Search tips
Search criteria 


Logo of acssdACS PublicationsThis JournalSearchSubmit a manuscript
Analytical Chemistry
Anal Chem. 2010 June 1; 82(11): 4558–4563.
Published online 2010 May 11. doi:  10.1021/ac100565b
PMCID: PMC2878175

NMR Method for Measuring Carbon-13 Isotopic Enrichment of Metabolites in Complex Solutions


Isotope-based methods are commonly used for metabolic flux analysis and metabolite quantification in biological extracts. Nuclear magnetic resonance (NMR) spectroscopy is a powerful analytical tool for these studies because NMR can unambiguously identify compounds and accurately measure 13C enrichment. We have developed a new pulse sequence, isotope-edited total correlation spectroscopy (ITOCSY), that filters two-dimensional 1H−1H NMR spectra from 12C- and 13C-containing molecules into separate, quantitatively equivalent spectra. The ITOCSY spectra of labeled and unlabeled molecules are directly comparable and can be assigned using existing bioinformatics tools. In this study, we evaluate ITOCSY using synthetic mixtures of standards and extracts from Escherichia coli. We show that ITOCSY has low technical error (6.6% for metabolites ranging from 0.34 to 6.2 mM) and can detect molecules at concentrations less than 10 μM. We propose ITOCSY as a practical NMR strategy for metabolic flux analysis, isotope dilution experiments, and other methods that rely on carbon-13 labeling.

Nuclear magnetic resonance (NMR) spectroscopy is one of the primary analytical tools used for investigating metabolites in complex biological extracts. Recent advances in NMR techniques,(1) data analysis software,2,3 and spectroscopic libraries of metabolite standards4,5 have considerably increased the number of metabolites that can be identified and quantified in routine studies. Currently, more than 80% of the signals observed in 2D 1H−13C spectra of biological extracts can be assigned,(1) and hundreds of spectra can be analyzed in a few hours.(2) However, these tools are primarily constrained to steady-state analyses of metabolites.

A wide variety of isotope-based methods have been developed for tracing metabolic pathways,(6) measuring reaction kinetics,(7) and quantifying molecules after extensive sample handling.(8) These techniques are of obvious utility to metabolomics but have not been widely applied to NMR-based studies because of complications related to signal overlap. NMR spectra of unenriched biological extracts can contain thousands of 1H resonances. Spectra of 13C-enriched extracts are further complicated by 1H−13C J-couplings. In isotope-based metabolomics studies, which involve complex mixtures of both 13C-labeled and unlabeled metabolites, NMR spectra are too heavily overlapped to support comprehensive quantitative analyses. As a result, studies that have employed isotope-based methods have been restricted to the subset metabolites with well-isolated resonances.911

To make isotope-based techniques more accessible to comprehensive metabolic analyses, we have developed a homonuclear 1H NMR experiment that separates signals from 12C- and 13C-containing molecules into distinct, quantitatively equivalent spectra. This pulse sequence, isotope-edited total correlation spectroscopy (ITOCSY), allows metabolomics studies to capitalize on isotope-based methods without increasing the complexity of spectra. ITOCSY is based on the established TOCSY pulse sequence,1214 which is desirable for metabolomics11,1519 because of its high sensitivity and compatibility with existing spectroscopic libraries and bioinformatics tools.25,20 We introduce the ITOCSY pulse sequence and evaluate its performance under both controlled and biologically relevant conditions. We show that ITOCSY is a reliable quantitative tool, and we foresee its potential utility in metabolic flux analysis, isotope dilution experiments, and other methods that rely on isotopic labeling.

Experimental Section

Strategy for Isotope Ratio-Based Quantification

This study uses an NMR adaptation of the traditional isotope dilution method for quantifying metabolites.(8) Isotope dilution studies derive concentration from the ratio of a metabolite’s signal to the signal from an isotope-labeled standard. Multiplying the observed isotope ratio by the amount of standard added allows one to calculate the amount of metabolite present in the sample at the time the standard was added.(8) In metabolomics studies, each observable metabolite is standardized against an isotopically enriched version of the same molecule. Alternatively, isotopically enriched extracts can be standardized against unlabeled molecules.

