Enzymatic Cleavage and Subsequent Facile Intramolecular

Oct 31, 2016 - However, the majority of currently available GGT fluorescent probes ... (2) Therefore, methods for detection GGT activities are greatly...
0 downloads 0 Views 881KB Size
Download PDF
Letter pubs.acs.org/ac

Enzymatic Cleavage and Subsequent Facile Intramolecular Transcyclization for in Situ Fluorescence Detection of γ‑Glutamyltranspetidase Activities Hongjuan Tong,†,§ Yongjun Zheng,†,§ Li Zhou,† Xiangmin Li,† Rui Qian,† Rui Wang,† Jianhong Zhao,† Kaiyan Lou,*,† and Wei Wang*,†,‡ †

Shanghai Key Laboratory of New Drug Design, Shanghai Key Laboratory of Chemical Biology, School of Pharmacy, and State Key Laboratory of Bioengineering Reactor, East China University of Science & Technology, 130 Meilong Road, Shanghai 200237, China ‡ Department of Chemistry and Chemical Biology, University of New Mexico, Albuquerque, New Mexico 87131-0001, United States S Supporting Information *

ABSTRACT: γ-Glutamyltranspetidase (GGT) is a cell-membrane-bound enzyme which selectively catalyzes cleavage of the γ-glutamyl bond of glutathione (GSH). It has been identified to be overexpressed in a number of malignant tumor cells. Therefore, fluorescent probes for fast and selective detection of GGT activities are greatly needed. However, the majority of currently available GGT fluorescent probes based on direct conjugation of a γ-glutamyl group to a specific fluorophore generally has slow enzymatic kinetics due to bulky fluorophore too close to the enzyme’s active site. Moreover, the uncaged fluorophore with a free amine group might undergo oxidation or other enzymatic transformation and resulted in a complicated timedependent fluorescence response. Herein, we reported design of a novel fluorescent GGT probe NM-GSH (2), which incorporated a fast intramolecular transcyclization cascade for rapid detection of GGT activities after enzymatic cleavage of the γglutamyl group. This design strategy allows introduction of bulky 1,8-naphthalimide fluorophore with improved enzymatic kinetics and lowered detection limit. The transcyclized product 4 gives more than 200-fold fluorescence increment. The probe NM-GSH showed both good selectivity and fast detection of GGT activities with the detection limit as low as 0.21 mU/mL. In addition, the fluorescent product 4 contains no free amine group and is more stable for detection. Most importantly, cell imaging studies showed that the transcyclized product 4 was enriched in lysosomes for selectively lighting up GGT-overexpressed ovarian cancer cells (OVCAR5) but not normal cells (HUVEC), indicating NMGSH’s potentials as an imaging agent in cancer diagnosis and treatment. he cell-membrane-bound enzyme γ-glutamyltranspetidase (GGT; EC 2.3.2.2) selectively catalyzes cleavage of the γglutamyl bond of extracellular glutathione (GSH). In this process, GSH is metabolized to cysteinyl-glycine (Cys-Gly) by GGT and is further converted to cysteine and glycine by dipeptidase activities.1 This GSH metabolism pathway provides a complementary supply of cysteine, which is advantageous for cancer cell’s survival, metathesis, and defense mechanism against chemotherapy.2,3 Indeed, a number of human cancer cell lines have been identified to overexpress GGT in their cell membranes, and the enzyme is thus recognized as a potential cancer biomarker.2 Therefore, methods for detection GGT activities are greatly needed for clinical diagnosis and treatment of certain cancers with elevated GGT levels.4 Conventionally, GGT activities were measured by p-nitroanilide-based colorimetric assays.5,6 However, these assays are not suitable for real-time biological imaging applications. In the past decade, fluorescence-based methods have attracted significant research endeavors for their high sensitivity, noninvasive, and real-time capabilities.7−10 Fluorescent GGT probes had demonstrated to be promising in optically guided

