ARTICLE pubs.acs.org/ac
Capillary Electrophoresis with Lamp-Based Wavelength-Resolved Fluorescence Detection for the Probing of Protein Conformational Changes Bregje J. de Kort,* Geert A. ten Kate, Gerhardus J. de Jong, and Govert W. Somsen Biomolecular Analysis, Department of Pharmaceutical Sciences, Utrecht University, Utrecht, The Netherlands ABSTRACT: Native protein fluorescence spectra encompass information on protein conformation. In this study, capillary electrophoresis (CE) combined with lamp-based wavelength-resolved fluorescence detection (wrFlu) is presented as a novel tool for the analysis of protein mixtures and the monitoring of protein unfolding. The CE-wrFlu system provides three-dimensional data (time, emission wavelength, intensity) from which electropherograms and accurate emission spectra of separated proteins can be extracted. For model proteins, linear detector responses (peak height vs concentration) were obtained (R2 > 0.96) with detection limits (LODs) in the 632 nM range. The minimum protein concentration required for precise determination of the maximum emission wavelength by CE-wrFlu was about 15 times the LOD. Unfolding of various model proteins was induced by protein incubation and analysis in background electrolyte (BGE) containing 7.0 M urea. CE-wrFlu of the unfolded species revealed peaks with clear red-shifted spectra, which adequately corresponded to reference spectra obtained on a standard spectrophotometer. Moreover, unfolded proteins showed a significant decrease in effective electrophoretic mobility (after correction for BGE viscosity) due to the increase of their molecular hydrodynamic radii. It is concluded that the CE-wrFlu system provides two independent indicators for changes in protein folding and will allow the simultaneous assessment of protein purity and conformation.
F
luorescence is a well-recognized principle in separation sciences that is employed to achieve sensitive and selective detection of compounds of interest. Fluorescence spectroscopy is also an established technique for the study of protein conformation.13 Native protein fluorescence arises from the aromatic amino acid residues phenylalanine, tyrosine, and tryptophan after their excitation in the deep-UV region. Excitation of tryptophan residues provides a useful means to monitor protein conformational changes, as the emission of tryptophan is sensitive to, e.g., solvent polarity and pH. Upon protein unfolding, tryptophan residues that are located in the hydrophobic core of the protein will be exposed to the more polar environment of the solvent (i.e., water). Solvent relaxation processes then cause the maximum emission wavelength of tryptophan residues in proteins to shift to higher wavelengths.1 This so-called red shift is indicative for protein unfolding. Correct assignment of protein conformational changes via tryptophan fluorescence measurements can only be achieved for pure protein solutions. In case of protein mixtures or when a protein sample comprises impurities, degradation products, or other fluorescent compounds, an average emission spectrum is obtained that will hinder appropriate spectral interpretation. Therefore, protein separation prior to wavelength-resolved fluorescence detection (wrFlu) is indicated when protein conformational information has to be derived from mixtures. Capillary electrophoresis (CE) is an attractive tool for the analysis of intact proteins providing efficient and fast separations r 2011 American Chemical Society
and requiring only minute amounts of sample.47 CE separations can be performed in aqueous buffers under mild conditions. Hence, separation conditions that possibly affect protein conformation, like stationary phases or organic solvents as normally used in liquid chromatography, can effectively be avoided. Another interesting aspect of CE is that the electrophoretic mobility of a protein depends on its conformation. Electrophoretic mobility is governed by the molecular charge-to-size ratio. Therefore, protein unfolding—which normally induces an increase of the hydrodynamic radius—will lead to a shift in migration time with respect to the folded (i.e., native) species.811 Combination of CE with wrFlu detection clearly would provide an interesting tool for protein conformational analysis. So far, CE-wrFlu of native fluorescent analytes has only been developed for and applied to the analysis of low-molecularweight compounds.12 For instance, Fuller et al. used CE-wrFlu for the identification of fluorescent aromatic amines, amino acids, and peptides in single neurons,13 and Kok et al. used emission spectra recorded by wrFlu detection in CE to assign naphthalene sulfonates in wastewater samples.14 CE-wrFlu has also been utilized by Timperman et al. to distinguish between tyrosineand tryptophan-containing peptides,15 but CE with wr detection of native protein fluorescence has not been reported before. Received: May 3, 2011 Accepted: June 23, 2011 Published: June 23, 2011 6060
