Use of hybrid capillary tube apparatus on 400 MHz NMR for quantitation of crucial low-quantity metabolites using aSICCO signal

Article abstract of DOI:10.1124/dmd.117.077073

Metabolites of new chemical entities can influence safety and efficacy of a molecule and often times need to be quantified in preclinical studies. However, synthetic standards of metabolites are very rarely available in early discovery. Alternate approaches such as biosynthesis need to be explored to generate these metabolites. Assessing the quantity and purity of these small amounts of metabolites with a nondestructive analytical procedure becomes crucial. Quantitative NMR becomes the method of choice for these samples. Recent advances in high-field NMR (>500 MHz) with the use of cryoprobe technology have helped to improve sensitivity for analysis of small microgram quantity of such samples. However, this type of NMR instrumentation is not routinely available in all laboratories. To analyze microgram quantities of metabolites on a routine basis with lower-resolution 400 MHz NMR instrument fitted with a broad band fluorine observe room temperature probe, a novel hybrid capillary tube setup was developed. To quantitate the metabolite in the sample, an artificial signal insertion for calculation of concentration observed (aSICCO) method that introduces an internally calibrated mathematical signal was used after acquiring the NMR spectrum. The linearity of aSICCO signal was established using ibuprofen as a model analyte. The limit of quantification of this procedure was 0.8 mM with 10 K scans that could be improved further with the increase in the number of scans. This procedure was used to quantify three metabolites—phenytoin from fosphenytoin, dextrophan from dextromethorphan, and 4-OH-diclofenac from diclofenac—and is suitable for minibiosynthesis of metabolites from in vitro systems.

Full text of DOI:10.1124/dmd.117.077073

DMD Fast Forward. Published on September 21, 2017 as DOI: 10.1124/dmd.117.077073
This article has not been copyedited and formatted. The final version may differ from this version.
DMD # 77073
Use of hybrid capillary tube apparatus on 400MHz NMR for quantitation of crucial low
quantity metabolites using aSICCO signal
AUTHOR(S):
Ranjeet Tiwari, Deepak Ahire, Hemantha Kumar, Sarmistha Sinha, Siddheshwar Kisan Chauthe, Murali
Subramanian, Ramaswamy Iyer, Ramakanth Sarabu and Lakshmikant Bajpai
Discovery Analytical Sciences, BMS-Biocon Research Center, Bangalore, India (R.T., H.K., S.K.C.,
R.S., L.B.)
Pharmaceutical Candidate Optimization, BMS-Biocon Research Center, Bangalore, India (D.A., S.S.,
M.S.)
Pharmaceutical Candidate Optimization, Bristol-Myers Squibb Company, Princeton, NJ, USA (R.I.)
1

DMD Fast Forward. Published on September 21, 2017 as DOI: 10.1124/dmd.117.077073
This article has not been copyedited and formatted. The final version may differ from this version.
DMD # 77073
Running title: Hybrid capillary tube for quant of metabolites on 400 MHz NMR.
Author contact information:
Lakshmikant Bajpai: lakshmikant.bajpai@syngeneintl.com
Discovery Analytical Sciences
Biocon Bristol-Myers Squibb Research and Development Center (BBRC),
Syngene International Ltd, Biocon Park
Plot No.2 & 3, Bommasandra IV Phase,
Jigani Link Road, Bangalore,
India - 560 099
Tel: 91-80 2808 4690
Fax: 91-80 2808 4023
Number of text pages: 22
Number of tables: 1
Number of Figures: 11
Number of references: 41
Number of words in abstract: 248
Number of words in introduction: 1012
Number of words in discussion: 1721
Number of figures in supplement: 8
2

DMD Fast Forward. Published on September 21, 2017 as DOI: 10.1124/dmd.117.077073
This article has not been copyedited and formatted. The final version may differ from this version.
DMD # 77073
Abbreviations:
Quantitative nuclear magnetic resonance (qNMR), Broad Band Fluorine Observe (BBFO), Centre
for Drug evaluation and Research (CDER), multiple reaction monitoring (MRM), Electronic
Reference To access In vivo Concentrations (ERETIC), artificial Signal Insertion for Calculation
of Concentration Observed (aSICCO), recombinant CYPs (rCYPs), Hydrophilic-Lipophilic-
Balanced (HLB), 3, 5-Dinitrobenzene (DNB), Liquid Chromatography-Mass Spectrometry (LC-
MS). Alkaline phosphatase (ALP)
3