T

© XXXX American Chemical Society

surgery and endoscopy by Y. Urano and co-workers.4 Majority of GGT fluorescent probes reported so far were designed by conjugation of a γ-glutamyl group to a specific fluorophore through an amide linkage.4,11−19 Enzymatic cleavage of the amide bond would then generate the fluorophore with unprotected amine group (Scheme 1). However, steric hindrance posed by a bulky fluorophore next to the enzymatic cleavage site might significantly reduce the probe’s enzymatic kinetic profiles (for example, gGlu-HMJCR,15 Table S1) and hamper its real-time applications. Moreover, the released fluorophore containing free amine group would potentially undergo oxidation or modification by other enzymes such as Nacetyltransferase (NAT) in biological system, which sometimes led to complicated time-dependent fluorescence response.16 Alternatively, a short linker or spacer, for example, 4hydroxymethyl aniline or Cys-Gly, were used to separate the enzymatic cleavage site away from the fluorophore.20−22 In this Received: September 1, 2016 Accepted: October 31, 2016 Published: October 31, 2016 A

DOI: 10.1021/acs.analchem.6b03448 Anal. Chem. XXXX, XXX, XXX−XXX

Letter

Analytical Chemistry

Initial fluorescence titration experiments with increased amount of GGT showed a new fluorescence emission band emerged at 473 nm in dose-dependent manner. For 1 μM of probe 2, it required a minimum 95 mU/mL of GGT to achieve a maximum fluorescence turn-on signal in about 20 min (see Supporting Information, Figure S1). This probe to enzyme ratio was used in the further emission and absorption studies. Time-dependent fluorescence emission spectra of 1 μM probe 2 showed a drastic turn-on fluorescence response (>200-fold fluorescence increment) at 473 nm upon addition of GGT within only 20 min at 25 °C (Figure 1a,b). The short reaction

Scheme 1. Comparison of Direct Enzymatic Cleavage Approach (Example ANF-Glu from ref 16) with the Current Worka

For clarity, the γ-glutamyl group was labelled in red; the top right inset shows the structure of the starting material 1 for probe 2.

a

indirect conjugation design, a cascade reaction triggered by enzymatic activities converts the direct enzymatic cleavage product with little fluorescence change to the final product with distinctive fluorescence emissions. A fast cascade reaction is therefore preferred, as a slow cascade reaction may give delayed fluorescence response to enzymatic activities. Since enzymatic kinetics is calculated from time-dependent fluorescence changes, a slow cascade reaction would result in underestimated enzymatic activities. Moreover, for quantification purpose, the cascade reaction is required to give close to quantitative yield. Overall, though this two-step cascade sequence may reduce the steric effect of a bulky fluorophore, it also adds significant challenge in probe design for a fast cascade reaction with high yield. In this work, we reported a turn-on fluorescent probe NMGSH (2), which incorporated a fast intramolecular transcyclization cascade reaction, for rapid in situ detection of GGT activities. The cascade reaction was triggered by the removal of the γ-glutamyl group by the GGT enzyme and resulted in facile formation of a six-membered thiomorpholinone ring and ringopening of a five-membered succinimide ring (Scheme 1), a process called transcylization.23 To the best of our knowledge, this complex manipulation of ring systems was first adopted here for detection of enzymatic activities. It allows the introduction of bulky fluorophore away from the enzymatic cleavage site to improve the probe’s enzymatic kinetic profiles. Moreover, the transcyclized product 4 does not contain free amine group, thus eliminates potential problems from amine oxidation or modifications by other enzymes. Probe 2 was conveniently synthesized from the thiol-Michael addition of GSH toward N-(N′-butyl-1,8-naphthalimide-4yl)maleimide (1)23 (Scheme S1). As the thiol-Michael addition could happen from both faces of the maleimide group, probe 2 was obtained as a mixture of inseparable two diastereomers in a ratio of 1:1 based on NMR integrations. The identity of probe 2 was firmly confirmed by 1HNMR, 13CNMR, and HRMS (see Supporting Information, Figure S20). With the probe 2 in hand, we first set out to study its fluorescence and UV−vis response upon incubation with GGT.

