Subscriber access provided by Binghamton University | Libraries
Article
Anion-exchange chromatography coupled to high resolution mass spectrometry: a powerful tool for merging targeted and non-targeted metabolomics Michaela Schwaiger, Evelyn Rampler, Gerrit Hermann, Walter Miklos, Walter Berger, and Gunda Koellensperger Anal. Chem., Just Accepted Manuscript • Publication Date (Web): 05 Jun 2017 Downloaded from http://pubs.acs.org on June 10, 2017
Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.
Analytical Chemistry is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.
Page 1 of 18
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Analytical Chemistry
Anion-exchange chromatography coupled to high resolution mass spectrometry: a powerful tool for merging targeted and non-targeted metabolomics Michaela Schwaiger†, Evelyn Rampler†, Gerrit Hermann†,‡, Walter Miklos§, Walter Berger§, Gunda Koellensperger†,∫,* † Department of Analytical Chemistry, Faculty of Chemistry, University of Vienna, Waehringer Str. 38, 1090
Vienna, Austria
‡ ISOtopic solutions, Waehringerstr. 38, 1090 Vienna, Austria §
Institute of Cancer Research, Department of Internal Medicine I, Medical University of Vienna, Borschkegasse 8a, 1090
Vienna, Austria ∫
Vienna Metabolomics Center (VIME), University of Vienna, Althanstrasse 14, 1090 Vienna, Austria
* Corresponding author: E-Mail:
[email protected] ABSTRACT In this work, simultaneous targeted metabolic profiling by isotope dilution and non-targeted fingerprinting is proposed for cancer cell studies. The novel streamlined metabolomics workflow was established using anion-exchange chromatography (IC) coupled to high resolution mass spectrometry (MS). The separation time of strong anion-exchange (2 mm column, flow rate 380 µL min-1, injection volume 5 µL) could be decreased to 25 min for a target list comprising organic acids, sugars, sugar phosphates and nucleotides. Internal standardization by fully
13C
labeled Pichia
pastoris extracts enabled absolute quantification of the primary metabolites in adherent cancer cell models. Limits of detection (LODs) in the low nM range and excellent intermediate precisions of the isotopologue ratios (on average < 5% N=5 over 40 hours) were observed. As a result of internal standardization, linear dynamic ranges over 4 orders of magnitude (5 nM – 50 µM, R2 > 0.99) were obtained. Experiments on drug-sensitive versus resistant SW480 cancer cells showed the feasibility of merging analytical tasks into one analytical run. Comparing fingerprinting with and without internal standard proved that the presence of the
13C
labeled yeast extract required for absolute
quantification was not detrimental to non-targeted data evaluation. Several interesting metabolites were discovered by accurate mass and comparing MS2 spectra (acquired in ddMS2 mode) with 1 ACS Paragon Plus Environment
Analytical Chemistry
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
spectral libraries. Significant differences revealed distinct metabolic phenotypes of drug-sensitive and resistant SW480 cells.
Today, high resolution mass spectrometry (MS) has reached a point that hypothesis driven absolute quantification studies can be performed using accurate mass profiling methods rather than using multiple reaction monitoring by triple quadrupole MS. In fact, there is a paradigmatic shift recognizing the capabilities of modern high resolution MS as quantitative tools.1 High resolution MS and simplified workflows,2 will be the key to improve the analytical throughput in metabolomics. Starting point of the establishment of any metabolomics toolbox should be the targeted approach. Accuracy of quantification demands validation of sample preparation procedures regarding extraction efficiency and recovery, quenching efficiency, and cell leakage upon quenching.3,4 In turn, this facilitates the successful establishment of non-targeted procedures, as the developed sample preparations routinely address metabolites of widely differing chemical and physical properties.5 Apart from this fact, comprehensive metabolomics workflows involve both hypothesis driven targeted quantification in parallel to non-targeted analysis for the discovery of metabolic rearrangements drawing a clear line between the two essential tasks. Up to now, hypothesis generation and hypothesis validation is addressed as separate measurement tasks on different MS platforms.2,6 As a consequence, state of the art metabolomic experiments require multiple analytical runs on one sample. Typically, hydrophilic interaction liquid chromatography (HILIC) and reversed phase (RP) chromatography (MS measurements in positive and negative mode using acidic and basic pH) in combination with high resolution MS are implemented as discovery tools.7 As both separation methods show poor selectivity for the central carbon metabolome, comprehensive workflows often comprised targeted analysis dedicated to exactly this metabolite panel next to nontargeted analysis.8 Only recently, a seminal critical review indicated the potential of merging different analytical tasks into simplified workflows.2 Reducing the complexity of the sample by on line derivatization with p-toluenesulfonylhydrazine, merging non-targeted and targeted analysis of aldehydes and ketones in one analytical run was enabled.9 In this work, we address a merged workflow based on anion-exchange chromatography (IC) in combination with high resolution MS detection. Anion chromatography proved to be a valuable separation method for key metabolites of the primary carbon metabolome,10–17 offering an alternative to the state of the art strategies based on gas chromatography (GC)18 and ion pairing LC.19–22 GC is unrivalled regarding the selectivity of the separation of sugars and sugar phosphates, 2 ACS Paragon Plus Environment
Page 2 of 18
Page 3 of 18
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Analytical Chemistry
