TECHNICAL NOTE pubs.acs.org/ac
Nanotube-Based Colorimetric Probe for Ultrasensitive Detection of Ataxia Telangiectasia Mutated Protein Qingzhi Zhang,†,‡ Bin Zhao,‡ Juan Yan,‡ Shiping Song,*,‡ Rui Min,*,† and Chunhai Fan‡ † ‡
Department of Naval Medicine, Second Military Medical University, Shanghai, 200433, China Laboratory of Physical Biology, Shanghai Institute of Applied Physics, Chinese Academy of Science, Shanghai 201800, China ABSTRACT:
We have developed a nanotube-based colorimetric probe using multiwalled carbon nanotubes (MWNTs), anti-immunoglobulin G (anti-IgG), and horseradish peroxidase (HRP). The probe was used as an alternative to conventional colorimetric conjugates to obtain amplified signals in a sandwich-type immunoassay for ataxia telangiectasia mutated (ATM), a potential biomarker for radiation doses and cancers. Results show that the MWNT-based probe colorimetry was 5000 times more sensitive than a conventional ELISA, while its concentration range was 10 000 times wider than that of the latter. Its limit of detection (LOD) was 0.2 fg/mL (54 aM, ∼32 molecules in 1 μL samples). Control experiments showed that detection of ATM molecules at the picogram-level could still be achieved in samples that contained protein makers present at more than 100 times the ATM concentration, demonstrating the high specificity of the technique. The MWNT-based probe also has the potential to become a universal probe for colorimetric assays of most protein markers because it can recognize the associated rabbit polyclonal antibodies.
A
taxia telangiectasia mutated (ATM) is a protein kinase that is recruited and activated by DNA damage, especially doublestrand breaks, which result from intrinsic factors such as errors during replication or extrinsic factors such as ionizing radiation. ATM plays an important role in cell cycle delay after DNA damage by acting as a sensor protein to transduce the signal of the damage. In ataxia telangiectasia (AT) disease, or AT-like disorders, the kinase activity of ATM proteins increases rapidly immediately following DNA damage.1 Because these types of diseases have extreme cellular sensitivity to radiation and a predisposition to cancer, ATM has been considered a potential biomarker for monitoring radiation doses and detecting some cancers.24 Usually, immunofluorescence methods and traditional ELISA are used to detect ATM activation. However, these techniques cannot meet the needs for higher detection sensitivity as earlier diagnosis becomes more and more important. Recently, nanomaterial-based bioprobes have been widely studied to improve the sensitivity of biodetection.5,6 Typically, the effect of high surface-to-volume ratio of nanomaterials was involved in developing nanobased bioprobes such that one unit may accommodate many biomolecules such as DNA, antibodies, and signaling enzymes, through high surface density.6 The highly r 2011 American Chemical Society
dense signaling enzymes at the surface of the nanomaterials can offer more opportunities to amplify detection signals. Among the nanomaterials tested, gold nanoparticles (AuNPs) were the most studied and were used as carriers to design multicomponent and multifunctional nanoprobes.7,8 For example, Li et al. developed an enzyme-based multicomponent colorimetric gold nanoprobe that integrated DNA recognition and signal amplification.9 The nanoprobes were prepared by coassembling a thiolated DNA detection probe and horseradish peroxidase at the surface of AuNPs. This nanoprobe was employed in a magnetic particlebased sandwich-type assay, providing colorimetric signal readout arising from enzymatic catalysis. After the signal amplification, the color contrast between the 100 pM target DNA sample and the blank control could be easily distinguished even with the naked eye. However, the amplification that could be achieved is limited because of the limited size of the AuNPs, especially when enzymes and large biomolecules such as immunoglobulins need to be coassembled on one nanoparticle. Received: September 6, 2011 Accepted: October 27, 2011 Published: October 27, 2011 9191