Figure 1. (a) Time-dependent fluorescence emission spectra of the probe 2 (1 μM) in the presence of GGT (95 mU/mL); (b) fluorescence emission spectra of probe 2 (1 μM) before and after addition of GGT (95 mU), isolated product 4 (1 μM), and reference compound 3a (1 μM); (c) time-dependent fluorescence intensity at 473 nm of probe 2 (1 μM) alone, probe 2 (1 μM) toward addition of GGT (95 mU/mL), and probe 2 (1 μM) toward addition of GGT (95 mU/mL) together with the inhibitor acivicin (400 μM); (d) stability of probe 2 (1 μM) at 473 nm in PBS buffer solution toward different biological samples and enzymes including 10% human plasma (HP), 10% fetal bovine serum (FBS), amyloglucosidase (Amy, 1 U/mL), collagen hydrolase (Hyd, 1 U/mL), chymotrypsin (Chy, 1 U/mL), pepsase (PEP, 1 U/mL), trypsin (Try, 1 U/mL), aprotinin (Apr, 1 U/ mL), alkaline phosphatase (ALP, 1 U/mL), and GGT (95 mU/mL). (All measurement was taken in PBS buffer (pH 7.4) at 25 °C with λex = 362 nm.)

time was comparable to those of previous reported probes designed with enzymatic cleavage and cyclization cascade at much higher concentration (10 μM) and more favorable temperature at 37 °C,21,22 indicating the probe 2 is a highly sensitive probe for GGT activities. The detection limit was determined to be 210 μU/mL of GGT at S/N = 3 (see Supporting Information, Figure S2a), significantly lower than two other reported GGT probes based on direct conjugation strategy using the same 1,8-naphthalimide fluorophore (see Supporting Information, Figure S2b).13,16 Different from drastic fluorescence difference, UV−vis studies showed only an 11 nm red-shift of maximum absorption wavelength from 343 to 354 nm with about the same intensity upon incubation of probe 2 with GGT (see Supporting Information, Figure S3). To confirm the observed fluorescence response was related to GGT activities, acivicin, a known inhibitor of GGT,24 was added in the presence of GGT. In this case, probe 2 showed a much slower fluorescence increment at 473 nm (Figure 1c) as B

DOI: 10.1021/acs.analchem.6b03448 Anal. Chem. XXXX, XXX, XXX−XXX

Letter

Analytical Chemistry expected from the suppressed GGT activity by this inhibitor. The above observations confirmed that probe 2 is an efficient turn-on fluorescence probe for GGT activities. We then started to verify the fluorescence sensing mechanism shown in Scheme 1. As the transient enzymatic product 3 could not be isolated, the reference compound 3a with acetylated amine group was synthesized (see Supporting Information, Scheme S3) and used to mimic fluorescence properties of 3. It turned out that 3a has very low fluorescence emission similar to probe 2 (Figure 1b), as both probe 2 and compound 3a share a very similar electron-withdrawing substituted succinimide group at 4-position of the 1,8naphthalimide fluorophore. The low fluorescence is likely caused by suppressed internal charge transfer (ICT)25 in the excited state. To confirm the transcyclized product 4 was responsible for the drastic turn-on fluorescence response, the standard sample of compound 4 was synthesized from a different synthetic route by mixing N-(N′-butyl-1,8-naphthalimide-4-yl)maleimide (1) with dipeptide Cys-Gly (see Supporting Information, Scheme S2). The structure of 4 was firmly confirmed by 1HNMR, 13CNMR, 2D-NOESY, and HRMS (see Supporting Information, Figure S21). We found the normalized fluorescence emission spectrum of the probe 2 upon incubation with GGT overlapped exactly with that of the standard sample 4 (see Supporting Information, Figure S4b) with a slightly lower maximum intensity at 473 nm than that of the solution of the standard sample 4 at the same concentration (Figure 1b), indicating close to quantitative yield of product 4 when probe 2 was treated with GGT. Besides, high-pressure liquid chromatography−mass spectrometry (HPLC−MS) analysis confirmed that the major product formed in the enzymatic reaction solution has the same retention time at 6.55 min and MS profiles as those of the standard sample of 4 (see Supporting Information, Figure S13). All the above observations strongly supported the formation of the transcyclized product 4 in high yield after the enzymatic cleavage of the γglutamyl bond by GGT. Moreover, fluorescence quantum yields of the probe 2, product 4, and the reference compound 3a were measured to be 0.026, 0.608, and 0.045, respectively (see Supporting Information, Figure S5). The slightly increased quantum yield of 3a compared with the probe 2 suggested the intermediate 3 did not contribute much to the observed drastic fluorescence increase. Besides, the transcyclized product 4 has much higher quantum yield than probe 2 and 3a, likely caused by the increased ICT in the excited state due to more electrondonating amide group instead of highly electron-withdrawing succinimide group at the 4-position. DFT calculations at the B3LYP/6-31G(d,p) level on the optimized ground state structures of the compound 4 and intermediate (R,R)-3 showed a relatively shorter C−N bond at 6-position for compound 4 (141.5 vs 143.1 pm, see Supporting Information, Figure S16) consistent with increased p-π conjugation from more electron-donating nitrogen atom at 6-position and thus more double bond character of the C−N bond. Moreover, fluorescence lifetimes measured by time-correlated single photon counting (TCSPC) method for probe 2 and probe 2 after treatment with GGT were 1.46 and 2.41 ns, respectively (see Supporting Information, Figure S12). The observed longer fluorescence lifetime after GGT treatment suggested a more stabilized excited state of product 4, which agreed well with increased solvent-fluorophore interactions when the excited state has more ICT induced charge separation and increased dipole moment.25