however, falls short when nucleotides are of interest. Compared to IC-MS, LODs in GC-MS/MS are on average somewhat higher (factor of 2-10). E.g. for sugar phosphates LODs ranging between 1200 fmol were obtained by GC-MS/MS.18 As a drawback, derivatization required in GC-MS is depending on the sample matrix and on the degree of automation (just-in-time derivatization by a robotic system is desirable). Ion pairing chromatography offers the highest coverage of the primary metabolome.22,23 Nevertheless, it has been only rarely coupled to high resolution MS,20,24 as the use of an ion pairing reagent demands for the establishment of a dedicated instrumentation.25 In fact, the contamination of ion pairing reagent is detrimental for other applications. Moreover, the general sensitivity is compromised due to suppression effects. Especially for nucleoside diphosphates and triphosphates rather high LODs ranging from 50 to 200 ng mL-1 (sub-µM) have been reported.20 Ion chromatography, more specifically anion-exchange chromatography, provides comparable separation selectivity to ion pairing chromatography. However, molecules that can be protonated (e.g. amino acids) are lost in the suppression system required for MS detection and are hence not amenable to subsequent MS analysis. As a matter of fact, many metabolites of the central carbon cycle occur in multiple isomers and are prone to in-source fragmentation. Accordingly, their separation becomes an absolute prerequisite for the two essential tasks accurate quantification and compound identification from MS fingerprints. Anion-exchange chromatography is a powerful separation method for metabolomic applications. Table S-1 gives an overview on the state of the art anion-exchange MS couplings in the field.10–17 So far, few studies covered the different compound classes amenable for anion-exchange by normal flow chromatography comprehensively. If the selected metabolite panel was small, the run time was reduced to 20 min. E.g. a recent work15 addressed IC-high resolution MS for absolute quantification of 6 organic acids based on standard addition via
13C
labeled standards was
implemented as quantification strategy. Moreover, several sugar phosphates were identified by their established non-targeted evaluation pipeline using the Metlin database (La Jolla, CA, USA).15 Moreover, a dedicated method for the quantification of 28 organic acids with run times of 19 min was published recently.17 Capillary IC shows longer run times (45 – 75 min, see Table S-1) but was propagated with the idea of increasing robustness and sensitivity, as microflow would decrease the potential salt contamination of the ionization source. The increased sensitivity results from maximizing the injection volume as typically 5 µL are injected into a flow of 25 µL min-1.11,14 In this work, we propose a streamlined workflow by IC-high resolution MS (Q Exactive HF) integrating targeted quantification and non-targeted fingerprinting into one analytical run. We addressed internal standardization based on fully
13C
labeled Pichia pastoris extracts. The
3 ACS Paragon Plus Environment
Analytical Chemistry
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
capabilities of this versatile internal standardization strategy have been shown in several studies.23,26–28 In order to ensure absolute quantification in metabolomics, isotope dilution is required. As a consequence, in this work we investigated whether the addition of the internal standard impeded non-targeted fingerprinting with regard to compound annotations and differential analysis between different sample groups. A proof of principle study addressed the metabolome of drug sensitive colon cancer cell as compared to the model of acquired resistance.
EXPERIMENTAL SECTION Metabolite standards, internal standard and solvents The metabolite standards for organic acids, sugar phosphates and nucleotides were purchased from Sigma-Aldrich or Fluka (Vienna, Austria) except of malic acid, which was purchased from Merck (Vienna, Austria). Standard stock solutions of 1 or 10 mM were prepared in water and used for the preparation of a multi-component standard of 0.5, 1, 5, 10, 50, 100, 500 nM and 1, 5, 10, 50 µM. Additionally, a quality control (QC) of 1 µM was prepared. A fully 13C labeled yeast extract of Pichia pastoris (2 billion cells) from ISOtopic solutions e.U., (Vienna, Austria) was reconstituted in 2 mL water and added in same amounts to the calibration standards, the QC as well as to the samples. The final dilution of the internal standard for the measurement was 1:10 (v/v). Deionized water from an ultra-pure water system (resistance 18.2 MΩ) was used for ion chromatography. Furthermore, LC-MS grade methanol from Fluka (Vienna, Austria) was used as make-up flow.
Cancer cell culture and sample preparation The human colon carcinoma cell line SW480 was purchased from the American Tissue Collection Center (ATCC). The sensitive cells were compared to its subline of acquired triapine resistance.29 Five times 1 x 106 of each cell type were seeded in 6-well plates in minimal essential medium (MEM) containing 10% fetal calf serum without antibiotics and incubated for 24 h (37 °C, 5% CO2). The sampling procedure was based on the “direct solvent scraping” method.30 After 24 h, the medium was removed and the wells were washed three times with 1 mL of PBS solution. Then, 50 µL internal standard and 1 mL ice-cold methanol (80% methanol, 20% water, v/v) were added. The cells were scraped into the extraction solvent and transferred to an Eppendorf tube. Then, the cell scraper and the wells were washed two times with 475 µL extraction solvent to reach a final volume of 2 mL extract. After thorough mixing and centrifugation (20,000 RCF, 5 min, 4 °C), the 4 ACS Paragon Plus Environment
Page 4 of 18
Page 5 of 18
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Analytical Chemistry
supernatant was aliquoted into HPLC vials (400 µL were dried under reduced pressure and reconstituted in 100 µL water) whereas the cell pellet was used for protein determination with the 2D-Quant Kit (GE Healthcare, Munich, Germany). For the non-targeted approach, additionally samples without internal standard were prepared, thus the wells were washed with 2 x 500 µL to reach the final volume of 2 mL. Moreover, a pooled sample was prepared by mixing same aliquots of all samples (separately for extracts with and without internal standard). Samples were measured in a randomized way. After each tenth injection, the QC sample was analyzed. Additionally, calibration standards were measured in the beginning, the middle and the end of the entire sequence. For method validation, samples without internal standard were measured in a separate block.