dx.doi.org/10.1021/ac2023684 | Anal. Chem. 2011, 83, 9191–9196
Analytical Chemistry Alternatively, carbon nanotubes (CNTs) provide an extremely large surface area for biomolecular conjugation and subsequent signal amplification. Wang and co-workers were the first to immobilize both alkaline phosphatases (ALP) and DNA on multiwalled carbon nanotubes (MWNTs), which coupled DNA hybridization with catalytic amplification of ALP.10,11 The nanoprobe was used to develop an electrical biosensor to determine DNA targets. Because of the presence of numerous enzymes on the CNTs, this electrochemical DNA sensor reached a remarkably low detection limit of 54 aM. Rusling and co-workers developed a CNT-based electrochemical platform for immunoassays, which combined single-wall carbon nanotube (SWNT) forest platforms with multilabel antibodynanotube bioconjugates.12 This approach provided a detection limit of 100 fM for prostate specific antigens. Yet, electrochemical bioanalytical methods are still not used very often in practice because of their inability to provide stable biodetection, especially protein detection. For optical biodetection, CNTs should also provide an extremely larger surface area for biomolecular conjugation than nanoparticles, resulting in stable signal amplification. Here, we report a novel CNT-based colorimetric probe and its use for ultrasensitive detection of ATM. Unlike the previous monocomponent CNT-based probes which only had improved binding ability,13,14 our probe is multicomponent, based on a coassembly technique, which provides both improved binding ability and signal-amplification ability. Also, comparing with the assembly of bioconjugates such as enzyme-linked antibody or enzyme-linked DNA to nanomaterials,1517 coassembly could avoid chemical conjugation between biomolecules, keeping their bioactivity well. To our knowledge, this is the first example of using a CNT-based multicomponent probe for colorimetric detection of protein markers. The nanoprobe allows detection of ATM down to 54 aM (∼32 molecules in 1 μL samples). This is the lowest detection limit reported thus far for the detection of ATM. The MWNT-based probe colorimetry was 5000 times more sensitive than a conventional ELISA. Also, its concentration range was 10 000 times wider than that of the latter. Importantly, the nanoprobe was prepared by consistent loading of enzymes and goat-antirabbit IgG (secondary antibody, instead of a primary antibody) on multiwalled carbon nanotubes (MWNTs). As this secondary antibody can react with rabbit polyclonal antibodies for any antigen, the probe has the potential to be used for sandwich-based colorimetric assays of any target protein by using unlabeled rabbit antibody as the detection antibody.
’ EXPERIMENTAL SECTION Reagents and Materials. MWNTs were purchased from Shenzhen Nanotech Port Co. Ltd. (NTP, China) and pretreated in our own lab. Horseradish peroxidase (HRP), lyophilized 99% bovine serum albumin (BSA), Tween 20, and anti-immunoglobulin G labeled with HRP (anti-IgG-HRP) were purchased from Sigma-Aldrich (St. Louis, MO). TMB substrate (TMB = 3,30 ,5,50 tetramethylbenzidine; Neogen K-blue low activity substrate) was purchased from Neogen. ATM polyclonal antibody, ATM monoclonal antibody, and ATM antigen were purchased from R&D Systems (Minneapolis, MN). Goat antirabbit immunoglobulin G (anti-IgG), goat antirabbit IgG (Poly-HRP40), prostatespecific antigen (PSA), and alpha-fetoprotein (AFP) were purchased from Fitzgerald (Acton, MA). Antigens and antibodies were dissolved in pH 7.0 phosphate buffered saline (PBS) (0.01 M in phosphate, 0.14 M NaCl, 2.7 mM KCl) unless otherwise
TECHNICAL NOTE