We further determined the enzymatic kinetic constants between GGT and probe 2 according to the Michaelis−Menten equation based on the fluorescence turn-on response, assuming the transcyclization reaction was sufficiently faster than the enzymatic cleavage reaction. The Km and Vmax values were obtained as 17.64 μM and 1.44 μM/min, respectively (see Supporting Information, Figure S6 and Table S1). We also measured the pseudo-first-order rate constant of the transcyclization reaction to be 0.516 min−1 with reaction half-time equal to 1.34 min (see Supporting Information, Figures S8 and S9). On the basis of the value of half-time, the transcyclization step is sufficiently fast to achieve a high yield at an incubation time of 20 min. On the basis of the theoretical calculations, the transcyclized product 4 is 49.0 kJ/mol more stable in the ground state than the direct enzyme product (R,R)-3 (see the Supporting Information). This large energy difference is likely caused by the released ring-strain presented in the fivemembered succinimide ring. Moreover, the transcyclization was initiated by the intramolecular nucleophilic attack of amine group to the nearby carbonyl group on the succinimide ring in 6-exo-trig fashion, which is kinetically feasible based on the Baldwin’s rules.26 We then evaluated probe 2’s stability and selectivity toward a range of different biological fluids and specific enzymes, including 10% human plasma (HP), 10% fetal bovine serum (FBS), amyloglucosidase, collagen hydrolase, chymotrypsin, pepsase, trypsin, aprotinin, and alkaline phosphatase. Results showed probe 2 had good selectivity for GGT detection at all tested conditions (Figure 1d). Moreover, the effect of pH (3− 11) on the fluorescence intensity of probe 2 and product 4 were studied (see Supporting Information, Figure S10). It was concluded that probe 2 was suitable for GGT detection at pH ≤ 8. We further investigated the potential interference from some common metal ions (see Supporting Information, Figure S11). The probe retained its efficacy in GGT detection in the presence of metal ions examined including Ca2+, Mg2+, Al3+, Zn2+, Fe3+, and Fe2+ except Cu2+. The interference of Cu2+ might be caused by the inhibition of the GGT activity by copper(II) ion,27 or through decomposition of the probe 2 through reverse thiol-Michael addition to form GSH-Cu(I) complex.28 Inspired by the above in vitro fluorescence studies, we are convinced that the probe 2 could be used for the determination of GGT activities in live cells. Human ovarian cancer OVCAR5 cells, which overexpress GGT, and noncancer human umbilical vein endothelial cells (HUVECs) as control, were used as model cell-lines for study. Cell viability assay showed no significant toxicity of probe 2 up to 20 μM (see Supporting Information, Figure S17). OVCAR5 cells were then incubated in cell culture media containing probe 2 (10 μM) at 37 °C for 30 min before being washed thoroughly with D-Hanks (pH = 7.4) to remove the extracellular probe. The cells were imaged using confocal fluorescence microscope at the blue channel (420−475 nm). The confocal fluorescence images obtained clearly showed bright blue fluorescence in cells (Figure 2b). In contrast, when HUVECs cells were treated with probe 2 under the same conditions, the fluorescence intensity in the cell images was greatly reduced (Figure 2a). Besides, when OVCAR5 cells were pretreated with acivicin (1 mM), fluorescence intensity in the images was drastically reduced (Figure 2c). Similarly, when cells were pretreated with Cu2+ ions (5 μM) and washed before loading with probe 2, the fluorescence intensity were also greatly suppressed (see C

DOI: 10.1021/acs.analchem.6b03448 Anal. Chem. XXXX, XXX, XXX−XXX

Analytical Chemistry

■

Letter

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.analchem.6b03448. Synthetic procedures, theoretical calculations, NMR, and other UV−vis and fluorescent spectra (PDF)

■

AUTHOR INFORMATION

Corresponding Authors

*E-mail: [email protected]. *E-mail: [email protected]. Author Contributions §

H.T. and Y.Z. contributed equally.