DMD Fast Forward. Published on September 21, 2017 as DOI: 10.1124/dmd.117.077073
This article has not been copyedited and formatted. The final version may differ from this version.
DMD # 77073
Abstract
Metabolites of new chemical entities can influence safety and efficacy of a molecule and often
times need to be quantified in preclinical studies. However, synthetic standards of metabolites are
very rarely available in early discovery. Alternate approaches such as bio-synthesis needs to be
explored to generate these metabolites. Assessing the quantity and purity of these small amounts
of metabolites with a non-destructive analytical procedure becomes crucial. Quantitative nuclear
magnetic resonance (qNMR) becomes the method of choice for these samples. Recent advances
in high field NMR (>500 MHz) with the use of cryoprobe technology has helped to improve
sensitivity for analysis of small microgram quantity of such samples. However, this type of NMR
instrumentation is not routinely available in all laboratories. In order to be able to analyze
microgram quantities of metabolites on a routine basis with lower-resolution 400MHz NMR
instrument fitted with a BBFO room temperature probe a novel hybrid capillary tube set up was
developed. To quantitate the metabolite in the sample an aSICCO method that introduces an
internally calibrated mathematical signal was used after acquiring the NMR spectrum. The
linearity of aSICCO signal was established using Ibuprofen as a model analyte. The limit of
quantification of this procedure was 0.8 mM with 10K scans which could be improved further with
the increase in the number of scans. This procedure was used to quantify three metabolites-
phenytoin from fosphenytoin, dextrophan from dextromethorphan and 4-OH-diclofenac from
diclofenac and is suitable for mini-biosynthesis of metabolites from in vitro systems.
4

DMD Fast Forward. Published on September 21, 2017 as DOI: 10.1124/dmd.117.077073
This article has not been copyedited and formatted. The final version may differ from this version.
DMD # 77073
Introduction
The importance of identification and characterization of metabolites in early stage drug
development has been highlighted in the guideline on “Safety Testing of Drug Metabolites” issued
by the U.S. Food and Drug Administration Centre for Drug evaluation and Research (CDER)
(Timmerman et al., 2010; Briggs et al., 2014; Timmerman et al., 2016). This guidance suggests
that metabolites that circulate at 10% or more of total drug related material in humans at steady
state needs to be further characterized in safety studies, which may require re-analysis of IND-
enabling toxicity studies with quantitative determination of metabolite levels, or conducting fresh
safety studies to match the exposure levels of the metabolites observed in humans. These studies
are often conducted in the absence of synthetic metabolite standards. Additionally, proactive
assessment of metabolite levels in IND toxicity studies anticipated to be major in humans is a
common mitigation strategy.
Advances in mass spectrometry have contributed to the improvements in identification and
characterization of metabolites. However, getting the exact structure from a mass spectrum is still
challenging. For any LCMS-based quantification assay, for example multiple reaction monitoring
(MRM), a qualified metabolite reference standard is needed (Picà and BarcelÃ, 2008; Zhou and
Yin, 2016). Thus, many unknown metabolites detected in LCMS profiling experiments at
discovery stage cannot be quantified due to lack of suitable reference standards (Scalbert et al.,
2009). Synthesis of many metabolite standards is chemically challenging so alternate strategies
such as the biological synthesis of metabolites followed by their isolation and quantification has
become routine in drug discovery (Tao et al., 2009; Huttanus et al., 2016). Certification of these
biologically generated metabolites using the classical methods of analysis is challenging, since
there is very limited quantity available. So there is a need for a method of quantification of these
5

