Magnetic Bead-Sensing-Platform-Based Chemiluminescence

Article pubs.acs.org/ac

Magnetic Bead-Sensing-Platform-Based Chemiluminescence Resonance Energy Transfer and Its Immunoassay Application Guoxin Qin, Shulin Zhao,* Yong Huang, Jing Jiang, and Fanggui Ye Key Laboratory for the Chemistry and Molecular Engineering of Medicinal Resources (Ministry of Education), College of Chemistry and Chemical Engineering, Guangxi Normal University, Guilin, 541004, China S Supporting Information *

ABSTRACT: A competitive immunoassay based on chemiluminescence resonance energy transfer (CRET) on the magnetic beads (MBs) is developed for the detection of human immunoglobulin G (IgG). In this protocol, carboxylmodified MBs were conjugated with horseradish peroxidase (HRP)-labeled goat antihuman IgG (HRP-anti-IgG) and incubated with a limited amount of fluorescein isothiocyanate (FITC)-labeled human IgG to immobilize the antibody− antigen immune complex on the surface of the MBs, which was further incubated with the target analyte (human IgG) for competitive immunoreaction and separated magnetically to remove the supernatant. The chemiluminescence (CL) buffer (containing luminol and H2O2) was then added, and the CRET from donor luminol to acceptor FITC in the immunocomplex on the surface of MBs occured immediately. The present protocol was evaluated for the competitive immunoassay of human IgG, and a linear relationship between CL intensity ratio (R = I425/I525) and human IgG concentration in the range of 0.2−4.0 nM was obtained with a correlation coefficient of 0.9965. The regression equation was expressed as R = 1.9871C + 2.4616, and a detection limit of 2.9 × 10−11 M was obtained. The present method was successfully applied for the detection of IgG in human serum. The results indicate that the present protocol is quite promising for the application of CRET in immunoassays. It could also be developed for detection of other antigen−antibody immune complexes by using the corresponding antigens and respective antibodies.

R

effective chemiluminescence (CL) donor or reaction that can excite a fluorescent acceptor by energy transfer. Recently, Zhang et al.11 developed a new method based on DNA molecular beacons and CRET involving luminol to fluorescein on magnetic nanoparicles for detection of ATP. Huang et al.12 reported a new strategy of CRET by using gold nanoparticles as efficient long-range energy acceptors for the detection of α-fetoprotein. Willner and coworkers13,14 demonstrated the application of CRET for the detection of aptamer substrate complexes, metal ion, and DNA. These studies indicate that the CRET system was simple and applicable for the detection of biomolecules with low detection limit. Magnetic beads (MBs) that are biocompatible can effectively conjugate biological molecules on their surface and provide convenient methods for the detection of biomolecules. In the past two decades, MBs have been widely used in immunoassays;15 enzyme immobilization;16 and other biochemical fields, such as the detection of proteins,17 nucleic acid hybridization,18 and telomerase activity.19 Herein, we report on the development of a competitive immunoassay based on CRET on the MBs for the detection of

esonance energy transfer is known as extremely sensitive to separation distance between the donor and acceptor, in which a luminescent donor transfers energy via nonradiative dipole−dipole interaction to a fluorescent or nonfluorescent acceptor when brought in proximity by a bioaffinity reaction. Resonance energy transfer that occurs between two fluorophores is intitled fluorescence resonance energy transfer (FRET), which is particularly useful for applications such as single-molecule spectroscopy,1,2 protein folding,3,4 cellular imaging,5 and immunoassay.6 However, as with any fluorescence technique, severe autofluorescence of biosystems and photobleaching of fluorescence dyes limit the usefulness of FRET. Resonance energy transfer that occurs between a light-emitting donor enzyme and a fluorophore is known as bioluminescence resonance energy transfer (BRET), which is emerging as a useful tool for detecting protein oligomerization,7 studying protein−protein interaction,8 and the detection of biomolecules.9 Chemiluminescence resonance energy transfer (CRET) which is similar to BRET, involves nonradiative (dipole−dipole) transfer of energy from a chemiluminescent donor to a fluorophore acceptor.10 In contrast to FRET, CRET occurs by the oxidation of a luminescent substrate without an excitation source, which makes CRET an improved sensitivity for bioassays owing to the absence of sample autofluorescence. However, little study has been reported so far on CRET. A major difficulty is to identify an © 2012 American Chemical Society