We used both the 12C- and 13C-standardized isotope dilution approaches in this study. To evaluate the qualitative performance and quantitative accuracy of ITOCSY-derived isotope ratios, we used 24 complex synthetic mixtures of unlabeled metabolites containing varying amounts of 13C-labeled molecules (Table S-1 in Supporting Information). We calculated the concentrations of the 13C-labeled metabolites by isotope dilution and compared them to the known concentrations of each molecule. We then applied ITOCSY to a metabolic investigation of the osmotic stress response in Escherichia coli. We describe how a mixture of 13C-labeled metabolites isolated from E. coli cells grown on [U−13C] glucose can be calibrated for use as a standard.

ITOCSY Pulse Sequence

ITOCSY data are collected as two 1H−1H DIPSI-TOCSY spectra,1214 one specific to 12C and the other isotopically nonspecific (12C + 13C). Signals from 13C-bound protons are removed from 12C-specific spectra by the use of 90°(1H)−1/41JHC−180°(1H,13C)−1/41JHC−90°(13C)−PFG filter elements. These low-pass J-filters convert antiphase 13C-bound 1H magnetization to undetectable double- and zero-quantum coherence.(21) ITOCSY employs three consecutive filtering elements with 1/41JHC delays tuned to aromatic (170 Hz), aliphatic (140 Hz), and anomeric (125 Hz) coupling constants (Figure (Figure1).1). Isotopically nonspecific spectra are collected using the same pulse sequence but with the carbon transmitters tuned 300 ppm off resonance to prevent the filtering elements from functioning. This approach ensures that the complementary spectra are collected under identical conditions. 12C + 13C and 12C data are collected in interleaved scans, and both utilize carbon decoupling to minimize spectral complexity. 13C-specific data are derived as a difference spectrum from the nonspecific (12C + 13C) and 12C-specific spectra.

Figure 1
Pulse program for acquiring ITOCSY. Filled and open bars indicate nonselective 90 and 180° hard pulses, respectively. Shaped 13C pulses denote 1 ms (200 ppm) hyperbolic secant inversions.(28) Spectra are acquired with the 1H carrier set on resonance ...

Preparation of Synthetic Mixtures

We evaluated the quantitative efficacy of ITOCSY by analyzing synthetic mixtures containing both 13C-labeled and unlabeled standards prepared at known concentrations. Synthetic mixtures (N = 24) contained 30 biologically relevant compounds, six of which included both U−13C-labeled and unlabeled species. These differentially labeled compounds were alanine, fructose, glucose, glutamate, glutamine, and lactate. All unlabeled molecules were prepared at 2 mM, and the concentrations of the U−13C compounds ranged from 130 μM to 6.5 mM (Table S-1 in Supporting Information). All synthetic samples were prepared in D2O containing 300 μM NaN3 (to inhibit microbial growth) and 300 μM DSS (4,4-dimethyl-4-silapentane-1-sulfonic acid, the NMR chemical shift standard). Samples were titrated with DCl/NaOD as needed to achieve an observed (glass electrode) pH of 7.400 ± 0.004.

In Vivo Isotopic Labeling Strategy

ITOCSY-based quantification requires isotopically labeled standards, but it is impractical to purchase the number of enriched compounds required for comprehensive metabolic analyses. For the osmotic stress study (see below), [U−13C] standards were produced in vivo from E. coli incubated in [U−13C]-glucose under the control condition. To quantify metabolites in the labeled extracts, a second set of unlabeled cultures was grown from the same starter in a medium containing unlabeled glucose. Metabolites in the unlabeled cultures were quantified using established methods,(1) and these levels were used as a benchmark for the labeled extracts. Regression coefficients used for quantifying enriched metabolites (see below) were derived from solutions of enriched extracts prepared at three dilutions (1:0, 1:1, 1:3) in D2O.