DMD Fast Forward. Published on September 21, 2017 as DOI: 10.1124/dmd.117.077073
This article has not been copyedited and formatted. The final version may differ from this version.
DMD # 77073
metabolites that should be sensitive and nondestructive so that recovered sample solutions could
be used as reference standard for subsequent analyses.
Nuclear magnetic resonance (NMR) spectroscopy is becoming an important tool for the
quantitation of small and large molecules (Bharti and Roy, 2012; Pauli et al., 2014). NMR has
excellent reproducibility and quantitative accuracy (Wishart, 2008; Nagana Gowda and Raftery,
2015). The response of the signal from NMR is proportional to molar concentrations of the analyte
and is independent of the type of other atoms present which makes it a more accurate quantitative
technique compared to other traditional analytical methods of quantitation. At the same time, it is
a non-destructive technique so analytes can be recovered after analysis for further use (Simmler et
al., 2014).
Quantitation with qNMR has been done by utilizing both an internal as well as an external standard
(Bharti and Roy, 2012; Pauli et al., 2014; Barding et al., 2012; Siddiqui et al., 2013; RundlÃf et
al., 2014; Simmler et al., 2014). There are several criteria for a compound to be used as NMR
internal standard: a) it should be available in pure form, b) it should have well resolved and
distinguished signal(s) from the analyte peak of interest, c) it should be inert when mixed with
analyte, and d) it should be nonvolatile (Saito et al., 2009; Weber et al., 2014). Use of external
reference standards for qNMR was helpful to overcome a few of these limitations like
compatibility and signal interference issues because in this method, standard and analyte sample
with the same volume is taken separately in identical precise NMR tubes and a spectrum is
acquired separately (Huang et al., 2014).
Another reported method for quantitation with NMR uses ERETIC (Electronic Reference To
access In vivo Concentrations) signal as a reference. Applications of ERETIC signal on low
6

DMD Fast Forward. Published on September 21, 2017 as DOI: 10.1124/dmd.117.077073
This article has not been copyedited and formatted. The final version may differ from this version.
DMD # 77073
resolution instruments is limited due to variation in response factor caused by variation of several
experimental factors (Barantin et al., 1997; Farrant et al., 2010; Liang et al., 2010; Esteve et al.,
2012; Jung et al., 2016). Another challenge in the ERETIC signal method is the insertion of the
reference signal into the acquisition data which requires specialized hardware and pulse sequences.
Walker et al. reported the use of a mathematically generated aSICCO signal (artificial signal
insertion for calculation of concentration observed) inserted post-acquisition as a reference for the
quantitation of metabolites. Any region in the analyte spectra can be selected for the insertion of
aSICCO reference signal. The magnitude of the inserted signal can be modified to appropriately
match the intensity of the sample resonances. The major limitation of Walker’s aSICCO signal
technique is that it uses high field magnets (500MHz and above with respect to 1H frequency) and
requires high sensitivity probe such as microprobe or cryoprobe (Walker et al., 2011; Walker et
al., 2014).
Another major challenge in the metabolite identification and quantification is the choice of
appropriate NMR solvents to get better resolution of signals. Use of incorrect solvent may lead to
improper signal with poor resolution or merging peaks. Generally, metabolite samples are very
limited in quantity and difficult to recover and reuse so there is no opportunity to try different
solvents to get the best possible resolution.
For improving the sensitivity in NMR experiments use of Shigemi NMR tubes have been reported
(Krunic et al; 2015) on 900 MHz instrument with cryoprobe. These tubes are matched in terms of
magnetic susceptibility to each of the solvents they are identified with. But there are certain
limitations with these tubes. These can only be used for the specified solvent, reported for the high
sensitivity NMR instruments, available only till 3 mm diameter, have no option to add another
solvent for screening, are very costly and do suffer with appearance of air bubbles between the
7

DMD Fast Forward. Published on September 21, 2017 as DOI: 10.1124/dmd.117.077073
This article has not been copyedited and formatted. The final version may differ from this version.
DMD # 77073
plunger and the solvent at high temperature which tends to destroy the field homogeneity (Jean-
Michael et al 2008). In order to extend the application of the aSSICO signal method for metabolite
quantitation we have developed a novel hybrid capillary tube set up which could be used on
400MHz NMR instrument with a room temperature BBFO probe. This configuration could also
be used to screen multiple NMR solvents for structure elucidation of crucial small quantity
metabolites using the same sample without recovery or contamination with uncommon solvents.
8