dx.doi.org/10.1021/ac201136y | Anal. Chem. 2011, 83, 6060–6067
Analytical Chemistry Fluorescence detection (Flu) of underivatized proteins is very well possible in CE and has predominantly been carried out using laser excitation.16,17 CE-Flu allows selective and quite sensitive detection of intact proteins, but no spectral information is acquired. In order to avoid the use of expensive UV lasers for protein excitation and to facilitate ease of application, our group recently has described the use of a detector for CE that employs lamp-based excitation for fluorescence detection of native proteins.18 It was demonstrated that with this system sensitivities similar to CE with laser-induced fluorescence detection could be obtained. In a follow-up study, a wrFlu detector for protein CE was designed.19 In this setup a dedicated fluorescence detection cell was combined with a spectrograph equipped with a chargecoupled device (CCD) camera. Optical detection parameters and data handling procedures were evaluated and optimized.19 Our ultimate goal was to develop a CE-wrFlu system that allows efficient and reproducible protein separation and provides sensitive and accurate protein emission data so that reliable information on protein conformation can be obtained. Therefore, we investigated the suitability of the earlier developed CEwrFlu system for protein mixture analysis. Aspects such as separation performance, migration time repeatability, linearity, and detection limits were evaluated. Subsequently, attention was devoted to the accuracy of the recorded protein emission spectra and the reliable determination of the maximum emission wavelength. Finally, the capability of the CE-wrFlu system to monitor protein conformational changes was studied by analyzing various test proteins in a background electrolyte (BGE) containing 7.0 M urea (forced unfolding) and comparing the obtained electrophoretic and spectral data with CE-wrFlu analyses of the same proteins in the absence of urea.
’ MATERIALS AND METHODS Chemicals. L-Tryptophan, bovine serum albumin, carbonic anhydrase II (from bovine erythrocytes), R-chymotrypsinogen A (from bovine pancreas), R-lactalbumin (from bovine milk), β-lactoglobulin A, β-lactoglobulin B (both from bovine milk), lysozyme (from chicken egg white), Polybrene (PB, hexadimethrine bromide, Mw ∼ 15 kDa), dextran sulfate sodium salt (DS, Mw ∼ 500 kDa), and indole were purchased from SigmaAldrich (Steinheim, Germany). Sodium dihydrogen phosphate and phosphoric acid (85%) were obtained from Merck (Darmstadt, Germany), sodium hydroxide was from BUFA Pharmaceutical Products (Uitgeest, The Netherlands), and urea was from J.T. Baker (Deventer, The Netherlands). Protein stock solutions of 1.0 mg/mL were prepared in deionized water, stored at 20 °C, and daily diluted to the required concentration. BGEs were prepared by dissolving 0.3449 g of sodium phosphate in 50.0 mL of deionized water and adjusting the pH to pH 3.0 with phosphoric acid. Urea-containing BGE was prepared by dissolving the desired amount of urea into BGE (pH 3.0) and readjusting the pH with phosphoric acid. To induce protein unfolding, proteins were dissolved in BGE containing 7.0 M urea (pH 3.0). For protein CE, indole was added as an electroosmotic flow (EOF) marker to a final concentration of 0.25 μg/mL (2.1 μM). Solutions of 10% (w/v) PB and 1% (w/v) DS were prepared in deionized water and filtered over a 0.45 μm filter type HA (Millipore, Molsheim, France) prior to use. CE System. CE experiments were carried out on a Beckman P/ ACE MDQ instrument (Beckman Coulter, Fullerton, CA, U.S.A.), with a modified capillary cartridge holder as described previously.19
ARTICLE
CE was performed using a fused-silica capillary (Polymicro Technologies, Phoenix, AZ, U.S.A.) coated with a triple-layer of PBDSPB. The coating was prepared as described by Haselberg et al.20 The coating does not interfere with fluorescence detection. The capillary had an i.d. of 75 μm, an o.d. of 375 μm, and total/effective lengths of 73/47 cm. New capillaries were rinsed with 1 M sodium hydroxide and deionized water for 10 min at 20 psi before the coating was applied. A coated capillary was daily rinsed for 10 min at 10 psi with subsequently 0.1 M phosphoric acid, 10% (w/v) PB, deionized water, and BGE. Between protein analyses the capillary was rinsed with BGE for 5 min at 10 psi, or 10 min at 10 psi when the BGE contained urea. Separations were performed in reversed polarity mode with a separation voltage of 25 or 20 kV. Sample injection was performed hydrodynamically