noted. 1-(3-(Dimethylamino)-propyl)-3-ethylcarbodiimide hydrochloride (EDC) and N-hydroxysulfosuccinimide (NHS) were purchased from Sigma and dissolved in pH 6.0 MES (2-(Nmorpholino) ethanesulfonic acid) buffer immediately before use. Preparation of MWNT-Based Colorimetric Probes. A concentration of 1 mg/mL of the pretreated MWNT was mixed with 250 μL of 400 mM EDC and 250 μL of 100 mM NHS in pH 6.0 MES buffer. The mixture vortexed at room temperature for 15 min. The resulting mixture was then centrifuged at 15 000 rpm at 4 °C for 10 min and the supernatant was discarded. The washing was repeated to remove excessive EDC and NHS. To the mixture 1 mg/mL of both anti-IgG and HRP were added and stirred in a small vial overnight at room temperature. The reaction mixture was then centrifuged at 15 000 rpm at 4 °C for 10 min and the supernatant was removed. To the solid conjugate remaining in the vial, 1 mL of PBST (phosphate buffered saline with Tween 20) buffer was added. After mixing well and centrifuging at 15 000 rpm at 4 °C for 10 min, the supernatant was discarded to remove any free HRP and anti-IgG. The process was repeated four times. To the precipitate collected, 200 μL of 2% casein was added and vortexed to form a homogeneous dispersion. The solution was stored at 4 °C and diluted immediately before use with PBS buffer containing 2% casein. Optimization and Characterization of MWNT-Based Colorimetric Probes. The amount and ratio of the anti-IgG and HRP are very important to the preparation of the probes and subsequent assays. The optimized conditions for probe preparation should be obtained. To optimize the amount of anti-IgG used in probe preparation, the HRP/IgG concentration ratio was fixed at 200/1, and 5, 10, 20, 30, 40, 50, 60, or 70 μL of the HRP solution was, respectively, added to 50 μL of a MWNT solution. The optimized anti-IgG amount then fixed. Next, the HRP/IgG ratios were changed to 30/1, 40/1, 50/1, 100/1, or 200/1 to obtain the optimized ratio of the HRP to anti-IgG in probe preparation. The probes were evaluated using colorimetric tests with a rabbit IgG coated 96-well plate. Carboxylated MWNTs, before and after bioconjugation with HRP and anti-IgG, were observed using a scanning electron microscope (SEM). Samples, diluted 20-fold in PBS, were dropped on the silicon base. After the sample was dried at 37 °C, the silicon base was washed with milli-Q water. This process was repeated three times. The samples were then imaged on a SEM. Detection of ATM with MWNT-Based Colorimetric Probes. Capture antibody was diluted to a working concentration (10 μg/mL) in PBS. The solution was immediately used to coat a 96-well microplate by transferring 100 μL to each well. The plate was covered and incubated overnight at room temperature. Each well was aspirated and washed with washing buffer two times. Then the plate was blocked by adding 300 μL of blocking buffer (1% BSA in PBS, pH 7.4) to each well. The plate was incubated at 37 °C for 1 h and then washed three times. To each well, 100 μL of ATM was added at different concentrations (100 fg/mL to 200 ng/mL). The plate was then covered with a plate sealer and incubated for 1 h at 37 °C. The washing steps were repeated three times. Next, 100 μL of ATM polyclonal antibodies of ATM were added to each well. The plate was then covered and incubated. The plate was then washed again, and 100 μL of anti-IgG-HRP conjugates or anti-IgG-MWNT-HRP (colorimetric probe) were added to each well. The plate was again covered with and incubated at 37 °C for 1 h. To each well, 100 μL of TMB is added and the plate was incubated at room temperature for 20 min. Finally, 100 μL of stop solution was added to each well to 9192
dx.doi.org/10.1021/ac2023684 |Anal. Chem. 2011, 83, 9191–9196
Analytical Chemistry
TECHNICAL NOTE
Figure 1. Scheme of the colorimetric strategy for the detection of ATM using a MWNT-based probe compared with conventional ELISA. A “sandwich” configuration was used to capture and detect ATM. In a 96-well plate, the target molecule first binds to the capture antibody followed by the binding of the detection antibody to form an immunocomplex. Then, the MWNT-based probe or anti-IgG-HRP was added to bind to the detection antibody. In the presence of TMB, HRP catalyzes the substrate to give colorimetric signals.
stop the reaction. The optical density of each well was determined immediately using a microplate reader set to 450 nm.