DMD Fast Forward. Published on September 21, 2017 as DOI: 10.1124/dmd.117.077073
This article has not been copyedited and formatted. The final version may differ from this version.
DMD # 77073
Materials and Methods
Chemicals and Reagents
Analytical grade Fosphenytoin, dextromethorphan, and diclofenac drug standards were purchased
from Sigma-Aldrich (St. Louis, USA). HPLC grade solvents and chemicals such as acetonitrile
(ACN), methanol, formic acid, malic acid, potassium dihydrogen phosphate (KH2PO4),
dipotassium hydrogen phosphate (K2HPO4) and tris (hydroxymethyl) amino methane (Tris buffer)
were purchased from Merck Specialties Private Limited (Mumbai, India). NADPH was purchased
from SISCO research laboratory Ltd. (Mumbai, India). Recombinant cytochrome P450s (rCYP
2D6 and 2C9) were purchased from Corning Life Sciences (Gurgaon, India). The rALP microsome
was prepared by in house methods (Subramanian M et al., 2017). Oasis HLB cartridges were
purchased from Waters India. All deuterated solvents such as DMSO-d6, Benzene-d6, and CDCl3
were purchased from Eurisotop Inc (Saclay, France). All other solvents and reagents were highest
grade commercially available materials and were used without further purification.
Equipment
The following equipment were used in these preparations unless otherwise noted: Eppendorf 5810
R centrifuge (Unit 4 112 Talavera Road, North Ryde NSW 2113Australia); Turbovap LV nitrogen
evaporator: (Caliper life sciences, 23 Mill Street Arcade, NY 14009United States); HLB cartridges
-Oasis HLB 1cc extraction cartridges (Milford, Massachusetts, United States); 100 µL X-type
syringe (CTC analytics AG, Zwingen, Switzerland); Cyclo-Mixer (Remi laboratory instruments,
Remi house, Mumbai, India). HPLC column SunFire C18 (Waters, Milford, MA 01757, USA),
HPLC column Inetrsil-ODS (GL Sciences Inc. Tokyo, Japan).
Preparation of calibration standards and internal standard solution
9

DMD Fast Forward. Published on September 21, 2017 as DOI: 10.1124/dmd.117.077073
This article has not been copyedited and formatted. The final version may differ from this version.
DMD # 77073
In this study, Ibuprofen was used as model analyte for checking the calibration and linearity of
aSSICO and traditional qNMR methods. 3, 5-Dinitrobenzene DNB was used as an internal
standard for quantification using the traditional method, whereas maleic acid was used for external
calibration using the aSICCO method. DNB is a nonvolatile compound, soluble in a variety of
solvents common to drug metabolism studies and has a sharp resonance at δ 8.9 ppm. A stock
solution of 3,5-dinitrobenzene (DNB) and ibuprofen were prepared by dissolving 6.12 mg and
10.6 mg solid in 5.00 mL (5.77mM) and 2.00 mL (25.7 mM) DMSO-d6, respectively. A series of
five standard solutions ranging from 0.80 mM (8.24 µg) to 12.85 mM (132.68 µg) concentration
of the Ibuprofen were prepared by serial dilution of the stock solution with DMSO-d6. Each sample
contained a 2.35 mM 3, 5 DNB acid as an internal standard. All solutions were vortexed and stored
in the refrigerator when not in use.
Generation and calibration of aSICCO Signal
NMR-SIM (Bruker TopSpin, Version 3.2) was used to create an aSSICO signal. A single pulse
sequence (zg) with uncoupled resonance was used to generate an artificial signal with 1s as
relaxation time. The chemical shift of the artificial signal was decided based on the spectral data
of the metabolite of interest. A line broadening of 1 Hz was used to reproduce the processing of
the isolated sample. Before each experiment, the aSSICO signal was calibrated with a known
concentration of a standard solution with constant parameters set for both known standard solution
and unknown sample data collection. Phase and baseline were manually corrected and the aSSICO
signal and signal from standard were integrated equally. An aSSICO signal relative concentration
was assigned using the equation below (Walker et al., 2011)
Conc of aSSICO signal = Conc.of Std x IC x NS
IS
NC
10

