase (HRP) and anti-digoxin alkaline phosphatase (ALP) for subsequent CL and FL detection. The results show that NSE, CEA and Cyfra21-1 could be sensitively determined with a common 96-well plate reader and with equal detection limits down to the ng mL−1 level. Furthermore, the proposed method has been successfully applied to the determination of three cancer markers in human samples without cross-reaction. Because it is straightforward to adapt this strategy to detect a spectrum of other proteins by using different antibodies or aptamers, this new CL strategy might create a universal technology for developing simple biosensors in the sensitive and selective detection of multiple targets in a variety of clinical, environmental, and biodefense applications.
In addition, chemiluminescence (CL) has been exploited for a wide range of applications in various fields owing to its extremely high sensitivity along with its other advantages, such as simple instrumentation, wide calibration ranges and suitability for miniaturization in analytical chemistry. CL methods have been generally employed in the CL flat surface arrays. For example, spatial separation along with a single label was employed in a multianalyte hybridization assay for detection of various pathogens by constructing microtiter plates containing main wells with built-in subwells, each subwell corresponding to a single assay;17 A substrate zone resolved multianalyte immunoassay has been developed for the detection of CA125, CA 153, CA 199 and carcinoembryonic antigen (CEA) by employing horseradish peroxidase (HRP) and alkaline phosphatase (ALP) as labels, respectively.18,19 Recently, a few reaction encoding-based CL methods have also been developed in several groups. A temperature-resolved technique has been proposed in a homogeneous CL immunoassay system by using HRP to label the tracer antibody for the sequential CL detection of two proteins.20 Four kinds of CL labels, i.e. HRP, ALP, aequorin and galactosidase, have also been simultaneously determined for the quadruple-analyte CL hybridization assay.21
It is presumed that the CL labels hyphenated with FL labels could increase significantly the number of targets which can be simultaneously determined. Most importantly, there are many instruments to be designed for CL and FL simultaneous detection, high-performance optical system enables the machine with high detection sensitivity and as a combination instrument, it also covers the full range of FL as well as glow-type and flash-type CL applications. As a proof-of-concept, herein we developed an enzyme CL hyphenated with QD FL technology for hybrid multiplex detection of three tumor markers exemplified by neuron-specific enolase (NSE), CEA and cytokeratin fragment (Cyfra21-1). The optimization and attractive performance characteristics of this new hyphenated CL and FL detection of three model proteins are reported in the following sections. It was found that this new technique was competitive with or even better than previously reported single-analyte or multianalyte detection systems employing other technologies.
Carboxymethoxylamine hemihydrochloride (31 mg) and sodium acetate (23 mg) were dissolved in 0.3 mL of water and placed into a two-necked flask. The digoxin dialdehyde, dissolved in 200 μL of methanol, was placed into a pressure-equalized funnel and slowly added to the flask while the reaction mixture was stirred under nitrogen. The reaction was complete within 10 min. The reaction mixture was evaporated to dryness and the residue dissolved in 2 mL of ethyl acetate and 300 μL of water. The organic layer was separated and the aqueous layer was washed three times with ethyl acetate. The organic layers were combined and dried over anhydrous magnesium sulfate. The resultant solution was filtered and evaporated to dryness. The residue, digoxin (bis[O-(carboxymethyl)oxime], was further dried for 30 min under high vacuum (0.1 mmHg) and used immediately for the next step.
Dicyclohexylcarbodiimide (DCC, 28 mg) was dissolved in 600 μL of dry dimethylformamide (DMF) and placed into a two-necked flask. The solution was cooled in an ice bath (4 °C). Digoxin (bis[O-(carboxymethyl)oxime], dissolved in 8 mL of DMF, was slowly added while the reaction mixture was stirred under nitrogen. Then N-hydroxysuccinimide (NHS) solution (15 mg in 600 μL of DMF) was added. The reaction continued at 4 °C under nitrogen for 18 h. The white solid precipitate in the mixture was removed by filtration. The filtrate was then evaporated on a rotary evaporator under vacuum (0.1 mmHg). An oil, dioxime active ester (60 mg), was left and stored in a desiccator before use.
Scheme 1 Schematic representation of quantum dot fluorescence labels hyphenated with enzyme chemiluminescence for multiplex detection of lung cancer markers. |
The ALP signal was observed to decrease considerably when measured after HRP. This was attributed to ALP inactivation due to exposure of the dimethyl sulfoxide (DMSO) solution of the HRP substrate. A washing step was introduced after each measurement to remove the substrate and possible signal carryover before proceeding to the next step. However, for simplicity, here we divided the resultant sample into three equal parts after the last wash. Each part was then used for direct measurement of one specific label after adding the corresponding substrate. It should be noted that this simple and quick detection procedure decreased the sensitivity three fold but avoided the in-between washing steps and shortened the detection time by more than 30 min.