’ RESULTS AND DISCUSSION MWNT-Based Probe and Colorimetric Strategy. In this study, a probe bearing multiple HRP labels attached to a MWNT was developed for multilabel amplification in a colorimetric assay. The MWNT-based probe was used to replace the conventional anti-IgG-HRP conjugate. Anti-IgG and HRP were covalently bound to carboxylated MWNTs in an optimized molar ratio using a two-step process. To achieve this, carboxylic acid groups on MWNT were first activated by EDC and NHS. The redundant chemical regents were then removed. The activated MWNT was then reacted with the amine residues on proteins (anti-IgG and HRP). The MWNT-based probes can be stored ready for use. For assays, a “sandwich” configuration was used (Figure 1), which involved a capture antibody and a detection antibody to form a “sandwich” type complex with an antigen (target). Then, the MWNT-based probe was used to bind to detection antibodies and produce optical signals. As a control experiment, anti-IgG-HRP conjugate was used to bind to detection antibodies instead of the MWNT-based probe. Optimization and Characterization. The process for preparation of the MWNT-based probes was optimized to achieve the best signal amplification efficiency in colorimetric assays (Figure 2). We studied the HRP/MWNT ratio by mixing different amounts of HPR with pretreated MWNT solution. Because MWNTs are size-inhomogeneous inherent and some of them were lost during pretreatment, it was difficult to estimate the molar ratio of HRP to MWNT. Instead, a volume ratio was used while the concentration of HRP was fixed. We found that a ratio of 8/10 produced the highest signal amplification efficiency for the HRP-MWNT conjugates (Figure 2A). The results show that high HRP/MWNT ratios did not give high signal amplification
efficiency because the linking efficiency of HRP and MWNT decreased with the increase in mixture volume. After the HRP became saturated in the mixing solution, the addition of more HRP solution decreased the concentration of MWNTs because the MWNTs were diluted. Figure 2B demonstrates that a molar ratio of 1/50 for HRP/ anti-IgG resulted in the highest signal amplification efficiency for the HRP-MWNT-anti-IgG conjugates (i.e., the final probes). The results indicate that the signal amplification efficiency decreased when the molar ratio was greater or less than 1/50. When the molar ratio was greater than 1/50, more anti-IgG molecules were linked to a MWNT, decreasing the amount of HRP that could be bind. Decreased binding of HRP resulted in less efficient catalysis of the MWNT-based probes to TMB. When the molar ratio was less than 1/50, fewer anti-IgG molecules were bound to a MWNT, which decreased the ability of the MWNT-based probes to bind to the rabbit IgG molecules (i.e., detection antibody). The result was that the signal amplification effects of the MWNTbased probes would not work well due to a decrease in immunocomplex formation. Several experimental steps are important to keep the MWNTbased probes stable. First, MWNTs were sonicated for a sufficient amount of time before activation to ensure their dispersion in solution. The better dispersed MWNTs resulted in better dispersed MWNT-based probes. Second, the redundant chemical reagents needed to activate MWNTs were completely removed before binding the assembly proteins to avoid protein cross-linking. Without this step, the carboxylic acid groups of proteins could be activated by the EDC and NHS. The activated carboxylic acid groups could then link to the free amido groups of the proteins themselves, which results in crossing linking. If there is remnant EDC, we find that anti-IgG, HRP and the MWNTs were inclined to aggregate and form bundles, which made the conjugates very unstable and decreased their performance. Third, efficient 9193
dx.doi.org/10.1021/ac2023684 |Anal. Chem. 2011, 83, 9191–9196
Analytical Chemistry
TECHNICAL NOTE
Figure 2. Optimization for the preparation of MWNT-based probes. (A) Different amounts of HPR were mixed with pretreated MWNT solution. The volume ratio (HRP/MWNT) was as follows: 1/10, 2/10, 6/10, 8/10, 10/10, 12/10, and 14/10. The obtained conjugates were evaluated with simple colorimetric tests; (B) anti-IgG and HRP with different molar ratios were mixed with pretreated MWNT solution while the concentration of HRP was fixed. The molar ratio (anti-IgG/HRP) was as follows: 1/30, 1/40, 1/50, 1/100, and 1/200. The obtained probes were used in colorimetric tests with a rabbit IgG-coated 96-well plate.
blocking was performed to prevent the MWNT-based probes from depositing on the bottom or attaching to the wall of the vials. In this work, the assembled probes were blocked in blocking solution overnight, while it takes no more than 1 h to do a conventional blocking. SEM imaging was used to characterize the MWNT-based probes. Figure 3 shows SEM images of MWNTs before and after linking with HRP and anti-IgG. Before protein linking, an average MWNTs diameter of about 44 nm was measured (Figure 3A). After protein linking, an average diameter of about 73 nm was measured (Figure 3B). This indicated that the MWNTs did bind to the protein and that the thickness of the protein shell was about 15 nm. On the basis of the size of the antiIgG molecules, the shell should be a monolayer. Detection of ATM. The MWNT-based probe was used to detect ATM by incorporating it into a sandwich-type immunoassay. Figure 4A shows the comparison between MWNT-based probe colorimetry and conventional ELISA with anti-IgG-HRP conjugates. The former had several remarkable advantages over the later. First, MWNT-based probe colorimetry could detect at least 0.2 pg/mL ATM (54 aM) while the conventional ELISA could only detect 1 ng/mL ATM. The high sensitivity should be attributed to the high signal amplification efficiency of the
Figure 3. SEM images of MWNTs and MWNT-based probes. (A) Images of MWNTs before linking with HRP and anti-IgG. Insert: isolated MWNT with the measurement of its diameter; (B) images of MWNT-based probes. Insert: isolated HRP-MWNT-anti-IgG conjugate with the measurement of its diameter.