Ion chromatography A Dionex Integrion HPIC System (Thermo Scientific) was used for anion-exchange chromatography. The separation was conducted on a Dionex IonPac AS11-HC column (2 x 250 mm, 4 µm particle size, Thermo Scientific) equipped with a Dionex IonPac AG11-HC guard column (2 x 50 mm, 4 µm, Thermo Scientific) at 30 °C. A potassium hydroxide gradient was produced by an eluent generator with a potassium hydroxide cartridge that was supplied with deionized water. The separation was carried out with a step gradient at a flow rate of 0.380 mL min-1 beginning with 10 mM KOH over 3 min, 10-50 mM from 3 to 12 min, 50-100 mM from 12 to 19 min, held at 100 mM from 19-21 min and re-equilibrated at 10 mM for 4 min. This resulted in a total run time of 25 min. A Dionex AERS 500, 2 mm suppressor was used to exchange potassium ions against protons in order to produce water instead of potassium hydroxide before entering the mass spectrometer. It was operated with 95 mA at a temperature of 15 °C. Furthermore, methanol was provided as make-up flow at a flow rate of 0.150 mL min-1. The samples were introduced via a Dionex AS-AP autosampler and full loop injection on a 5 µL loop (overfill factor 3). The temperature of the autosampler was set to 6 °C.
Mass spectrometry High resolution mass spectrometry was conducted on a high field Thermo Scientific™ Q Exactive HF™ quadrupole-Orbitrap mass spectrometer equipped with an electrospray source. Full mass scan (full MS, 50 – 750 m/z) in negative mode was used at a resolution of 120,000 for all calibration standards and samples. The automatic gain control (AGC) target was set to 1 x 106 ions and the maximum injection time (IT) was 200 ms. The ESI source parameters were the following: sheath gas 50, auxilary gas 14, sweep gas 3, spray voltage 2.75 kV, capillary temperature 230 °C, S-Lens RF level 45 and auxiliary gas heater 380 °C. Spectrum data were acquired in profile mode. For the nontargeted approach, also data-dependent MS2 (ddMS2) fragmentation spectra of the 1 µM standard mix spiked with internal standard and of a pooled sample (prepared by mixing aliquots of all 5 ACS Paragon Plus Environment
Analytical Chemistry
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
samples) were acquired. A top 5 method with an AGC target of 1 x 105, a maximum injection time of 50 ms and a minimum AGC target of 1 x 103 was used. The ions were isolated with 1 m/z, fragmented with HCD energy NCE 30 and detected with a resolution of 30,000.
Data processing Targeted data evaluation for the quantification of organic acids, sugar, sugar phosphates and nucleotides was performed in TraceFinderTM 3.3 from Thermo Scientific™ with internal standardization. All calibration curves were linear and weighted 1/x. For non-targeted data processing, Thermo Scientific™ Compound Discoverer™ 2.0 software was used. This software combines feature detection with statistical data evaluation. Retention time alignment within 0.15 min and 5 ppm mass tolerance was performed. For the detection of unknown features on MS1 level, a minimum peak intensity of 10,000, 3 ppm mass tolerance and a minimum number of 2 isotopes were used. Unknown compounds were grouped according to a mass tolerance of 3 ppm and 0.15 min. The “Fill Gaps” node was used with 3 ppm and 0.1 min. mzCloud search was performed with 5 ppm mass tolerance and an assignment threshold for compound annotation of 70.
RESULTS AND DISCUSSION IC high-resolution MS of metabolites In this work, merging absolute quantification based on isotope dilution and non-targeted fingerprinting by normal flow IC-high resolution MS (Q Exactive HF) was evaluated for a 45 metabolite panel. Starting point of the study was the separation of a 45 metabolite standards including organic acids, nucleotides and sugar phosphates on the IonPac AS11-HC column using normal flow IC and relatively short separation times of 25 min. Establishing a gradient up to 100 mM KOH, nucleoside triphosphates could be successfully eluted obtaining good peak shapes (see Figure S-1). All mono-, di- and triphosphates of the nucleosides and the isomers deoxyguanosine-triphosphate (dGTP) and adenosine-triphosphate (ATP) were baseline separated. Due to the short gradient, only limited selectivity was obtained regarding sugar phosphates. In the case of hexose-monophosphates, G1P and F1P were baseline separated, whereas the 6-phosphates of glucose, fructose and mannose were not. Fructose-1,6-bisphosphate was successfully separated from all hexose-monophosphates enabling absolute quantification despite in source fragmentation (see Figure S-1). In fact, significant loss of one phosphate group was observed resulting in the isobar fragment at m/z 259.0224 otherwise overlapping with the hexose-monophosphates. No separation could be achieved for the pentose-5-phosphates (ribose-5-phosphate, ribulose-5-phosphate). The 6 ACS Paragon Plus Environment
Page 6 of 18
Page 7 of 18
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Analytical Chemistry
two isomers citric acid and isocitric acid were almost baseline separated from each other and separated from cis-aconitic acid. The latter was crucial, as citric and isocitric showed in source fragmentation which could not be resolved from cis-aconitic acid by mass selectivity (m/z of 173.0092 [M-H]-). Oxaloacetic acid was separated from pyruvic acid and malic acid from fumaric acid, respectively. Again this was crucial since pyruvic and fumaric acid are produced as fragments of oxaloacetic and malic acid, respectively, in the ESI source (see Figure S-1). Table 1 gives an overview on the selected metabolite panel and their respective retention times.