Received: November 8, 2011 Accepted: February 15, 2012 Published: February 15, 2012 2708

dx.doi.org/10.1021/ac202959d | Anal. Chem. 2012, 84, 2708−2712

Analytical Chemistry

Article

the PBS (3 × 300 μL) and resuspended to a final volume of 1.0 mL with the PBS. This suspension was stored at 4 °C for further use. Preparation of Human Serum Samples. Human blood samples were kindly provided by the No. 5 People’s Hospital (Guilin, China). Human blood samples were centrifuged at 2000 rpm for 15 min to obtain serum. These samples were stored at −20 °C until analysis, and diluted 2.0 × 104 fold with PBS before analysis. Determination of Human IgG. A 250 μL volume of 0.5 nM MB−HRP-anti-IgG suspension was mixed with 500 μL of 0.05 nM FITC-IgG solution in a 5 mL Eppendorf tube, and the resulting mixture was incubated for 30 min at 37 °C12 to obtain the MB−HRP-anti-IgG−FITC-IgG immune complexes. The suspension was separated magnetically and resuspended in 750 μL of PBS, then 250 μL of human IgG standard solution or sample solution was added to the suspension and incubated for 30 min at 37 °C for competitive immunoreactions. Subsequently, the immunocomplexes on the surface of the MBs were separated magnetically and redispersed in 1.0 mL of PBS. The resultant mixture was poured into a 1 cm path length quartz cell, then 200 μL of CL buffer solution was quickly added to the quartz cell, and the CL spectrum was recorded immediately with a LS-55 luminescence spectrometer.20 The luminescence intensity ratio R (I425/I525) was used for quantification of human IgG.

human immunoglobulin G (IgG, as model analyte). In this protocol, carboxyl-modified MBs were conjugated with horseradish peroxidase (HRP)-labeled goat antihuman IgG (HRP-antiIgG) and incubated with a limited amount of fluorescein isothiocyanate (FITC)-labeled human IgG (FITC-IgG) to immobilize the antibody−antigen immunocomplex on the surface of the MBs. In the presence of luminol and H2O2, HRP-anti-IgG catalyzes the oxidation of luminol by H2O2 to yield CL. The close proximity between the CL light source and the FITC-IgG in immune complexes on the surface of the MBs leads to CRET to FITC, a process that activates the luminescence of FITC. This process was used as a CRET immunosensor for the detection of human IgG. Quantitative measurements of IgG in human sera have been demonstrated, and thus provide a promising potential in clinical diagnosis.

■

EXPERIMENTAL SECTION Apparatus and Reagents. CL spectra were measured with a LS-55 luminescence spectrometer (Perkin-Elmer, USA); UV−visible spectra were measured with a TU-1901 UV−visible spectrophotometer (Beijing Purkinje General Instrument Co., Ltd., China). HRP-anti-IgG, FITC-IgG, human IgG, and human IgM were obtained from Beijing Biosynthesis Biotechnology Co. (Beijing, China). Carboxyl-modified MBs (1 μm diameter, 20.0 mg/mL), N-hydroxysuccinimide (NHS), 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide hydrochloride (EDC), and bovine serum albumin (BSA) were purchased from Sigma-Aldrich (St. Louis, MO). 2(N-Morpholino)-ethanesulfonic acid monohydrate (MES) was purchased from Fluka BioChemika (Buchs, Switzerland). Luminol was purchased from Fluka (Buchs, Switzerland). Hydrogen peroxide (H2O2) was provided by Taopu Chemicals (Shanghai, China). All other chemicals used in this work were of analytical grade. Water was purified with a Milli-Q Plus 185 apparatus from Millipore (Bedford, MA) and used throughout the work. Buffer solution for washing and activating of MBs was 0.05 M MES solution (pH 6.0). Buffer solution for incubating immune complexes was 0.01 M phosphate solution containing 0.1 M NaCl (PBS; pH 7.4). Buffer solution for CL reaction was 0.01 M phosphate solution containing 5.0 × 10−4 M luminol and 1.0 × 10−3 M H2O2 (pH 9.0). Preparation of Antibody-Immobilized MBs. The HRPanti-IgG was immobilized on the MBs according to the procedure described by Bi et al.20 with a slight modification. Briefly, a suspension of MBs (200 μL), in a 5 mL Eppendorf tube was separated magnetically. The MBs were washed three times with MES buffer (3 × 300 μL) and then suspended to a final volume of 200 μL in the same buffer solution. A 0.2 M NHS solution (200 μL) and a 0.8 M EDC solution (200 μL) were added to the Eppendorf tube, and the mixture was incubated at room temperature for 30 min to activate the carboxylate groups on the MBs. The MBs were then washed three times with the PBS (3 × 300 μL) and resuspended to a final volume of 2 mL. An aqueous solution of the HRP-anti-IgG (200 μL, 6.25 × 10−7 M) was then added, and the resulting suspensions were allowed to stand for 24 h at 25 °C for the immobilization of the HRPanti-IgG on the surface of the activated MBs. Finally, the resulting MB−HRP-anti-IgG conjugates were separated magnetically and resuspended in 400 μL of PBS containing 1.0% BSA. The solution was incubated at 25 °C for 2 h to eliminate the risk of unspecific binding and separated magnetically. Then, MB−HRP-anti-IgG conjugate was washed three times with