Fig. 1 FL and CL intensity versus the amount of coating antibody. Experimental conditions: NSE (☆, left Y-axis), CEA (△, right Y-axis), or Cyfra21-1 (○, right Y-axis), 78 μg PS beads was used; antigens were 10 ng mL−1; biotin-modified NSE detection antibody was 2.8 μg well−1, digoxin-modified CEA detection antibody was 1.2 μg well−1 and FITC-modified Cyfra21-1 detection antibody was 3.45 μg well−1; SA-QD was 1:500, anti-FITC HRP was 1:1000 and anti-digoxin ALP was also 1:1000. The detection procedure was carried out as described in the Experimental Section. |
The amount of the bead–antibody conjugates was then optimized to give the highest FL or CL intensity (Fig. 2). Higher FL or CL intensity was achieved with 107 μg NSE bead–antibody conjugates, 80 μg CEA bead–antibody conjugates, and 187 μg Cyfra21-1 bead–antibody conjugates. These amounts were then selected for further studies. The amount of the detection antibody was also optimized for FL or CL intensity (Fig. 3), and consequently subsequent work employed 2.8 μg well−1 of biotin-modified NSE detection antibody, 1.2 μg well−1 of digoxin-modified CEA detection antibody, and 3.45 μg well−1 of FITC-modified Cyfra21-1 detection antibody. Next the amounts of SA-QDs, anti-digoxin ALP and anti-FITC HRP were optimized for FL or CL intensity. Optimum amounts were found to be 1:200 for SA-QD, 1:500 for anti-digoxin ALP and 1:1000 for anti-FITC HRP (Fig. 4).
Fig. 2 FL and CL intensity versus the amount of bead–antibody conjugates. Experimental conditions: NSE (☆, left Y-axis), CEA (△, right Y-axis), or Cyfra21-1 (○, right Y-axis), 0.031 μg NSE coating antibody μg−1 PS beads, 0.041 μg CEA coating antibody μg−1 PS beads, and 0.0062 μg Cyfra21-1 coating antibody μg−1 PS beads were used. Antigens were 10 ng mL−1; biotin-modified NSE detection antibody was 2.8 μg well−1, digoxin-modified CEA detection antibody was 1.2 μg well−1 and FITC-modified Cyfra21-1 detection antibody was 3.45 μg well−1; SA-QD was 1:500, anti-digoxin ALP was 1:1000 and anti-FITC HRP was 1:1000. The detection procedure was carried out as described in the Experimental section. |
Fig. 3 FL or CL intensity versus the amount of detection antibody. Experimental conditions: NSE (☆, left Y-axis), CEA (△, right Y-axis), or Cyfra21-1 (○, right Y-axis), 2.4 μg NSE coating antibody, 3.2 μg CEA coating antibody and 0.48 μg Cyfra21-1 coating antibody. 107 μg bead–NSE coating antibody conjugates, 80 μg bead–CEA coating antibody conjugates, 187 μg bead–Cyfra21-1 coating antibody conjugates; antigens were 10 ng mL−1; SA-QD was 1:500, anti-digoxin ALP was 1:1000 and anti-FITC HRP was 1:1000. The detection procedure was carried out as described in the Experimental section. |
Fig. 4 FL and CL intensity versus the amount of signaling molecules. Experimental conditions: NSE (☆, bottom X-axis, left Y-axis), CEA (△, top X-axis, right Y-axis), or Cyfra21-1 (○, top X-axis, right Y-axis), 2.4 μg NSE coating antibody, 3.2 μg CEA coating antibody and 0.48 μg Cyfra21-1 coating antibody. 107 μg bead–NSE coating antibody conjugates, 80 μg bead–CEA coating antibody conjugates, 187 μg bead–Cyfra21-1 coating antibody conjugates; antigens were 10 ng mL−1; biotin-modified NSE detection antibody at 2.8 μg well−1, digoxin-modified CEA detection antibody at 1.2 μg well−1, FITC-modified Cyfra21-1 detection antibody at 3.45 μg well−1; The detection procedure was carried out as described in the Experimental section. |
Fig. 5 Study of the specificity of QD FL labels hyphenated with enzyme CL for the detection of three tumor markers. Experimental conditions: 107 μg bead–NSE coating antibody conjugates, 80 μg bead–CEA coating antibody conjugates, 187 μg bead–Cyfra21-1 coating antibody conjugates; biotin-modified NSE detection antibody at 2.8 μg well−1, digoxin-modified CEA detection antibody at 1.2 μg well−1, FITC-modified Cyfra21-1 detection antibody at 3.45 μg well−1; and SA-QDs at 1:200, anti-digoxin ALP at 1:500, anti-FITC HRP at 1:1000. Each antigen was at 50 ng ml−1. The detection procedure was carried out as described in the Experimental section. |
The three antigen mixtures were analyzed in a quantitative fashion to determine the FL or CL response to increasing levels of the three tumor markers in a sample (Fig. 6). CL and FL intensities were proportional to the amount of the corresponding antigen. The calibration graphs in the concentration range of 0.5–50 ng mL−1 showed good linear correlations between the intensity and the concentration of NSE (y = 0.0204x − 0.0231 R2 = 0.9984), CEA (y = 3768x + 11657 R2 = 0.9844), and Cyfra21-1 (y = 2944x + 434, R2 = 0.9987). Our novel hyphenated technique allowed detection of three lung cancer markers down to the ng mL−1 level and was competitive with or even better than other assay formats (Table 1). The recoveries of NSE (10, 25 and 50 ng mL−1) in the presence of 50 ng mL−1 CEA and Cyfra21-1 were 89.8, 104.2 and 110.1%, respectively. Recoveries of CEA (10, 25 and 50 ng mL−1) in the presence of 50 ng mL−1 NSE and Cyfra21-1 were 98.4, 95.6 and 107.7%, respectively. Recoveries of Cyfra21-1 (10, 25 and 50 ng mL−1) with 50 ng mL−1 NSE and CEA were 101, 88.1 and 97.6%, respectively.