Internal standardization with uniformly 13C labeled metabolites produced from Pichia pastoris It is a commonly accepted fact that an isotope ratio based approach is a prerequisite for accurate quantification in metabolomics.23 In this study, we implemented a fully
13C
labeled cell extract
derived from the yeast Pichia pastoris as compound specific internal standard providing nearly full coverage of the selected metabolite panel (see Table 1). In the past, this internal standardization strategy enabled accurate quantification of primary metabolites as proven by interlaboratory31 and inter-platform comparisons.32 For the selected metabolite panel of this study, the concentrations of the uniformly following
13C
13C
labeled analogs were found in the low nM up to several µM range. Only the
compounds revealed concentrations too low for internal standardization: the four
deoxytriphosphates present in the DNA (dATP, dCTP, dGTP, dTTP), dCMP, the cyclic forms cAMP and cGMP, CMP, isocitric acid, oxaloacetic acid and mannitol-1-phosphate. In the case of the poorly separated hexose-6-phosphates, the sum of the 13C peaks was used to correct for G6P, F6P and M6P. The internal standardization strategy for all compounds is given in Table 1.
Table 1. List of 45 metabolite standards, retention times and the uniformly 13C labeled metabolites used for internal standardization. Compound
2-Phosphoglyceric acid (2PG) and 3PG AMP 6-Phosphogluconic acid ADP alpha-Ketoglutaric acid ATP cAMPa
RT [min]
12C
metabolite
Uniformly 13C labeled internal standard ISTD [M-H]-
Formula
[M-H]-
ISTD compound
ISTD formula
13.65
C3H7O7P
184.9857
U13C 2PG + 3PG
13C3H7O7P
187.9957
11.77
C10H14N5O7P
346.0558
13C10H14N5O7P
356.0894
12.67
C6H13O10P
275.0174
13C6H13O10P
281.0375
16.90
C10H15N5O10P2
426.0221
13C10H15N5O10P2
436.0557
10.68
C5H6O5
145.0143
13C5H6O5
150.0310
19.82
C10H16N5O13P3
505.9885
U13C AMP U13C 6Phosphogluconic acid U13C ADP U13C alphaKetoglutaric acid U13C ATP
13C10H16N5O13P3
516.0220
11.95
C10H12N5O6P
328.0452
U13C AMP
7 ACS Paragon Plus Environment
Analytical Chemistry
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Compound
RT [min]
12C
metabolite
Page 8 of 18
Uniformly 13C labeled internal standard
CDP
14.29
C9H15N3O11P2
402.0109
U13C CDP
13C9H15N3O11P2
ISTD [M-H]411.0411
cGMPa
19.40
C10H12N5O7P
344.0402
cis-Aconitic acid
15.49
C6H6O6
173.0092
13C6H6O6
179.0293
Citric acid
14.45
C6H8O7
191.0197
U13C GMP U13C cis-Aconitic acid U13C Citric acid
13C6H8O7
197.0399
C9H14N3O8P
322.0446
U13C GMP 13C9H16N3O14P3
491.0074
13C6H14O12P2
345.0089
13C6H13O9P
265.0426
13C6H13O9P
265.0426
Formula
[M-H]-
ISTD compound
ISTD formula
CMPa
9.02
CTP
17.72
C9H16N3O14P3
481.9772
U13C CTP
dATPa
19.58
C10H16N5O12P3
489.9936
U13C ATP
dCMPa
7.91
C9H14N3O7P
306.0497
U13C GMP
dCTPa
17.12
C9H16N3O13P3
465.9823
U13C CTP
dGTPa Fructose-1,6bisphosphate (FBP) Fructose-1phosphate (F1P) Fructose-6phosphateb (F6P) Fumaric acid
22.14
C10H16N5O13P3
505.9885
17.29
C6H14O12P2
338.9888
9.81
C6H13O9P
259.0224
10.59
C6H13O9P
259.0224
11.28
C4H4O4
115.0037
U13C GTP U13C Fructose-1,6bisphosphate U13C Fructose-1phosphate U13C Hexose-6phosphates U13C Fumaric acid
GDP Glucose-1phosphate (G1P) Glucose-6phosphateb (G6P) GMP
21.20
C10H15N5O11P2
442.0171
7.60
C6H13O9P
259.0224
10.32
C6H13O9P
259.0224
18.35
C10H14N5O8P
GTP
22.67
Hexoses
2.27
IMP Isocitric
acida
13C4H4O4
119.0171
13C10H15N5O11P2
452.0506
13C6H13O9P
265.0426
13C6H13O9P
265.0426
362.0507
U13C GDP U13C Glucose-1phosphate U13C Hexose-6phosphates U13C GMP
13C10H14N5O8P
372.0843
C10H16N5O14P3
521.9834
U13C GTP
13C10H16N5O14P3
532.0169
C6H12O6
179.0561
U13C Hexoses
13C6H12O6
185.0762
13C10H13N4O8P
357.0734
18.08
C10H13N4O8P
347.0398
U13C IMP
15.00
C6H8O7
191.0197
U13C Citric acid
Lactic acid
2.70
C3H6O3
89.0244
U13C Lactic acid
13C3H6O3
92.0345
Malic acid
9.12
C4H6O5
133.0143
U13C Malic acid
13C4H6O5
137.0277
13C6H14O6
187.0919
13C6H13O9P
265.0426
13C5H11O8P
234.0287
13C3H5O6P
169.9852
Mannitol Mannitol-1phosphatea Mannose-6phosphateb (M6P) Oxaloacetic acida Pentose-5phosphates Phosphoenolpyruvic acid Pyruvic acid Sedoheptulose-7phosphate Succinic acid
2.12
C6H14O6
181.0718
7.47
C6H15O9P
261.0381
10.75
C6H13O9P
259.0224
12.47
C4H4O5
130.9986
11.13
C5H11O8P
229.0119
15.40
C3H5O6P
166.9751
3.57
C3H4O3
87.0088
C7H15O10P
289.0330
C4H6O4
117.0193
U13C Mannitol U13C Hexose-6phosphates U13C Hexose-6phosphates U13C Fumaric acid U13C Pentose-5phosphates U13C Phosphoenolpyruvic acid U13C Pyruvic acid U13C Sedoheptulose7-phosphate U13C Succinic acid
11.28 9.07
TMP
14.63
C10H15N2O8P
321.0493
U13C TMP
TTPa
20.76
C10H17N2O14P3
480.9820
U13C ATP
UDP
19.34
C9H14N2O12P2
402.9949
U13C UDP
8 ACS Paragon Plus Environment
13C3H4O3
90.0188
13C7H15O10P
296.0565
13C4H6O4
121.0328
13C10H15N2O8P
331.0829
13C9H14N2O12P2
412.0251
Page 9 of 18
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Analytical Chemistry
Compound
RT [min]
12C
metabolite
Uniformly 13C labeled internal standard
UMP
15.80
C9H13N2O9P
323.0286
U13C UMP
13C9H13N2O9P
ISTD [M-H]332.0588
UTP
21.11
C9H15N2O15P3
482.9613
U13C UTP
13C9H15N2O15P3
491.9914
Formula
[M-H]-
ISTD compound
ISTD formula
In case no 13C equivalent was available, a similar compound was used as internal standard (italic font). b G6P, F6P, and M6P were not baseline separated. For internal standardization the sum of the 13C peaks (U13C Hexose-6-phosphates) was used. a