by applying 0.5 psi for 9 s, or 0.7 psi for 9 s when samples contained 7.0 M urea. These injections corresponded to about 1% of the capillary volume. Wavelength-Resolved Fluorescence Detection. WrFlu detection was performed using a fluorescence detection cell with lamp excitation (Argos 250B from Flux, Basel, Switzerland) and a SR-163 spectrograph connected to a DV420A CCD camera (Andor Technologies, Darmstadt, Germany). Excitation light from a XeHg lamp was selected by use of a 280 nm interference filter (Melles Griot, Didam, The Netherlands) and a 300 nm short-pass cutoff filter (Asahi Spectra USA, Torrance, CA, U.S.A.). The spectrograph was calibrated daily using the spectral lines of a Hg pen-ray light source (L.O.T.-Oriel, Darmstadt, Germany). Acquisition parameters were as follows: slit width, 200 μm; vertical shift speed, 16.25 μs; horizontal read-out rate, 33 kHz; chip temperature 60 °C; CCD exposure time, 3 s. Spectra were acquired using the full vertical binning mode. For every analysis, a background emission spectrum was constructed by averaging 15 spectra during the first minute of the CE run, i.e., measuring BGE only. The resulting spectrum was used for background correction of all emission spectra recorded after 1.0 min. Data acquisition and analysis was performed using the software program Solis (Andor Technologies, Darmstadt, Germany). Electropherograms were extracted from the three-dimensional (3D) data by plotting the signal intensity against time at a selected emission wavelength. In order to enhance signal-to-noise ratios, average electropherograms were constructed by integrating the protein emission signal over a 30 nm interval, including the proteins’ maximum emission wavelengths. Protein emission spectra were extracted from the 3D data by plotting signal intensity against wavelength at a selected migration time. Average protein spectra with enhanced signal-to-noise ratio were constructed by averaging the seven most intense emission spectra (i.e., a time interval of 21 s) recorded in a protein peak. A signal-intensity adjustment was performed on the emission spectra to correct for a lower transmission efficiency of the optics at low-UV wavelengths (3%). For a protein concentration of 5 μg/mL, the measured maximum emission wavelengths approach the true values and RSDs are below 1%. However, for reliable determination of the maximum emission wavelength, the RSD should be below 0.3% (i.e., 1.0 nm), which is within the spectral resolution of the detector.19 So, taking the data in Tables 1 and 2 into account, we conclude that for carbonic anhydrase II, β-lactoglobulin B, and lysozyme concentrations should be about 15 times and for R-chymotrypsinogen A about 28 times above their LOD to allow adequate determination of their maximum emission wavelength with the CE-wrFlu system. Probing Protein Unfolding. The capability of the CE-wrFlu system to monitor protein conformational changes was evaluated by analyzing various model proteins, which were incubated in 7.0 M of the chaotropic agent urea, in order to induce protein unfolding. The presence of urea changes the internal hydrogen bonds and destabilizes the protein. Urea is uncharged and UV transparent and was added to the BGE so that the incubated proteins remained unfolded during CE-wrFlu analysis. The PBDSPB coating appeared not to be affected by the presence of 7.0 M urea in the capillary. The EOF obviously decreased when urea was added to the BGE, due to the increased viscosity of the solution. However, both with and without urea, the EOF was constant and migration time RSDs for the EOF marker indole were below 1.5%, which is consistent with earlier findings.27 The viscosity of the BGE should be taken into account when protein electrophoretic mobilities measured in presence and absence of urea are compared. Therefore, protein mobilities obtained with CE-wrFlu using the urea-containing BGE were adjusted using a viscosity correction factor (see the Materials and Methods section for the procedure). As in most protein fluorescence studies, we assumed that changes in protein emission spectra induced by urea were caused by protein conformational changes only. This assumption was strongly supported by the fact that the emission spectra of L-tryptophan recorded in presence and absence of 7.0 M urea were virtually identical, yielding the same maximum emission wavelength. The model proteins R-lactalbumin, β-lactoglobulin A, β-lactoglobulin B, bovine serum albumin, and carbonic anhydrase II were incubated individually for 2.5 h in 50 mM sodium
Figure 3. Raw (gray) and fitted (black) emission spectra of lysozyme obtained during CE-wrFlu analysis of the protein test mixture. Protein concentration: (A) 10 μg/mL, (B) 2 μg/mL. Further conditions: see Figure 1.