Osmotic Stress Study

As a biological application of ITOCSY, we investigated the metabolic response of E. coli incubated under varying levels of osmotic stress. Cultures (500 mL each; N = 4 per condition) of E. coli (MG1655) were grown in M9 media containing 22.8 mM glucose and variable concentration of salt: 0 (control), 150, 300, or 500 mM NaCl. To probe the dependence of the observed metabolic responses on osmolarity, we prepared additional cultures (N = 4) in M9 medium containing 500 mM NaCl and 10 mM glycine betaine, an established osmoprotectant.(22) Isotopically enriched cultures used for metabolite standards (250 mL each; N = 20) were grown in M9 medium containing 22.1 mM [U−13C]-glucose (Cambridge Isotope Laboratories). All cultures were incubated on a shaking platform at 37 °C. When cultures reached an optical density of 0.80 (660 nm), they were transferred to an ice−water bath to minimize metabolic activity. All subsequent sample preparation was conducted in a 4 °C cold room. Chilled samples (500 mL for natural abundance, 250 mL for 13C cultures) were centrifuged (18 500g), and the supernatant was discarded. Pellets were washed with 10 mL of glucose-free M9 media osmotically adjusted with NaCl to match each of the culture conditions. Washed pellets were recentrifuged, and the resulting pellets were flash frozen in liquid N2.

Preparation of Biological Extracts

Frozen E. coli pellets were resuspended in 16 mL of boiling water containing 250 μM MES. Samples were incubated in sealed reaction vials in a boiling water bath for 7.5 min, then centrifuged to remove cellular debris (8000g). The pellet was discarded, and extracts were microfiltered (Vivaspin 20; 3000 MWCO) to remove high molecular weight components. Filtrates were lyophilized and redissolved in 800 μL of D2O with 300 μM NaN3 and 300 μM DSS. Samples were titrated with DCl/NaOD as needed to an observed pH of 7.400 (±0.004). All 13C-labeled samples were pooled to create a single 13C-enriched library. Labeled standards were mixed with unlabeled samples (300 μL each) and transferred to 5 mm NMR tubes (Wilmad) for spectroscopic analysis.

NMR Spectroscopy

All NMR spectroscopy was conducted at the National Magnetic Resonance Facility at Madison. Two-dimensional 1H−1H ITOCSY spectra were collected on a 600 MHz Varian spectrometer equipped with a cryogenic probe. Spectra were collected with a 1.5 s repetition delay, 64 steady-state transits, 2 acquisition transits, 128 increments, and an acquisition time of 0.5 s (7530 points). Sweep widths for the direct and indirect dimensions were 7500 and 5400 Hz, respectively. 12C + 13C and 12C-specific spectra were collected concurrently using interleaved scans. Time-domain data were Fourier transformed with a shifted sine bell window function, zero-filled, phased, and referenced to DSS using automated NMRPipe(23) macros written in-house. 13C-specific difference spectra were derived from the 12C + 13C and 12C-specific data using custom NMRPipe macros.

Metabolite Quantification

All NMR data analyses were performed using the rNMR software package.(2) Metabolite signals in ITOCSY spectra were assigned using previously established methods.(1) Briefly, metabolites were identified by submitting peak lists to the Madison Metabolomics Consortium Database;(20) possible metabolite matches were verified by visually inspecting overlaid spectral standards from the BioMagResBank.(4) Dispersed resonances were selected for each of the assigned metabolites, and the peak intensities of these signals were measured.

Deriving accurate metabolite concentrations from multidimensional NMR signal intensities requires the use of calibration coefficients.(1) These isotope- and resonance-specific coefficients (m12, m13) are the linear regression slopes describing concentration as a function of intensity (Δconcentration/Δintensity) for metabolite signals in standards prepared at three concentrations. In the E. coli study, m12 was calculated from the signals of unlabeled standards prepared at 2, 5, and 10 mM; m13 was calculated from signals of 13C-enriched metabolite extracts prepared at 1:0, 1:1, and 1:3 dilutions. Calibration coefficients in the synthetic samples were calibrated independently using standards prepared at the minimum, median, and maximum concentration of each metabolite in the synthetic mixtures (Table S-1 in Supporting Information).

Metabolite concentrations were calculated from ITOCSY signals via three methods (eqs 13). The established fast metabolite quantification (FMQ) method,(1) which was used as a benchmark for E. coli study, determines the total concentration (Utot) of unlabeled (U12) and natural abundance 13C (Una) resonances from signal intensities (I12) observed in 12C-specific ITOCSY spectra (eq 1).