Analytical figures of merit of targeted metabolomics A 1 µM multi-metabolite standard spiked with the yeast based
13C
internal standard was injected
five times over a period of 40 h in order to assess the intermediate repeatability of retention time, calculated concentrations and peak areas. The obtained results based on full MS data acquisition are shown in Table 2. The internal standard derived from yeast added to the calibration standard is at the same time a perfect mimic of a biological matrix. In spite of a gradient up to 100 mM and the high flow rate of 380 µL min-1, the intermediate repeatability regarding retention time, peak area and determined metabolite concentration was excellent. Except for GTP (1.3%), the relative standard deviation (RSD) of the retention time was below 1% (Table 2). 29 out of 45 metabolites could be quantified with an experimental repeatability RSD < 3%. Being based on area ratio measurement, the RSDs for quantification were significantly smaller than the RSDs observed for the peak areas, respectively. The implemented normal flow set-up showed a superior performance compared to capillary IC showing typically RSDs between 5 and 8% for intensities and retention times.14 Finally, in this work excellent LODs and LOQs in the low nM range were obtained (Table S-2). Overall, the analytical figures of merit of isotope dilution based on IC-high resolution MS were comparable to an earlier study based on LC-MS/MS.33 Table S-3 compares LODs, LOQs, correlation coefficients and the linear dynamic range of calibration for 8 metabolites investigated by both MS platforms using internal standardization by uniformly
13C
labeled yeast. A recent IC-triple
quadrupole MS study reported LOQs between 0.25 µM and 50 µM regarding the quantitative analysis of organic acids17 proving once more the validity of high resolution MS as alternative to MS/MS quantification providing an internal standard is used. Typical metabolomics experiments involve quantification of different metabolites varying over 4 orders of magnitude in complex matrices requiring internal standardization.
9 ACS Paragon Plus Environment
Analytical Chemistry
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Page 10 of 18
Table 2. RSDs (N=5) of retention times, calculated concentrations and peak areas for all compounds present in the 1 µM metabolite mix spiked with uniformly 13C labeled yeast cell extract as internal standard and measured five times over 40 hours.
Retention time
Calculated concentration
Peak area
RSD # of [%] metabo.
RSD # of [%] metabo.
RSD # of [%] metabo.
≤ 0.1
21
≤1
11
≤ 5.0
12
≤ 0.2
12
≤3
18
≤ 7.5
23
≤ 0.4
8
≤5
11
≤ 10
9
≤ 1.0
3
≤ 7.5
5
≤ 12
1
≤ 1.3
1
Non-targeted data evaluation IC-high resolution MS in combination with isotope dilution proofed to be a suitable platform for absolute quantification. The excellent figures of merit such as retention time stability (repeatability of < 1% as shown before) and mass accuracy (< 1.5 ppm for the target metabolite panel) qualify the method for simultaneous non-targeted analysis. Given that, the next step was the thorough investigation of non-targeted fingerprinting in samples spiked with yeast based
13C
internal
standards. First, compound annotation was tested by measuring a calibration solution containing the 45 metabolite panel (see Table 1) at a concentration of 1 µM spiked with
13C
internal standards
(concentrations ranging from low nM to several µM) in the ddMS2 mode. On the basis of MS2 spectra, a library search in mzCloud (HighChem LLC, Slovakia) was performed (level 2 compound annotation after nomenclature proposed by the Chemical Analysis Working Group of the Metabolomics Standards Initiative).34 All compounds present in the standard mix and contained in the mzCloud library at the time of analysis (Jan 11, 2017) were successfully annotated, showing that the acquisition of MS2 spectra was triggered in the ddMS2 mode despite the presence of internal standard. However, mzCloud annotation was restricted by four aspects: (1) co-elution (for the coeluting pentose-5-phosphates four isomers were suggested), (2) ambiguous isomeric compounds despite chromatographic separation (for the hexose-monophosphates six possible isomers were proposed with matching scores > 85 but higher scores are conveniently correlated with the correct isomeric structure), (3) in-source fragmentation (seven metabolites were annotated twice due to insource fragmentation resulting in the loss of a phosphate unit) and (4) library coverage (10 10 ACS Paragon Plus Environment
Page 11 of 18
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Analytical Chemistry
metabolites were not contained in mzCloud: FBP, isocitric acid, 6-phosphogluconic acid, CMP, cGMP, CDP, CTP, UTP, FBP, and 2PG). Summarizing, compound identification by comparison of MS2 spectra only (level 2), was not straightforward and demanded manual curation. Finally, only by chromatographic selectivity and measurement of standards level 1 identification can be achieved as level 1 metabolite identification requires at least two orthogonal data (e.g. retention time and mass spectrum) compared to a reference standard analyzed under the same conditions. Without reference standards, metabolites can only be putatively annotated (level 2) based on spectral comparison with libraries.34 The presence of
13C
labeled internal standards was not detrimental to the number of annotated
compounds in standards, despite the fact that some