Table 2. Average Maximum Emission Wavelength (λem,max) Values Obtained after Fitting Emission Spectra Recorded during CE-wrFlu Analysis (n = 3) of the Protein Test Mixture at Various Concentrationsa 10 μg/mL
5 μg/mL
2 μg/mL
referenceb
protein
λem,max (nm)
RSD (%)
λem,max (nm)
RSD (%)
λem,max (nm)
RSD (%)
R-chymotrypsinogen A
332.2
0.06
331.2
0.28
325.5
3.35
carbonic anhydrase II
341.9
0.11
342.3
0.33
321.3
8.57
β-lactoglobulin B
333.3
0.08
331.7
0.83
318.7
lysozyme
340.7
0.04
339.9
0.06
340.0
13.6 0.09
λem,max (nm) 331.8 341.1 332.8 340.2
Fitted with a log-normal distribution described by Burstein et al. (ref 21). b From reference spectra (100 μg/mL) recorded on a standard spectrophotometer. a
6064
dx.doi.org/10.1021/ac201136y |Anal. Chem. 2011, 83, 6060–6067
Analytical Chemistry
ARTICLE
Figure 4. Electropherograms (A and C) and emission spectra (B and D) obtained during CE-wrFlu of carbonic anhydrase II (50 μg/mL) in folded (A and B) and unfolded states (C and D). Carbonic anhydrase was incubated for 2.5 h in BGE before injection. The EOF marker indole (0.25 μg/mL) is indicated with an asterisk. Conditions: BGE, 30 mM sodium phosphate (pH 3.0) with 0 M (A and B) or 7.0 M (C and D) urea; separation voltage, 20 kV; other conditions, see the Materials and Methods section.
Figure 5. Emission spectra obtained during CE-wrFlu of (A) R-lactalbumin, (B) β-lactoglobulin A, (C) β-lactoglobulin B, and (D) bovine serum albumin. Proteins were incubated for 2.5 h in BGE before injection. Conditions: protein concentration, 50 μg/mL; BGE, 50 mM sodium phosphate (pH 3.0) with 0 M (black) or 7.0 M (gray) urea; separation voltage, 20 kV; other conditions, see the Materials and Methods section.
phosphate (pH 3.0) containing 0 or 7.0 M urea. The incubated protein solutions were analyzed by CE-wrFlu using the incubation media as BGE, and from the recorded data electropherograms and protein emission spectra were extracted. For each sample a single protein peak was obtained from which the
(viscosity-corrected) effective electrophoretic mobility as well as the maximum emission wavelength was determined. Protein emission spectra of the native and unfolded proteins were also recorded on a stand-alone spectrophotometer in order to allow verification of the CE-wrFlu results. 6065
dx.doi.org/10.1021/ac201136y |Anal. Chem. 2011, 83, 6060–6067
Analytical Chemistry
ARTICLE
Table 3. Effective Electrophoretic Mobility (μeff,c) and Maximum Emission Wavelength (λem,max) Values of Model Proteins Determined during CE-wrFlu Using a BGE with 0 or 7.0 M Urea μeff,c (108 m2 V1 s1) protein
0M
a
7.0 M
λem,max (nm) Δμeff,c
0M
7.0 M
red shift
expected shift (nm)b
carbonic anhydrase II
2.47
1.68
0.79
341
354
13
14
bovine serum albumin
2.68
2.14
0.54
329
352
23
24
R-lactalbumin β-lactoglobulin A
2.25 2.60
1.95 1.59
0.30 1.01
344 333
352 353
8 20
8 21
β-lactoglobulin B
2.65
1.67
0.98
333
354
21
21
a
Corrected for the increase of viscosity of the BGE (see the Materials and Methods section). b Determined from emission spectra recorded on a standalone spectrophotometer (see the Materials and Methods section).