■

RESULTS AND DISCUSSION Principle of CRET Immunosensor. The luminol−H2O2 reaction catalyzed by HRP is one of the most sensitive CL reactions and has been widely used in CL analysis. Therefore, this CL system was chosen as the CRET donor. FITC is a highly fluorescent compound and a widely used fluorescent label in immunoassays. It was noted that the CL spectrum of luminol was largely overlapped with the absorption spectrum of FITC, as shown in Figure.1, which indicated that they were

Figure 1. CL spectrum of luminol (1) and UV−visible absorption spectrum of FITC (2).

likely a nice pair of CRET donor/acceptor. Thus, FITC was selected as the acceptor to label the IgG antigen. The principle of a CRET immunosensor is illustrated in Scheme 1. The carboxyl-modified MBs were conjugated with HRP-anti-IgG and incubated with a limited amount of FITC-IgG to immobilize the antibody−antigen immunocomplex on the surface of the MBs. In the presence of luminol and H2O2, HRPanti-IgG catalyzes the oxidation of luminol by H2O2 to yield 2709

dx.doi.org/10.1021/ac202959d | Anal. Chem. 2012, 84, 2708−2712

Analytical Chemistry

Article

Scheme 1. The schematic illustration of the determination of human IgG based on CRET immunosensor for preliminary immunoassay (A), and competitive immunoassay (B)

the dynamic range. Therefore, R can be used for the quantification of human IgG. Characterization of Antibody-MBs Conjugate and Antigen−antibody-Mbs Conjugate. MBs which are biocompatible, can effectively conjugate biological molecules on their surface and provide convenient methods for the detection of biomolecules in biological samples. To confirm the formation of the conjugates, MB−HRP-anti-IgG conjugate was characterized by CL spectrum. As seen in Figure S1 of the Supporting Information, the MB−HRP-anti-IgG conjugate shows a maximal emission at 425 nm in the presence of luminol and H2O2. This indicates that HRP-anti-IgG has been immobilized on the surface of the MBs. A UV−vis absorption spectrum was also performed to characterize the formation of the antibody−MBs conjugate and antigen−antibody−MBs conjugate. As shown in Figure S2 of the Supporting Information, an absorption peak at 280 nm appeared in curves b and c that corresponds to the typical protein absorption peak, indicating the HRP-anti-IgG has been immobilized on the MBs, whereas a 495 nm absorption peak appeared for FITC-IgG (curve c), demonstrating the formation of the antigen−antibody−MBs conjugate. The stability of the antibody−MBs conjugate and antigen− antibody−MBs conjugate was assessed. A suspension of antibody−MBs conjugate or antigen−antibody−MBs conjugate was stored at 4 °C for 7 days, and the suspension was then again characterized. It was found that their UV−vis absorption spectra did not obviously change. Optimization of Detection Conditions. In generally, the antigen−antibody immunoreactions were performed in the PBS,20 and the immune complexes were stable in PBS. Therefore, we chose 0.01 M PBS (pH 7.4) as the reaction medium of antigen and antibody. After the antigen and antibody binding and magnetic separation, the immune complexes were resuspended in PBS. To ensure the stability of the immune complexes and maximize the sensitivity of the CL detection, a 0.01 M phosphate solution (pH 9.0) was chosen as the CL buffer solution. As the chemiluminescent substrate, the concentrations of luminol strongly influenced the luminescence intensity ratios (R = I425/I525), which were used to measure human IgG. As shown in Figure S3 of the Supporting Information, with and