13C
labeled components were present at µM
concentration. Next, compound annotation was considered in a proof of principle study addressing an in vitro cancer model. Comparative non-targeted analysis was performed using SW480 cancer cell preparations (1 x 106 cells seeded and extracted after 24 h incubation) with and without
13C
labeled internal standards. The added amount of yeast based 13C labeled standard was equivalent to the prior investigated multi-metabolite standard. As can be readily observed in Table 3, 41 compounds were identically annotated in samples with and without internal standard. The number of annotated compounds (mass accuracy 5 ppm, assignment threshold 70) that were not found either in the spiked cancer cell extract or in the pure extract or vice versa was very small. Table 3. Comparison of a cancer cell extract (preparation of 106 SW480 cells) with and without internal standard regarding putatively annotated compounds (level 2 annotation34). SW480 ISTD
SW480 no ISTD
# of mzCloud annotations
50
46
MS2 available in one sample type only
5
4
Features and MS2 available in one sample type only
4
1
Event
Based on MS/MS spectra acquired in the pooled sample, mzCloud search revealed a multitude of putatively annotated metabolites that were not included in the targeted analysis. These comprised e.g. sugar nucleotides (such as guanosine 5'-diphospho-ß-L-fucose, uridine 5'-diphosphogalactose, UDP-N-acetylglucosamine),
different
other
carboxylic
acids
(such
as
galacturonic acid,
pantothenic acid, 3,3-dimethylglutaric acid) and numerous modified amino acids (such as N-acetyl11 ACS Paragon Plus Environment
Analytical Chemistry
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Page 12 of 18
methionine, N-formyl-methionine, N-acetyl-1-aspartylglutamic acid). Independent of the presence of 13C internal standard, half of the metabolites that were putatively annotated (level 2) was affected by either in-source fragmentation or isomers (see Figure 1). These results confirm an earlier study.35 Thus, the importance of chromatographic selectivity and retention time as additional qualifier for peak annotation is evident.
Figure 1. Reliability of putatively annotated compounds of a cancer cell extracts (preparation of 106 SW480 cells) without internal standard found by mzCloud search. Comparison with standards revealed a number of identified false positives due to in-source fragmentation and due to isomers. Isomeric interference leads to features with more than one proposed structure in case the MS2 spectrum is not unique for one isomer.
Merging targeted metabolomics with non-targeted analysis The cancer cell line SW480 and its subline of acquired resistance was studied by the proposed workflow merging targeted absolute quantification with non-targeted fingerprinting. Five biological replicates of sensitive versus resistant SW480 cancer cell extracts (1 x 106 cells seeded and extracted after 24 h incubation) spiked with yeast based fully
13C
labeled internal standard upon
preparation were investigated. In order to prove the validity of the approach, additionally, samples prepared without internal standards were studied by non-targeted analysis. Figure S-2 gives a general schematic overview on measurements required for non-targeted data evaluation by the commercially available software Compound Discoverer 2.0. Feature detection and fold change (FC) calculations between the two sample groups were based on Full MS measurements. Data dependent acquisition of MS/MS spectra was performed on a pooled sample to allow spectral comparison with the library for compound annotation. The same Full MS data files of the samples with internal
12 ACS Paragon Plus Environment
Page 13 of 18
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Analytical Chemistry
standard were also used for the targeted quantification based on isotope dilution. Thereby, the measurement time and required sample volume could be significantly decreased. Isotope dilution IC-MS revealed concentrations ranging over nearly 4 orders of magnitude, from nM to several µM (5 nM – 50 µM, R2> 0.99) for the investigated metabolite panel. Scaled to the protein content, typical concentrations for organic acids were between 1 and 10 nmol mg-1 protein (approx. 0.8 to 8 µM in the measurement solution) whereas the concentration of sugar phosphates was in general lower (20 – 200 pmol mg-1 protein, 10 to 130 nM measured). A wide concentration range was observed in the case of nucleotides with several deoxynucleotides < LOD, up to 20 nmol mg-1 protein (14 µM measured) found for ATP (Table S-4). 16 metabolites showed an RSD ≤ 10% (N=5) and further 10 metabolites RSDs below 20% (N=5). The obtained biological repeatability was in the expected range for replicates of adherent cancer cell cultures extracted consecutively following an established protocol.30 Differences in absolute concentrations obtained by targeted IC-MS between the two cancer cell models are shown in Figure 2 and detailed results on all investigated metabolites are given in Table S-4.
Figure 2. Different concentrations in resistant and sensitive SW480 cancer cells measured by targeted IC-high resolution MS and quantified with yeast based fully 13C labeled internal standard. Error bars denote standard deviations and statistical significance (t-test) is indicated as follows: * p < 0.05, ** p < 0.01.