As an example, Figure 4 shows the electropherograms and emission spectra obtained for carbonic anhydrase. The migration time of the EOF marker increased considerably in presence of 7.0 M urea (cf. Figure 4, parts A and C), due to the higher viscosity of the BGE. The unfolded carbonic anhydrase exhibits a much broader peak than the folded species, which might be explained by the occurrence of several states of unfolding. The wrFlu detector clearly revealed a red-shifted emission spectrum for carbonic anhydrase in 7.0 M urea (cf. Figure 4, parts B and D). The maximum emission wavelength shifted from 341 nm for the folded protein to 354 nm for the unfolded protein. This shift of 13 nm matched well with the shift observed for the spectra recorded on the spectrophotometer. The measured migration time yielded (viscosity-corrected) effective electrophoretic mobilities of 2.47 108 and 1.68 108 m2 V1 s1 for the folded and unfolded carbonic anhydrase, respectively. So, the unfolding of the protein (i.e., increase of the molecular hydrodynamic radius) is nicely reflected in a decrease of effective electrophoretic mobility. The emission spectra obtained during CE-wrFlu of the other model proteins are depicted in Figure 5, and the measured maximum emission wavelengths and effective electrophoretic mobilities are summarized in Table 3. The spectral red shift of the unfolded protein species is correctly monitored by the wrFlu detector. The magnitude of the shift varies from 8 to 23 nm among the proteins, but interestingly, the emission spectra of the unfolded species all show a maximum emission wavelength of 353 ( 1 nm. This corresponds with the emission wavelength of L-tryptophan in water, suggesting that the protein tryptophan residues are strongly exposed to water and, thus, that the proteins are unfolded to a large extent. Protein unfolding commonly leads to a lower quantum yield,2 and the observed intensity for the unfolded species indeed was lower for most proteins (Figure 5). The fact that the unfolded proteins also showed considerably wider peaks (and thus lower peak heights) also contributed to the lower intensity of the emission spectra. The protein R-lactalbumin exhibits a remarkable gain in quantum yield upon unfolding,28 which was nicely reflected in the emission spectrum recorded during CE-wrFlu (Figure 5A). The mobility results in Table 3 show that denaturation by urea leads to a measurable decrease of effective electrophoretic mobility for all tested proteins, which can be attributed to the increase of molecular size upon unfolding. When considering the CE-wrFlu data obtained for carbonic anhydrase, R-lactalbumin and β-lactoglobulin A and B, one might presume a more or less linear correlation between the magnitude of the spectral red shift
and the change in effective electrophoretic mobility observed upon protein unfolding. However, the large red shift (23 nm) obtained for denatured bovine serum albumin is accompanied by a relatively modest decrease of effective electrophoretic mobility (0.54 108 m2 V1 s1). Apparently, unfolding of bovine serum albumin causes a substantial change in local tryptophan residue environment, whereas the increase of the protein’s hydrodynamic size is relatively limited. It is important to note that protein fluorescence and electrophoretic mobility reflect protein unfolding in fundamentally different ways. So, CE-wrFlu yields two essentially independent parameters that can provide information on protein conformational changes.
’ CONCLUSION The present study shows that CE combined with lamp-based wrFlu detection allows the probing of protein conformational changes. The red shift in maximum emission wavelength indicative for protein unfolding can be readily detected by the CEwrFlu system. Moreover, the decrease in effective electrophoretic mobility related to unfolding can also be measured with CEwrFlu. The wrFlu detector shows good linearity, and protein LODs are in the 632 nM range, which is similar to sensitivities obtained for native proteins with laser-induced fluorescence detection. So, introduction of wrFlu detection does not reduce detection sensitivity, while essential spectral information on proteins can be obtained. Overall, we conclude that the CEwrFlu system offers a promising new tool for the characterization of intact protein samples in terms of composition, purity, and conformation. Protein samples that contain impurities, degradation products, and/or denatured species could effectively be separated, and the individual proteins could then be analyzed for changes in conformation. Such an approach might be particularly useful in the biopharmaceutical field which holds a great demand for techniques that allow quality assessment of intact proteins. CE-wrFlu might also be valuable for the conformational studies of individual proteins that, e.g., show multiple stages of unfolding or different conformations at equilibrium (unfolding intermediates). It will be interesting to study whether CE-wrFlu is able to provide fluorescence emission spectra of unfolding intermediates. A final important aspect could be a study of protein unfolding mechanisms, since a whole class of diseases is related to protein misfolding. Currently we are evaluating these aspects by investigating the unfolding behavior of several relevant proteins by CE-wrFlu. 6066
dx.doi.org/10.1021/ac201136y |Anal. Chem. 2011, 83, 6060–6067
Analytical Chemistry
ARTICLE
’ AUTHOR INFORMATION Corresponding Author
*E-mail:
[email protected]. Phone: +31 6-20291145. Fax: +31 30-2536655.