Figure 2. Fluorescence confocal images (a−c) collected at 420−475 nm (λex = 405 nm) and the corresponding bright field view (d−f) of normal cells (HUVEC cells, a, d), cancer cells (OVCAR5 cells, b, c, e, f): (a,d) HUVEC cells incubated with probe 2 (10 μM) for 30 min; (b, e) OVCAR5 cells incubated with probe 2 (10 μM) for 30 min; (c, f) OVCAR5 cells pretreated with acivicin (1 mM) for 30 min then loaded with probe 2 (10 μM) for 30 min (scale bar = 20 μm).

Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS The work was supported by East China University of Science and Technology (start-up funds, W.W.), the Fundamental Research Funds for the Central Universities (Grant WY1213013, K. L.), the Pujiang Talent Project (Grant 14PJ1402200, K. L.), the National Science Foundation of China (Grant No. 21577037, K. L.), and the China 111 Project (Grant B07023, W.W.).

Supporting Information, Figure S18), suggesting the potential inhibition of GGT activity by Cu2+ ions.27 On the basis of the above fluorescence sensing mechanism and imaging studies, we proposed rationalization of the probe 2 for selective lighting-up GGT overexpressed OVCAR5 cancer cells. The probe 2’s γ-glutamyl bond is first cleaved upon binding with GGT on the cell membrane. And then the uncaged amino group undergoes facile intramolecular transcyclization reaction and generates fluorescent product 4 with reduced polarity, which readily enters the cancer cells and selectively stains the cancer cells with strong blue fluorescence.4,21 Evidence from imaging studies (see Supporting Information, Figure S19) showed that the blue fluorescence from the product 4 colocalized well with the Lyso-Tracker Green according to the value of the Pearson’s correlation coefficient (86%), while poorly with the Memb-Tracker Green and Mito-Tracker Green (40% and 50%, respectively), indicating it mainly enriches in lysosomes after entering the cancer cells. In conclusion, we have developed an efficient in situ fluorescent probe NM-GSH (2), which adopted a facile and fast intramolecular transcyclization reaction for rapid detection of γ-glutamyltranspetidase activities. Probe 2 showed drastic turn-on fluorescence response (>200-fold fluorescence increment) at 473 nm within 20 min and allowed fast detection of GGT activities with good selectivity. Moreover, the transcyclized fluorescent product 4 does not contain reactive free amine group for further modifications by other enzymes. Most importantly, it could easily pass through cell-membrane and enriched in lysosomes and selectively lit up the GGToverexpressed cancer cells for easy visualization. This property not only could be utilized for easy localization of cancerous tissues in clinical applications, such as optically guided operation, but also could open the road toward novel therapeutical opportunities such as targeted drug delivery and photodynamic therapy, which are currently under investigation in our lab.