13 ACS Paragon Plus Environment
Analytical Chemistry
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Page 14 of 18
Figure 3. Principal component analysis of sensitive versus resistant SW480 cancer cell extracts (A) without internal standard and (B) with fully 13C labeled yeast cell extract used as internal standard (ISTD) in the same amount in all samples analyzed by anion-exchange chromatography high resolution mass spectrometry. For both preparations a clear separation could be obtained.
In non-targeted data evaluation, a clear clustering between resistant and sensitive SW480 cancer cells in principal component analysis (PCA) could be obtained for the cell extractions with and without internal standard (Figure 3). The median of the RSDs of all peak areas resulting from the fully 13C labeled isotopologues selected for the targeted quantification approach in all samples was 14%. Hence, successful clustering of the cancer cell groups was possible as all samples contain the same amount of internal standard and the internal standard is equally affected by sample preparation. Setting the criteria of peak area RSD < 15% in at least one of the sample groups, 459 and 574 aligned features (consisting of different isotopologues and adducts) were found in the samples without and with internal standard, respectively. 153 and 129 showed differences with an adjusted p-value of < 0.05 (Benjamini-Hochberg), respectively. These differences are displayed as heat maps in Figure S-3. Out of these differential compounds, several could be putatively annotated in both experimental settings, such as e.g. acetylated amino acids and other organic acids. From the putatively annotated compounds, which were not already included in the targeted analysis, Nacetyl-methionine and N-acetyl-glutamic acid showed the most significant decrease in resistant cells with fold changes (FC, ratio resistant/sensitive) of 0.3. On the other hand, galacturonic acid showed 14 ACS Paragon Plus Environment
Page 15 of 18
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Analytical Chemistry
a significant upregulation in resistant cells (FC 2). For pentose-phosphates (m/z 229.0119) more than one peak was putatively annotated indicating that ribose-1-phosphate is separated from the pentose-5-phosphates which were analyzed in the standard mix. All of those peaks are decreased in the resistant cells (FC 0.4 – 0.6). Using our novel IC-high resolution MS workflow for merging targeted and non-targeted metabolomics distinct metabolic phenotypes of drug-sensitive and resistant SW480 cells were observed.
CONCLUSIONS In this work, we introduced a novel IC-MS workflow that allows merging targeted quantification based on isotope dilution with non-targeted data analysis. By using the same data files for both approaches, the measurement time and sample amounts could be considerably decreased supporting high-throughput analysis. Fully 13C labeled extracts of the yeast Pichia pastoris proved to be ideally suitable for compound specific internal standardization of human derived samples extending the linear dynamic range to 4 orders of magnitude. The presented strategy based on anion chromatography and high resolution MS facilitating internal standardization is a promising approach to unravel changes in the central carbon metabolism and nucleotides in different biological systems.
ASSOCIATED CONTENT Supporting information Additional information as noted in the text. The Supporting Information is available free of charge on the ACS Publications website (http://pubs.acs.org).
AUTHOR INFORMATION Corresponding Author * E-mail:
[email protected] Author Contributions The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript. 15 ACS Paragon Plus Environment
Analytical Chemistry
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Page 16 of 18
Notes The authors declare no competing financial interest.
ACKNOWLEDGEMENTS We want to thank Petra Volejnik and Yasin El Abiead for continuous support and fruitful discussions. Thermo Fisher Scientific is acknowledged for providing the Dionex Integrion HPIC System and the Mass Spectrometry Center (MSC), Faculty of Chemistry, University of Vienna for providing mass spectrometric instrumentation. Lastly, Fellinger Krebsforschung is gratefully acknowledged for financial support.
REFERENCES (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16) (17) (18)