MWNT-based probe which has improved binding ability and increased signaling ability. Anti-IgG and HRP were assembled onto the surface of the MWNT to form a monolayer (Figure 3). On the basis of the size of anti-IgG (∼15 nm in diameter) and HRP (∼6 nm diameter), there should be approximately 40 antiIgG molecules and more than 2000 HRP molecules on a 500 nmlong MWNT. More importantly, the combination of both capture and catalysis capabilities increases the signal amplification efficiency of the probes. The results show that our novel method was 5 000 times more sensitive than the conventional technique, although the slope of the regression lines indicates that the later had better accuracy than the former. Second, we found that MWNT-based probe colorimetry has a wider range of detectable concentrations (0.2100 000 pg/mL) than conventional ELISA (150 ng/mL), indicating that our probes are able 9194
dx.doi.org/10.1021/ac2023684 |Anal. Chem. 2011, 83, 9191–9196
Analytical Chemistry
TECHNICAL NOTE
were used herein to act as negative controls. Control experiments showed that 10 pg/mL ATM could be detected very clearly, while the signal resulting from 1 ng/mL PSA and AFP showed no significant difference to a blank control sample. This result indicated that ATM molecules could be detected even in the presence of interfering protein makers present with a more than 100-fold greater concentration in samples. The good specificity of the MWNT-based probe indicated that the affinity of anti-IgG was intact after its assembly onto the MWNTs because the monolayer binding retained its 3D structure (Figure 3). The standard deviations in Figure 4 also indicate that the MWNTbased probe colorimetry has good reproducibility.
Figure 4. Colorimetric assay for ATM with MWNT-based probes. (A) The comparison of sensitivity and concentration range between MWNT-based probe colorimetry and conventional ELISA with antiIgG-HRP conjugates and Poly-HRP. The concentration of ATM ranged from 0.2 pg/mL to 100 ng/mL in the MWNT-based probe colorimetry. The concentration of ATM ranged from 50 pg/mL to 100 ng/mL in the ELISA with Poly-HRP. The concentration of ATM ranged from 1 ng/ mL to 50 ng/mL in the conventional ELISA with anti-IgG-HRP conjugates. Insert: images of the ATM standard samples at different concentrations in the MWNT-based probe colorimetry. (B) The specificity of MWNT-based probe colorimetry. ATM (10 pg/mL), PSA (1 ng/mL), and AFP (1 ng/mL) were detected with MWNT-based probe colorimetry. The later two proteins acted as negative controls. A PBS buffer without any antigens was used as a blank control. Insert: images of the different samples used for the MWNT-based probe colorimetry.
to detect ATM in most samples without any dilution. This might be attributed to the improved binding ability of the MWNT-based probe (i.e., many antibodies on one MWNT). The improved binding ability makes it possible to recognize and capture targets at a wider concentration range. Third, the dose response curve of MWNT-based probe colorimetry was fitted better than that of the conventional ELISA. The R-square of the linear curve fit for the former was 0.99, while that for the latter was 0.98. The P value of the linear curve fit for the former was less that 0.0001, while that for the latter was less than 0.0012. These results show that the dose response with MWNT-based probe colorimetry was very reasonable even at a wide concentration range. Of note, the MWNTbased probe demonstrated better signal amplification than PolyHRP (Figure 4A). This result can be attributed to the high loading of anti-IgG and HRP on each probe. Our MWNT-based probe colorimetry also demonstrated good specificity in the detection of ATM (Figure 4B). PSA and AFP, known as certified cancer markers in clinical diagnostics,
’ CONCLUSIONS We have demonstrated the fabrication of MWNT-based probes and their use for a highly sensitive and selective colorimetric assay for ATM. The results demonstrate that our MWNT-based probe colorimetry had both high sensitivity and a wide detectable concentration range. The detection limit of 54 aM (∼32 molecules in 1 μL samples) surpassed that of a conventional ELISA and all known commercial techniques for the detection of ATM. Importantly, the MWNT-based probe has the potential to become universal for the optical detection of most protein markers because the appropriate rabbit polyclonal antibodies could be recognized by the novel probe. Also, other enzymes such as alkaline phosphatases could be coassembled on MWNTs along with anti-IgG, instead of HRP, to produce luminescence signal. Besides a microplate-based immunoassay system, our CNT probes could also be used in a magnetic affinity immunoassay system. We believe that such probes will become an available tool for molecular diagnostics and proteomics in the future. ’ AUTHOR INFORMATION Corresponding Author
*E-mail:
[email protected] (S.S.);
[email protected] (R.M.).