equation image

Metabolites were also quantified using the calibrated ITOCSY method (eqs 2 and 3), which relates unenriched signals (I12) to the corresponding resonances from [U−13C]-labeled molecules (I13). The calibrated ITOCSY approach can be used to either (eq 2) calculate concentrations of enriched molecules (U13) relative to known concentrations of unenriched standards (S12) or (eq 3) calculate concentrations of unenriched molecules (Utot) relative to concentrations of [U−13C] standards (S13). The equations used for calibrated ITOCSY calculations differ between S12- and S13-standarized experiments to account for the natural abundance levels of 13C (1.1%) in unenriched molecules. In this study, the S12 protocol (eq 2) was used for calculating metabolite concentrations in synthetic mixtures whereas the S13 protocol (eq 3) was used in the E. coli study.

equation image
equation image

Results and Discussion

We used complex mixtures of standards prepared with [U−13C]-labeled and unlabeled compounds (Table S-1 in Supporting Information) to investigate the efficacy of ITOCSY-based analyses. As expected, ITOCSY was effective for separating resonances on the basis of isotopic composition (Figure (Figure22 and Figure Figure3).3). We had hoped that quantification based on 13C/12C ratios for individual resonances would normalize the systematic quantitative defects inherent to multidimensional NMR(1) and eliminate the need for external signal calibration. Although enrichment-based quantification eliminated several major sources of systematic error, differential T1 relaxation and other isotope-related phenomena prevented accurate quantification from the uncalibrated ITOCSY signals.

Figure 2
Isotopically nonspecific (12C + 13C), difference edited (13C), and isotope-filtered (12C) 1D 1H−1H ITOCSY spectra of a mixture of unlabeled and uniformly 13C-labeled molecules. Signals in the 13C spectrum correspond to [U−13C]-glucose, ...
Figure 3
Isotope-filtered (12C) and difference edited (13C) 2D 1H−1H ITOCSY spectra of a synthetic mixture containing 30 unlabeled and six 13C-labeled metabolites. Resonance assignments for 13C-enriched compounds are shown. A complete listing of the unlabeled ...

We previously reported a general method for correcting systematic quantitative defects in NMR pulse sequences using empirically determined calibration coefficients.(1) ITOCSY-derived isotopic ratios were calibrated in this manner using standards prepared at multiple dilutions. As expected, the corrected 13C/12C values were quantitatively reliable; regression of 121 ITOCSY-derived concentrations versus known levels yielded a slope of 0.96 (ideal slope of 1) with R2 = 0.98. This translates to 6.6% average technical error for concentrations ranging from 0.34 to 6.2 mM (Figure (Figure44).

Figure 4
Metabolite concentrations (N = 121) measured by calibrated 2D 1H−1H ITOCSY vs known values. The dotted line indicates the ideal regression line (slope = 1). The data represent 24 synthetic mixtures; each mixture contained 30 metabolite standards, ...

To evaluate ITOCSY in the context of a biological study, E. coli cultures were prepared under varying levels of osmotic stress. Unlabeled extracts were mixed with a 13C metabolite library, and compounds were quantified using the FMQ(1) and corrected ITOCSY protocols. FMQ-based values were calculated from signals observed in 12C-specific ITOCSY spectra, whereas ITOCSY-based values were calculated from 12C/13C ratios. Regression of concentrations (N = 339; 0.009 to 11 mM) determined by ITOCSY versus FMQ showed a slope of 1.02 and an R2 of 0.99 (Figure (Figure5).5). Concentration-dependent error was observed, with values under 100 μM showing a substantial increase in error. These data indicate that the sensitivity limit of ITOCSY is less than 10 μM, but the limit for reliable quantification is 100 μM.

Figure 5
Metabolite concentrations (N = 339) and quantitative errors observed in 20 E. coli extracts as measured by calibrated ITOCSY vs the established FMQ protocol.(1) Each extract was mixed 1-to-1 with a 13C-enriched E. coli extract. FMQ values were based on ...

ITOCSY-based analyses of the E. coli extracts relied on isotopically enriched standards produced in vivo. These standards were quantified under the assumption that steady-state metabolite levels observed in labeled and unlabeled extracts are equivalent. The strong correlation observed between FMQ and ITOCSY values verifies this assumption. If steady-state levels differed between labeled and unlabeled samples, then compound-specific systematic error would have been observed.