Rochat, B. TrAC Trends Anal. Chem. 2016, 84, Part B, 151–164. Cajka, T.; Fiehn, O. Anal. Chem. 2016, 88 (1), 524–545. León, Z.; García-Cañaveras, J. C.; Donato, M. T.; Lahoz, A. ELECTROPHORESIS 2013, 34 (19), 2762–2775. Vuckovic, D. Anal. Bioanal. Chem. 2012, 403 (6), 1523–1548. Ritter, J. B.; Genzel, Y.; Reichl, U. Anal. Biochem. 2008, 373 (2), 349–369. Armitage, E. G.; Barbas, C. J. Pharm. Biomed. Anal. 2014, 87, 1–11. Ivanisevic, J.; Zhu, Z.-J.; Plate, L.; Tautenhahn, R.; Chen, S.; O’Brien, P. J.; Johnson, C. H.; Marletta, M. A.; Patti, G. J.; Siuzdak, G. Anal. Chem. 2013, 85 (14), 6876–6884. Dunn, W. B.; Broadhurst, D.; Begley, P.; Zelena, E.; Francis-McIntyre, S.; Anderson, N.; Brown, M.; Knowles, J. D.; Halsall, A.; Haselden, J. N.; Nicholls, A. W.; Wilson, I. D.; Kell, D. B.; Goodacre, R.; Consortium, T. H. S. M. (HUSERMET). Nat. Protoc. 2011, 6 (7), 1060–1083. Siegel, D.; Meinema, A. C.; Permentier, H.; Hopfgartner, G.; Bischoff, R. Anal. Chem. 2014, 86 (10), 5089–5100. van Dam, J. C.; Eman, M. R.; Frank, J.; Lange, H. C.; van Dedem, G. W. K.; Heijnen, S. J. Anal. Chim. Acta 2002, 460 (2), 209–218. Kvitvang, H. F. N.; Kristiansen, K. A.; Bruheim, P. J. Chromatogr. A 2014, 1370, 70–79. Kiefer, P.; Nicolas, C.; Letisse, F.; Portais, J.-C. Anal. Biochem. 2007, 360 (2), 182–188. Alonso, A. P.; Piasecki, R. J.; Wang, Y.; LaClair, R. W.; Shachar-Hill, Y. Plant Physiol. 2010, 153 (3), 915–924. Wang, J.; Christison, T. T.; Misuno, K.; Lopez, L.; Huhmer, A. F.; Huang, Y.; Hu, S. Anal. Chem. 2014, 86 (10), 5116–5124. Hu, S.; Wang, J.; Ji, E. H.; Christison, T.; Lopez, L.; Huang, Y. Anal. Chem. 2015, 87 (12), 6371– 6379. Burgess, K.; Creek, D.; Dewsbury, P.; Cook, K.; Barrett, M. P. Rapid Commun. Mass Spectrom. 2011, 25 (22), 3447–3452. Petucci, C.; Zelenin, A.; Culver, J. A.; Gabriel, M.; Kirkbride, K.; Christison, T. T.; Gardell, S. J. Anal. Chem. 2016. Cipollina, C.; ten Pierick, A.; Canelas, A. B.; Seifar, R. M.; van Maris, A. J. A.; van Dam, J. C.; Heijnen, J. J. J. Chromatogr. B 2009, 877 (27), 3231–3236. 16 ACS Paragon Plus Environment
Page 17 of 18
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Analytical Chemistry
(19) Coulier, L.; Bas, R.; Jespersen, S.; Verheij, E.; van der Werf, M. J.; Hankemeier, T. Anal. Chem. 2006, 78 (18), 6573–6582. (20) Lu, W.; Clasquin, M. F.; Melamud, E.; Amador-Noguez, D.; Caudy, A. A.; Rabinowitz, J. D. Anal. Chem. 2010, 82 (8), 3212–3221. (21) Wu, J.; Zhang, Y.; Wiegand, R.; Wang, J.; Bepler, G.; Li, J. J. Chromatogr. B 2015, 1006, 167–178. (22) Luo, B.; Groenke, K.; Takors, R.; Wandrey, C.; Oldiges, M. J. Chromatogr. A 2007, 1147 (2), 153–164. (23) Büscher, J. M.; Czernik, D.; Ewald, J. C.; Sauer, U.; Zamboni, N. Anal. Chem. 2009, 81 (6), 2135– 2143. (24) Kiefer, P.; Delmotte, N.; Vorholt, J. A. Anal. Chem. 2011, 83 (3), 850–855. (25) Yanes, O.; Tautenhahn, R.; Patti, G. J.; Siuzdak, G. Anal. Chem. 2011, 83 (6), 2152–2161. (26) Neubauer, S.; Haberhauer-Troyer, C.; Klavins, K.; Russmayer, H.; Steiger, M. G.; Gasser, B.; Sauer, M.; Mattanovich, D.; Hann, S.; Koellensperger, G. J. Sep. Sci. 2012, 35 (22), 3091–3105. (27) Wahl, S. A.; Seifar, R.; ten Pierick, A.; Ras, C.; van Dam, J.; Heijnen, J.; van Gulik, W. In Metabolic Flux Analysis; Krömer, J. O., Nielsen, L. K., Blank, L. M., Eds.; Methods in Molecular Biology; Springer New York, 2014; pp 91–105. (28) Bennett, B. D.; Yuan, J.; Kimball, E. H.; Rabinowitz, J. D. Nat. Protoc. 2008, 3 (8), 1299–1311. (29) Miklos, W.; Pelivan, K.; Kowol, C. R.; Pirker, C.; Dornetshuber-Fleiss, R.; Spitzwieser, M.; Englinger, B.; van Schoonhoven, S.; Cichna-Markl, M.; Koellensperger, G.; Keppler, B. K.; Berger, W.; Heffeter, P. Cancer Lett. 2015, 361 (1), 112–120. (30) Dettmer, K.; Nürnberger, N.; Kaspar, H.; Gruber, M. A.; Almstetter, M. F.; Oefner, P. J. Anal. Bioanal. Chem. 2011, 399 (3), 1127–1139. (31) Klavins, K.; Neubauer, S.; Chalabi, A. A.; Sonntag, D.; Haberhauer-Troyer, C.; Russmayer, H.; Sauer, M.; Mattanovich, D.; Hann, S.; Koellensperger, G. Anal. Bioanal. Chem. 2013, 405 (15), 5159–5169. (32) Ortmayr, K.; Hann, S.; Koellensperger, G. Analyst 2015, 3465–3473. (33) Klavins, K.; Drexler, H.; Hann, S.; Koellensperger, G. Anal. Chem. 2014, 86 (9), 4145–4150. (34) Sumner, L. W.; Amberg, A.; Barrett, D.; Beale, M. H.; Beger, R.; Daykin, C. A.; Fan, T. W.-M.; Fiehn, O.; Goodacre, R.; Griffin, J. L.; Hankemeier, T.; Hardy, N.; Harnly, J.; Higashi, R.; Kopka, J.; Lane, A. N.; Lindon, J. C.; Marriott, P.; Nicholls, A. W.; Reily, M. D.; Thaden, J. J.; Viant, M. R. Metabolomics Off. J. Metabolomic Soc. 2007, 3 (3), 211–221. (35) Xu, Y.-F.; Lu, W.; Rabinowitz, J. D. Anal. Chem. 2015, 87 (4), 2273–2281.
17 ACS Paragon Plus Environment
Analytical Chemistry
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
FOR TOC ONLY
18 ACS Paragon Plus Environment
Page 18 of 18