Fig. 6 FL and CL intensity versus the concentration of tumor marker. (A) A mixture of NSE (☆, left Y-axis), CEA (△, right Y-axis), or Cyfra21-1 (○, right Y-axis), and (B) individual markers. Experimental conditions: 107 μg bead–NSE coating antibody conjugates, 80 μg bead–CEA coating antibody conjugates, 187 μg bead–Cyfra21-1 coating antibody conjugates; biotin-modified NSE detection antibody at 2.8 μg well−1, digoxin-modified CEA detection antibody at 1.2 μg well−1, FITC-modified Cyfra21-1 detection antibody at 3.45 μg well−1; and SA-QD at 1:200, anti-digoxin ALP at 1:500, anti-FITC HRP at 1:1000. The detection procedure was carried out as described in the Experimental section. |
Label | Number of analytes | Detection method | Sensitivity | Ref. |
---|---|---|---|---|
HRP and ALP | 2 | CL | 1 ng mL−1 and 5 U mL−1 | 18 |
HRP and ALP | 4 | CL | 0.5, 2, 5, U mL−1 1 ng mL−1 | 19 |
HRP | 2 | CL | 2,1.5 ng mL−1 | 20 |
ALP | 4 | CL | 0.52 and 0.55 ng mL−1, 0.49 and 0.79 U mL−1 | 25 |
HRP | 10 | CL | 0.12–32 ng mL−1 | 26 |
HRP | 3 | CL | 5 ng mL−1, 105 cfu mL−1, 107 pfu mL−1 | 27 |
Cy5 | 3 | FL | 0.13 μg mL−1 | 28 |
Cy5 | 3 | FL | 105 cfu mL−1, 107 pfu mL−1, 10 ng mL−1 | 29 |
QD | 2 | FL | — | 30 |
QD | 4 | FL | — | 14 |
QD | 4 | FL | — | 15 |
CdS | 4 | Electrochemical | 10 ng mL−1 | 31 |
ALP | 4 | Electrochemical | 1, 2, 1.2 and 1 ng mL−1 | 32 |
AuNP | 2 | Electrochemical | 1 and 1.5 ng mL−1 | 33 |
ALP | 4 | Electrochemical | 3 ng mL−1 | 34 |
Label-free | 4 | Electrochemical | 0.5 ng mL−1 | 35 |
AuNPs | 2 | SPR | 30 ng mL−1 | 36 |
HRP | 2 | Colorimetric | 0.02 ng mL−1. | 37 |
Eu3+, Sm3+ | 2 | ICPMS | 1.2 and 1.7 ng mL−1 | 8 |
Label-free | 4 | Piezoelectric sensor | 4, 2, 3 and 12 ng mL−1 | 38 |
HRP | 1 | EIA | 1 ng mL−1 | CanAg kit 420-10 |
HRP | 1 | EIA | 0.25 ng mL−1 | CanAg kit 401-10 |
ALP, HRP and QD | 3 | FL/CL | 0.2 ng mL−1 | This work |
In addition to being very reproducible, we have also shown that our system is rapid. Even with 15 min incubations for each step, the total reaction time was within 1 h and the detection limit was almost unchanged. Consequently, our system should be at least as fast as current and commercially available assay formats.
We also challenged our method with 24 human serum samples from normal individuals. With an increase in human serum, the FL intensity decreased,24 which was attributed to interference from the human serum (Fig. 7). Thus 50% normal human serum was always included in the antigen-antibody reaction buffer for serum detection. The average recoveries for 20, 40 and 80 ng mL−1 from 24 normal serum samples were 93.0, 88.7 and 97.1% for NSE, respectively. For CEA the corresponding recoveries were 107.1, 116.4 and 96.2%, and 95.0, 93.9 and 90.8% for Cyfra21-1. Assays on serum from normal individuals clearly indicated that this method would be suitable for routine assay of clinical samples.
Fig. 7 FL intensity versus the amount of signaling molecules. Experimental conditions: 2.4 μg NSE coating antibody, 107 μg bead–NSE coating antibody conjugates; biotin-modified NSE detection antibody at 2.8 μg well−1; SA-QD was 1:200. The detection procedure was carried out as described in the Experimental section. |
This journal is © The Royal Society of Chemistry 2010 |