CL. The close proximity between the CL light source and the FITC-IgG in immune complexes on the surface of MBs lead to CRET to FITC. For a competitive immunoassay, human IgG competes with the FITC-IgG that is added at a fixed amount, and two complexes, that is, HRP-anti-IgG−FITC-IgG, and HRP-antiIgG−IgG, are formed; however, CRET occurs only in the HRP-anti-IgG−FITC-IgG complex, which leads to a CL intensity of luminol at 425 nm increasing and that of FTIC at 525 nm decreasing, respectively (Figure 2). It was also noted

Figure 2. Luminescence spectra of CRET immunosensor in the presence of different concentrations of IgG. The concentrations of IgG were 0 (a), 0.5 (b), 1 (c),1.5 (d), 2 (e), 2.5 (f), 3 (g), 3.5 (h), and 4 nM (i). Inset: plot of the luminescence intensity ratios (R = I25/I525). The concentration of luminol was 5.0 × 10−4 M; the concentration of H2O2 was 1.0 × 10−3 M; the concentration of HRP-anti-IgG was 5.0 × 10−10 M; the concentration of FITC-IgG was 5.0 × 10−11 M.

that the increase in the 425 nm signal and decrease in the 525 nm signal do not exhibited good linearity with the increase of human IgG concentrations, which may be due to slow recovery of CL for the luminol donor when the acceptor at very low concentrations were released, whereas when the acceptor at higher concentrations was released, the releasing amount of acceptor is close to saturation, resulting in a nonlinear regression of CL for FTIC acceptor. It was found that the CL intensity ratio (R = I425/I525) showed very good linearity in 2710

dx.doi.org/10.1021/ac202959d | Anal. Chem. 2012, 84, 2708−2712

Analytical Chemistry

Article

increase in the concentration of luminol from 1.0 × 10−4 M to 5.0 × 10−4 M, the luminescence intensity ratios increased gradually and reached their maximum value when the luminol was 5.0 × 10−4 M, then, with the a further increase in the luminol concentration, the luminescence intensity ratios decreased gradually. Therefore, 5.0 × 10−4 M luminol was used for subsequent studies. The concentrations of H2O2 also impacted the luminescence intensity ratios. As shown in Figure S4 of the Supporting Information, the luminescence intensity ratios increased gradually with the increasing concentration of H2O2 from 5.0 × 10−4 M to 1.0 × 10−3 M, and R reached its maximum value when the H2O2 was 1.0 × 10−3 M. Thus, 1.0 × 10−3 M H2O2 was used for the optimum. IgG Detection with CRET Immunosensor. We developed one CL biosensing platform for human IgG detection using the CRET immunosensor. Figure 2 shows the luminescence spectra of the system in the presence of different concentrations of IgG. As can be seen, the luminescence intensity ratios (R = I425/I525) increased with an increase in the human IgG concentrations. To quantify the IgG, eight point calibration curves were prepared by analyzing a series of IgG standard solutions at different concentrations. The analysis for each point was performed three times, and the mean was used to plot the working curve. As can be seen in Figure 2, the calibration curve showed a good linear relationship between the luminescence intensity ratios and IgG concentration in the range of 0.2−4.0 nM (inset in Figure 2), with a correlation coefficient of 0.9965. The relative standard deviations (RSD) for the measurement of each data point were less than 5.0%. The limit of detection (LOD) at 3σ (σ = S0/S; S0, standard deviation of blank sample; S, the slope of the calibration curve) was 2.9 × 10−11 M. The LOD is lower than those of many reported immunoassays methods for IgG, such as time-resolved fluoroimmunoassay (5 μg/L),21 flow immunoassay (5.0 × 10−7 M),22 electrochemical immunoassay (1.0 × 10−10 M),23 and luminescence resonance energy transfer (LRET)-based immunoassay (0.88 μg/mL).24 Assay reproducibility was investigated by analyzing a 2.0 nM IgG standard solution nine times. The results showed that the RSD was 4.1%. Specificity of CRET Immunosensor. To illustrate whether the strategy is applicable to the detection of human IgG in a biological sample, BSA and human IgM were used to test the selectivity of the CRET immunosensor. In a typical experiment, the CRET immunosensor was incubated with 2.5 nM BSA, human IgM, human IgG in PBS, and human IgG in BSA, respectively, and their luminescence intensities at 425 and 525 nM were measured. As shown in Figure 3, there was almost no change in the luminescence intensity ratios for BSA and human IgM compared with background buffer solution, but 2.5 nM human IgG in PBS and BSA led to an evident change of the luminescence intensity ratios. These phenomena indicated that the proposed strategy had good selectivity in human IgG detection and could distinguish IgG in complex samples from its analogues. The good selectivity for the CRET immunosensor is attributed to the high specificity of the antigen−antibody immune response and the magnetic separation for purification. Analysis of Human Sera Samples. The sera samples from five healthy volunteers were analyzed to demonstrate the feasibility of the proposed CRET immunosensor. Table 1 presents the estimated concentrations of IgG in these samples. It was shown that the concentrations of human IgG in five human serum samples were found from 3.64 × 10−5 M to 6.82 × 10−5 M,