One limitation of ITOCSY evident from the E. coli study is that reliable quantification requires both the labeled and unlabeled molecules to be present at concentrations above the detection limit. Six of the 24 metabolites quantified by FMQ were not observable in ITOCSY spectra of the enriched extracts. Four of these compounds were present at concentrations near the limit of detection for FMQ and were unobservable in enriched extracts because of the faster R1 relaxation of 13C-labeled molecules. The other two compounds, glucose and trehalose, were absent from the enriched library because these metabolites are only accumulated under high-salt conditions. The main biological findings of this study were salt-dependent increases in trehalose and glucose, diminished putrescine levels at high salt, and the reversal of these phenotypes by glycine betaine (Figure (Figure6).6). These findings, which had been previously reported,2427 were not quantifiable by isotope dilution because of our design of the in vivo labeling conditions. This limitation could be mitigated in future studies by including samples from each condition as standards.

Figure 6
Osmolarity-dependent alterations in metabolites observed in 2D 1H−1H ITOCSY spectra (12C-specific) of E. coli extracts. Each box shows a metabolite-specific region of an NMR spectrum; columns denote the four selected compounds; rows denote the ...


ITOCSY was effective in separating signals from labeled and unlabeled molecules. Although accurate quantification required calibration, the corrected signals had sufficiently low error to be useful for metabolomics studies. We foresee the primary applications of ITOCSY to be metabolic flux analysis, isotope dilution, and metabolite quantification using standards produced in vivo.

The in vivo labeling strategy used in this study has obvious applications beyond the limited case presented here. Enriched metabolites produced in cost-effective systems could be used to quantify extracts from unrelated species. The 13C-labeled metabolite library generated for this study, for example, could be used to quantify metabolites in human serum. The main limitation to this approach is that quantification requires metabolites to be present in both library and test samples. However, a more comprehensive library could be constructed using a variety of free-living organisms or under a range of conditions appropriate to a particular study.

The ITOCSY quantification protocol is an NMR adaptation of the traditional radioisotope dilution method for measuring metabolite concentrations.(8) Isotope dilution methods are advantageous because they can correct sample-to-sample variations in extraction efficiency, chromatography, and other sample handling errors if standards are added prior to any manipulation of the samples. Although extracts were analyzed as a complex solution in this study, the ITOCSY quantification protocol presented here allows samples to be fractionated and concentrated without affecting quantification.

Capitalizing on enriched libraries depends on our ability to quantify the labeled metabolites. ITOCSY provides a simple mechanism for achieving this; labeled compounds can be identified by using existing bioinformatics tools and quantified by using unlabeled standards. In summary, ITOCSY is a convenient tool for differentiating between molecules on the basis of enrichment and provides the framework for extending metabolomics into comprehensive isotope-based studies.


This work was supported by NIH Grant P41 RR02301 and by the DOE Great Lakes Bioenergy Research Center (DOE BER Office of Science DE-FC02-07ER64494). I.A.L. was the recipient of a fellowship from the NHGRI 1T32HG002760; NMR data were collected at the National Magnetic Resonance Facility at Madison (NMRFAM) funded by NIH Grants P41 RR02301 and P41 GM GM66326.

Funding Statement

National Institutes of Health, United States

Supporting Information Available

Supporting Information Available

ITOCSY pulse sequence for Varian, along with related macros for using the pulse sequence, is available from The rNMR analysis software and NMR data used in this study are available at A complete listing of metabolites and their concentrations in the synthetic mixtures and E. coli extracts can be found in the online Supporting Information. This material is available free of charge via the Internet at