■

REFERENCES

(1) Hanigan, M. H.; Ricketts, W. A. Biochemistry 1993, 32, 6302− 6306. (2) Pompella, A.; De Tata, V.; Paolicchi, A.; Zunino, F. Biochem. Pharmacol. 2006, 71, 231−238. (3) Daubeuf, S.; Leroy, P.; Paolicchi, A.; Pompella, A.; Wellman, M.; Galteau, M. M.; Visvikis, A. Biochem. Pharmacol. 2002, 64, 207−216. (4) Urano, Y.; Sakabe, M.; Kosaka, N.; Ogawa, M.; Mitsunaga, M.; Asanuma, D.; Kamiya, M.; Young, M. R.; Nagano, T.; Choyke, P. L.; Kobayashi, H. Sci. Transl. Med. 2011, 3, 110ra119. (5) Schiele, F.; Artur, Y.; Bagrel, D.; Petitclerc, C.; Siest, G. Clin. Chim. Acta 1981, 112, 187−195. (6) Fossati, P.; Melzi d’Eril, G. V.; Tarenghi, G.; Prencipe, L.; Berti, G. Clin. Chem. 1986, 32, 1581−1584. (7) Qian, X.; Xiao, Y.; Xu, Y.; Guo, X.; Qian, J.; Zhu, W. Chem. Commun. 2010, 46, 6418−6436. (8) Yang, Y. M.; Zhao, Q.; Feng, W.; Li, F. Y. Chem. Rev. 2013, 113, 192−270. (9) Li, X. H.; Gao, X. H.; Shi, W.; Ma, H. M. Chem. Rev. 2014, 114, 590−659. (10) Tang, Y. H.; Lee, D. Y.; Wang, J. L.; Li, G. H.; Yu, J. H.; Lin, W. Y.; Yoon, J. Y. Chem. Soc. Rev. 2015, 44, 5003−5015. (11) Prusak, E.; Siewinski, M.; Szewczuk, A. Clin. Chim. Acta 1980, 107, 21−26. (12) White, I. N. H.; Razvi, N.; Lawrence, R. M.; Manson, M. M. Anal. Biochem. 1996, 233, 71−75. (13) Hou, X.; Yu, Q.; Zeng, F.; Yu, C.; Wu, S. Chem. Commun. 2014, 50, 3417−3420. (14) Li, L. H.; Shi, W.; Wang, Z.; Gong, Q. Y.; Ma, H. M. Anal. Chem. 2015, 87, 8353−8359. (15) Iwatate, R. J.; Kamiya, M.; Urano, Y. Chem. - Eur. J. 2016, 22, 1696−1703. (16) Zhang, H.; Wang, K.; Xuan, X.; Lv, Q.; Nie, Y.; Guo, H. Chem. Commun. 2016, 52, 6308−6311. (17) Zhang, P.; Jiang, X.-f.; Nie, X.; Huang, Y.; Zeng, F.; Xia, X.; Wu, S. Biomaterials 2016, 80, 46−56. (18) Park, S.; Bae, D. J.; Ryu, Y.-M.; Kim, S.-Y.; Myung, S.-J.; Kim, H.-J. Chem. Commun. 2016, 52, 10400−10402. D

DOI: 10.1021/acs.analchem.6b03448 Anal. Chem. XXXX, XXX, XXX−XXX

Letter

Analytical Chemistry (19) Hou, X.; Zeng, F.; Wu, S. Biosens. Bioelectron. 2016, 85, 317− 323. (20) Park, S.; Lim, S.-Y.; Bae, S. M.; Kim, S.-Y.; Myung, S.-J.; Kim, H.-J. ACS Sens. 2016, 1, 579−583. (21) Wang, F.; Zhu, Y.; Zhou, L.; Pan, L.; Cui, Z.; Fei, Q.; Luo, S.; Pan, D.; Huang, Q.; Wang, R.; Zhao, C.; Tian, H.; Fan, C. Angew. Chem., Int. Ed. 2015, 54, 7349−7353. (22) Li, L.; Shi, W.; Wu, X.; Gong, Q.; Li, X.; Ma, H. Biosens. Bioelectron. 2016, 81, 395−400. (23) Li, X.; Zheng, Y.; Tong, H.; Qian, R.; Zhou, L.; Liu, G.; Tang, Y.; Li, H.; Lou, K.; Wang, W. Chem. - Eur. J. 2016, 22, 9247−9256. (24) Han, L.; Hiratake, J.; Kamiyama, A.; Sakata, K. Biochemistry 2007, 46, 1432−1447. (25) Lakowicz, J. R. Principles of Fluorescence Spectroscopy, 3rd ed.; Springer: New York, 2006. (26) Baldwin, J. E. J. Chem. Soc., Chem. Commun. 1976, 734−736. (27) Yuan, J.; Kang, X.; Hu, Z.; Zou, G. China Biotechnol. 2008, 28, 100−105. (28) Speisky, H.; Gomez, M.; Carrasco-Pozo, C.; Pastene, E.; LopezAlarcon, C.; Olea-Azar, C. Bioorg. Med. Chem. 2008, 16, 6568−6574.

E

DOI: 10.1021/acs.analchem.6b03448 Anal. Chem. XXXX, XXX, XXX−XXX