Figure 3. Specificity of the CRET immunosensor. The concentration of analytes was 2.5 nM. The concentration of luminol was 5.0 × 10−4 M; the concentration of H2O2 was 1.0 × 10−3 M; the concentration of HRP-anti-IgG was 5.0 × 10−10 M; the concentration of FITC-IgG was 5.0 × 10−11 M.

Table 1. Detection Results of IgG in Human Sera Using CRET Immunosensor found in total found in sample diluted RSD added found recovery original number sample (nM) (%,n = 4) (nM) (nM) (%, n = 4) sample (μM) 1 2 3 4 5

2.12 3.24 1.98 3.41 1.82

3.8 3.2 4.6 3.4 4.2

0.5 0.5 0.5 0.5 0.5

2.64 3.72 2.51 3.89 2.31

104 96.0 106 96.0 98.0

42.4 64.8 39.6 68.2 36.4

which was consistent with that (from 1.9 × 10−5 M to 26.6 × 10−5 M) in the literature.22,25 Recoveries of IgG from sera samples were also studied. IgG was spiked to these sera samples at 5.0 nM, and the samples were analyzed again. Recoveries were found to be in the range of 96.0−106%. In addition, the contents of human IgG in eight diluted human sera samples were determined by the proposed method (X) and a standard ezyme-linked immunosorbent assay (ELISA) method (Y), respectively. The detection results shows that two methods coincided well, with a regression equation of Y = 1.0213X + 0.0879 and a correlation coefficient of 0.9993 (Figure S5 in the Supporting Information). These results illustrate that the prepared CRET immunosensor exhibits great promise as a reliable technique for the detection of IgG in human sera samples. It was also noted that the detection limits and linear range of the present method are 4 orders of magnitude lower than the endogenous concentrations in the human sera samples analyzed. The discrepancy can cause an error in measurements because of large sample dilution, but as a proof of concept, the analysis results have indicated that the proposed assay is effective for detecting human IgG.

■

CONCLUSIONS A CRET immunosensor has been constructed on the basis of a HRP-anti-IgG−FITC-IgG immune complex on the surface of MBs. Human IgG was selected as a model analyte to demonstrate the generality of the strategy. A MB−HRP-antiIgG/FITC-IgG conjugate was proved to be an efficient CRET and magnetic separation protocol for the detection of human IgG and the purification of samples. The proposed CRET immunosensor has several remarkable advantages, including facile preparation without use of a complicated chemical procedure and a high specificity and sensitivity to target species. Furthermore, it is a novel sensing platform for biomolecule 2711

dx.doi.org/10.1021/ac202959d | Anal. Chem. 2012, 84, 2708−2712

Analytical Chemistry

Article

(21) Niu, C. G.; Liu, J.; Qin, P. Z.; Zeng, G. M.; Ruan, M.; He, H. Anal. Biochem. 2011, 409, 244−248. (22) Tang, D.; Niessner, R.; Knopp, D. Biosens. Bioelectron. 2009, 24, 2125−2130. (23) Zhong, Z.; Li, M.; Xiang, D.; Dai, N.; Qing, Y.; Wang, D.; Tang, D. Biosens. Bioelectron. 2009, 24, 2246−2249. (24) Wang, M.; Hou, W.; Mi, C. C.; Wang, W. X.; Xu, Z. R.; Teng, H. H.; Mao, C. B.; Xu, S. K. Anal. Chem. 2009, 81, 8783−8789. (25) Liu, Y. M.; Mei, L.; Liu, L. J.; Peng, L. F.; Chen, Y. H.; Ren, S. W. Anal. Chem. 2011, 83, 1137−1143.