Supplementary Material


  • Lewis I. A.; Schommer S. C.; Hodis B.; Robb K. A.; Tonelli M.; Westler W. M.; Sussman M. R.; Markley J. L. Anal. Chem. 2007, 79, 9385–9390. [PubMed]
  • Lewis I. A.; Schommer S. C.; Markley J. L. Magn. Reson. Chem. 2009, 47, S123–S126. [PubMed]
  • Xia J.; Bjorndahl T. C.; Tang P.; Wishart D. S. BMC Bioinf. 2008, 9, 507.
  • Markley J. L.; Anderson M. E.; Cui Q.; Eghbalnia H. R.; Lewis I. A.; Hegeman A. D.; Li J.; Schulte C. F.; Sussman M. R.; Westler W. M.; Ulrich E. L.; Zolnai Z.Pacific Symposium on Biocomputing; World Scientific Press: River Edge, NJ, 2007; pp 157−168. [PubMed]
  • Wishart D. S.; Tzur D.; Knox C.; Eisner R.; Guo A. C.; Young N.; Cheng D.; Jewell K.; Arndt D.; Sawhney S.; Fung C.; Nikolai L.; Lewis M.; Coutouly M. A.; Forsythe I.; Tang P.; Shrivastava S.; Jeroncic K.; Stothard P.; Amegbey G.; Block D.; Hau D. D.; Wagner J.; Miniaci J.; Clements M.; Gebremedhin M.; Guo N.; Zhang Y.; Duggan G. E.; Macinnis G. D.; Weljie A. M.; Dowlatabadi R.; Bamforth F.; Clive D.; Greiner R.; Li L.; Marrie T.; Sykes B. D.; Vogel H. J.; Querengesser L. Nucleic Acids Res. 2007, 35, D521–D526. [PubMed]
  • Vallino J. J.; Stephanopoulos G. Biotechnol. Bioeng. 1993, 41, 633–646. [PubMed]
  • Northrop D. B. Biochemistry 1975, 14, 2644–2651. [PubMed]
  • Rittenberg D.; Foster G. L. J. Biol. Chem. 1940, 133, 737–744.
  • Carvalho R. A.; Jeffrey F. M. H.; Sherry A. D.; Malloy C. R. FEBS Lett. 1998, 440, 382–386. [PubMed]
  • Lane A. N.; Fan T. W. M. Metabolomics 2007, 3, 79–86.
  • Massou S.; Nicolas C.; Letisse F.; Portais J. C. Metab. Eng. 2007, 9, 252–257. [PubMed]
  • Bax A.; Davis D. G. J. Magn. Reson. 1985, 65, 355–360.
  • Shaka A. J.; Lee C. J.; Pines A. J. Magn. Reson. 1988, 77, 274–293.
  • Braunschweiler L.; Ernst R. R. J. Magn. Reson. 1983, 53, 521–528.
  • Morvan D.; Demidem A.; Papon J.; Madelmont J. C. Magn. Reson. Med. 2003, 49, 241–248. [PubMed]
  • Sandusky P.; Raftery D. Anal. Chem. 2005, 77, 2455–2463. [PubMed]
  • Swanson M. G.; Keshari K. R.; Tabatabai Z. L.; Simko J. P.; Shinohara K.; Carroll P. R.; Zektzer A. S.; Kurhanewicz J. Magn. Reson. Med. 2008, 60, 33–40. [PubMed]
  • Zhang F.; Bruschweiler-Li L.; Robinette S. L.; Brüshweiler R. Anal. Chem. 2008, 80, 7549–7553. [PubMed]
  • Liu M.; Farrant R. D.; Nicholson J. K.; Lindon J. C. J. Magn. Reson. Ser. A 1995, 112, 208–219.
  • Cui Q.; Lewis I. A.; Hegeman A. D.; Anderson M. E.; Li J.; Schulte C. F.; Westler W. M.; Eghbalnia H. R.; Sussman M. R.; Markley J. L. Nat. Biotechnol. 2008, 26, 162–164. [PubMed]
  • Kogler H.; Sørensen O. W.; Bodenhausen G.; Ernst R. R. J. Magn. Reson. 1983, 55, 157–163.
  • Cayley S.; Record M. T. Biochemistry 2003, 42, 12596–12609. [PubMed]
  • Delaglio F.; Grzesiek S.; Vuister G. W.; Zhu G.; Pfeifer J.; Bax A. J. Biomol. NMR 1995, 6, 277–293. [PubMed]
  • Lewis B. A.; Cayley S.; Padmanabhan S.; Kolb V. M.; Brushaber V.; Anderson C. F.; Record M. T. J. Magn. Reson. 1990, 90, 612–617.
  • Measures J. C. Nature 1975, 257, 398–400. [PubMed]
  • Munro G. F.; Hercules K.; Morgan J.; Sauerbie W. J. Biol. Chem. 1972, 247, 1272–1280. [PubMed]
  • Record M. T.; Courtenay E. S.; Cayley D. S.; Guttman H. J. Trends Biochem. Sci. 1998, 23, 143–148. [PubMed]
  • Silver M. S.; Joseph R. I.; Hoult D. I. J. Magn. Reson. 1984, 59, 347–351.
  • Kupěe E.; Freeman R. J. Magn. Reson. Ser. A 1995, 115, 273–276.

Articles from ACS AuthorChoice are provided here courtesy of American Chemical Society