’ ACKNOWLEDGMENT This work was supported by the National Science Foundation (Grants 20725516, 30970874, 90913014, and 91123037) and the Ministry of Science and Technology (Grants 2008AA06A406 and 2012CB932600). ’ REFERENCES (1) Lee, J. H.; Paull, T. T. Oncogene 2007, 26, 7741–7748. (2) Barlow, C.; Eckhaus, M. A.; Schaffer, A. A.; Wynshaw-Boris, A. Nat. Genet. 1999, 21, 359–360. (3) Ch’ang, H. J.; Maj, J. G.; Paris, F.; Xing, H. R.; Zhang, J.; Truman, J. P.; Cardon-Cardo, C.; Haimovitz-Friedman, A.; Kolesnick, R.; Fuks, Z. Nat. Med. 2005, 11, 484–490. (4) Kuhne, M.; Riballo, E.; Rief, N.; Rothkamm, K.; Jeggo, P. A.; Lobrich, M. Cancer Res. 2004, 64, 500–508. (5) Giljohann, D. A.; Mirkin, C. A. Nature 2009, 462, 461–464. (6) Song, S. P.; Qin, Y.; He, Y.; Huang, Q.; Fan, C. H.; Chen, H. Y. Chem. Soc. Rev. 2010, 39, 4234–4243. (7) Giljohann, D. A.; Seferos, D. S.; Daniel, W. L.; Massich, M. D.; Patel, P. C.; Mirkin, C. A. Angew. Chem., Int. Ed. 2010, 49, 3280–3294. (8) Sperling, R. A.; Rivera Gil, P.; Zhang, F.; Zanella, M.; Parak, W. J. Chem. Soc. Rev. 2008, 37, 1896–1908. (9) Li, J.; Song, S. P.; Liu, X. F.; Wang, L. H.; Pan, D.; Huang, Q.; Zhao, Y.; Fan, C. H. Adv. Mater. 2008, 20, 497. 9195
dx.doi.org/10.1021/ac2023684 |Anal. Chem. 2011, 83, 9191–9196
Analytical Chemistry
TECHNICAL NOTE
(10) Wang, J.; Liu, G.; Jan, M. R. J. Am. Chem. Soc. 2004, 126, 3010–3011. (11) Munge, B.; Liu, G.; Collins, G.; Wang, J. Anal. Chem. 2005, 77, 4662–4666. (12) Yu, X.; Munge, B.; Patel, V.; Jensen, G.; Bhirde, A.; Gong, J. D.; Kim, S. N.; Gillespie, J.; Gutkind, J. S.; Papadimitrakopoulos, F.; Rusling, J. F. J. Am. Chem. Soc. 2006, 128, 11199–11205. (13) Lynam, C.; Gilmartin, N.; Minett, A. I.; O’Kennedy, R. Carbon 2009, 47, 2337–2343. (14) Yang, M.; Kostov, Y.; Bruck, H. A.; Rasooly, A. Anal. Chem. 2008, 80, 8532–8537. (15) Lee, A. C.; Ye, J. S.; Tan, S. N.; Poenar, D. P.; Sheu, F. S.; Heng, C. K.; Lim, T. M. Nanotechnology 2007, 18, 455102. (16) Ambrosi, A.; Castaneda, M. T.; Killard, A. J.; Smyth, M. R.; Alegret, S.; Merkoci, A. Anal. Chem. 2007, 79, 5232–5240. (17) Ambrosi, A.; Airo, F.; Merkoci, A. Anal. Chem. 2010, 82, 1151–1156.
9196
dx.doi.org/10.1021/ac2023684 |Anal. Chem. 2011, 83, 9191–9196