detection based on CRET and can be readily extended to develop a variety of analysis methods for detection of other biomolecules by using corresponding antigens and respective antibodies. Nevertheless, this study involves an exogenous enzyme (i.e., HRP), which may complicate the assay by disturbing the biological interactions under study. Therefore, further studies are being undertaken by employing a nonenzymatic CRET system to construct a new CRET immunosensor.

■

ASSOCIATED CONTENT

S Supporting Information *

Supplementary figures for the characterization of the antibody− MBs conjugate and antigen−antibody−MBs conjugate and optimization of detection conditions. This material is available free of charge via the Internet at http://pubs.acs.org.

■

AUTHOR INFORMATION

Corresponding Author

*Phone: +86 773 5856104. Fax: +86 773 5832294. E-mail: [email protected]. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS This work was supported by the National Natural Science Foundations of China (Nos. 20875019, 21175030) and the Natural Science Foundations of Guangxi Province (No. 2010GXNSFF013001).

■

REFERENCES

(1) Michalet, X.; Weiss, S.; Jaeger, M. Chem. Rev. 2006, 106, 1785− 1813. (2) Roy, R.; Hohng, S.; Ha, T. Nat. Methods 2008, 5, 507−516. (3) Royer, C. A. Chem. Rev. 2006, 106, 1769−1784. (4) Schuler, B.; Lipman, E. A.; Eaton, W. A. Nature 2002, 419, 743− 747. (5) Dehmelt, L.; Bastiaens, P. I. H. Nat. Rev. Mol. Cell Biol. 2010, 11, 440−452. (6) Liu, L.; Shao, M.; Dong, X.; Yu, X.; Liu, Z.; He, Z.; Wang, Q. Anal. Chem. 2008, 80, 7735−7741. (7) James, J. R.; Oliveira, M. I.; Carmo, A. M.; Iaboni, A.; Davis, S. J. Nat. Methods 2006, 3, 101−106. (8) Pfleger, K. D. G.; Eidne, K. A. Nat. Methods 2006, 3, 165−174. (9) Yao, H.; Zhang, Y.; Xiao, F.; Xia, Z.; Rao, J. Angew. Chem., Int. Ed. 2007, 46, 4346−4349. (10) Huang, X.; Li, L.; Qian, H.; Dong, C.; Ren, J. Angew. Chem., Int. Ed. 2006, 45, 5140−5143. (11) Zhang, S.; Yan, Y.; Bi, S. Anal. Chem. 2009, 81, 8695−8701. (12) Huang, X.; Ren, J. Anal. Chim. Acta 2011, 686, 115−120. (13) Freeman, R.; Liu, X.; Willner, I. J. Am. Chem. Soc. 2011, 133, 11597−11604. (14) Liu, X.; Freeman, R.; Golub, E.; Willner, I. ACS Nano 2011, 5, 7648−7655. (15) Fan, A.; Lau, C.; Lu, J. Anal. Chem. 2005, 77, 3238−3242. (16) Wang, J.; Liu, G.; Munge, B.; Lin, L.; Zhu, Q. Angew. Chem., Int. Ed. 2004, 43, 2158−2161. (17) Jianzhong, L.; Choiwan, L.; Kai, M. Chem.Commun. 2003, 2888−2889. (18) Wang, J.; Kawde, A. N.; Erdem, A.; Salazar, M. Analyst 2001, 126, 2020−2024. (19) Zhou, X.; Xing, D.; Zhu, D.; Jia, L. Anal. Chem. 2009, 81, 255− 261. (20) Bi, S.; Yan, Y.; Yang, X.; Zhang, S. Chem.Eur. J. 2009, 15, 4704−4709. 2712

dx.doi.org/10.1021/ac202959d | Anal. Chem. 2012, 84, 2708−2712