’ ACKNOWLEDGMENT This research was supported by the Dutch Technology Foundation STW, Applied Science Division of NWO, and the Technology Program of the Ministry of Economic Affairs. ’ REFERENCES (1) Lakowicz, J. R. Principles of Fluorescence Spectroscopy; Springer: New York, 2006. (2) Ladokhin, A. S. Fluorescence Spectroscopy in Peptide and Protein Analysis; John Wiley & Sons Ltd: Chichester, U.K., 2000. (3) Eftink, M. R. Biophys. J. 1994, 66, 482–501. (4) El Rassi, Z. Electrophoresis 2010, 31, 174–191. (5) Huang, Y. F.; Huang, C. C.; Hu, C. C.; Chang, H. T. Electrophoresis 2006, 27, 3503–3522. (6) Hu, S.; Dovichi, N. J. Anal. Chem. 2002, 74, 2833–2850. (7) Patrick, J. S.; Lagu, A. L. Electrophoresis 2001, 22, 4179–4196. (8) Hilser, V. J.; Freire, E. Anal. Biochem. 1995, 224, 465–485. (9) Righetti, P. G.; Verzola, B. Electrophoresis 2001, 22, 2359–2374. (10) Rochu, D.; Masson, P. Electrophoresis 2002, 23, 189–202. (11) Gavina, J. M. A.; Britz-McKibbin, P. Curr. Anal. Chem. 2007, 3, 17–31. (12) Zhang, X.; Stuart, J. N.; Sweedler, J. V. Anal. Bioanal. Chem. 2002, 373, 332–343. (13) Fuller, R. R.; Moroz, L. L.; Gillette, R.; Sweedler, J. V. Neuron 1998, 20, 173–181. (14) Kok, S. J.; Kristenson, E. M.; Gooijer, C.; Velthorst, N. H.; Brinkman, U. A. T. J. Chromatogr., A 1997, 771, 331–341. (15) Timperman, A. T.; Oldenburg, K. E.; Sweedler, J. V. Anal. Chem. 1995, 67, 3421–3426. (16) Dolnik, V. Electrophoresis 2008, 29, 143–156. (17) Garcia-Campana, A. M.; Taverna, M.; Fabre, H. Electrophoresis 2007, 28, 208–232. .; de Kort, B. J.; Somsen, G. W. J. Chromatogr., (18) Radenovic, D. C A 2009, 1216, 4629–4632. (19) De Kort, B. J.; De Jong, G. J.; Somsen, G. W. Electrophoresis 2010, 31, 2861–2868. (20) Haselberg, R.; de Jong, G. J.; Somsen, G. W. J. Sep. Sci. 2009, 32, 2408–2415. (21) Burstein, E. A.; Abornev, S. M.; Reshetnyak, Y. K. Biophys. J. 2001, 81, 1699–1709. (22) Gavina, J. M. A.; Das, R.; Britz-McKibbin, P. Electrophoresis 2006, 27, 4196–4204. (23) Kawahara, K.; Tanford, C. J. Biol. Chem. 1966, 241, 3228–3232. (24) Haselberg, R.; van der Sneppen, L.; Ariese, F.; Ubachs, W.; Gooijer, C.; de Jong, G. J.; Somsen, G. W. Anal. Chem. 2009, 81, 10172–10178. (25) Paquette, D. M.; Sing, R.; Banks, P. R.; Waldron, K. C. J. Chromatogr., B 1998, 714, 47–57. (26) Tseng, W. L.; Chang, H. T. Anal. Chem. 2000, 72, 4805–4811. (27) Katayama, H.; Ishihama, Y.; Asakawa, N. Anal. Chem. 1998, 70, 5272–5277. (28) Permyakov, E. A.; Morozova, L. A.; Burstein, E. A. Biophys. Chem. 1985, 21, 21–31.
’ NOTE ADDED AFTER ASAP PUBLICATION This paper was published on the Web on July 8, 2011 with minor errors in equations 13. The corrected version was reposted on July 13, 2011. 6067
dx.doi.org/10.1021/ac201136y |Anal. Chem. 2011, 83, 6060–6067