DMD Fast Forward. Published on September 21, 2017 as DOI: 10.1124/dmd.117.077073
This article has not been copyedited and formatted. The final version may differ from this version.
DMD # 77073
Where IC = Integral of aSSICO signal, IS = Integral of standard, NS = Number of protons in
standard, NC = Number of protons in sSSICO signal
The acquisition and processing parameters such as pulse length, relaxation delay, and sweep width
were kept identical for both analyte and the calibration data. After generating an artificial signal
and calibration data set, both the data points were merged with the help of macro in TopSpin. A
scaling factor which is variable within the macro is used to set aSSICO signal similar to that of
resonance signal. The chemical shift variable was also available in macro which allowed setting
of aSSICO signal chemical shift value as per the profile of the analyte.
General method for In vitro incubations to generate metabolites
In the general method, the parent compound (typically 30–100 µM) was incubated in an in vitro
system containing a source of drug metabolizing enzymes and cofactors necessary to effect the
biotransformation of interest (Table 1). The reaction volume was typically a multiple of 1 ml
depending on the requirement and turnover of the compound in the given matrix. All the
incubations were performed at 37°C in a shaking incubator at 60 rpm. At the end of the incubation,
the multiple reaction samples were pooled and passed through Oasis HLB (Hydrophilic-
Lipophilic-Balanced) cartridges to remove the matrix interference and concentrate the
metabolite(s) samples. A vendor (Waters) recommended protocol was used to process the samples
through HLB (Zhang X et al). Around 10 µL of the pooled sample was loaded on the Orbitrap for
the confirmation of the metabolite formation. The eluted samples from HLB cartridge were
subjected to nitrogen evaporation for reducing the sample volume. Resulting concentrated samples
were purified on analytical HPLC column. This general method for In vitro incubations to generate
metabolites was applied for fosphenytoin, dextromethorphan, and diclofenac drug standards. The
details of type of microsome and buffer conditions are given in (Table1)
11

DMD Fast Forward. Published on September 21, 2017 as DOI: 10.1124/dmd.117.077073
This article has not been copyedited and formatted. The final version may differ from this version.
DMD # 77073
HPLC conditions
Concentrated fractions from SPE after incubation were purified on Agilent 1200 series
chromatographic system (Santa Clara, CA, USA) equipped with a high-pressure quaternary pump
with PDA detector, heated column thermostat, and a Rheodyne universal loop injector with 100
µL injection capacity. Sample injection and data processing were controlled by Chemstation
software. The purifition of fosphenytoin, dextromethorphan metabolites were carried out on
Inertsil-ODS (250 mm×4.6 mm, 5 µm) column. The mobile phase consisted of 10 mM ammonium
acetate buffer in water (Solvent A) and acetonitrile (Solvent B) at a flow rate of 1ml/min. The
gradient started with 20% B and ramped linearly to 100% B over 25 minutes, held at 100% B for
5 minutes, returned to the initial condition over 1.0 minute, and allowed to equilibrate for 4.0
minutes. Diclofenac metabolites were purified on SunFire C18 (250 mm × 4.6 mm, 5 µm) column.
The mobile phase consisted of 0.1% TFA buffer in water (Solvent A) and acetonitrile (Solvent B)
at a flow rate of 1ml/min. The gradient started with 10% B and ramped linearly to 100% B over
25 minutes, held at 100% B for 5 minutes, returned to the initial condition over 2.0 minute, and
allowed to equilibrate for 4.0 minutes. All the metabolites were monitored at 220 nm and 254 nm
throughout the run time. During chromatography, the column temperature was maintained at room
temperature. Fractions containing the metabolite(s) of interest were combined, and the organic
solvent removed by nitrogen evaporation and the remaining aqueous sample was passed through
an HLB cartridge for a second time, following the same protocol. The eluted organic sample that
passed through the HLB cartridge was analyzed on the Orbitrap before nitrogen evaporation to
verify the sample integrity. Remaining organic samples were dried under nitrogen.
NMR Sample preparation and analysis
12

DMD Fast Forward. Published on September 21, 2017 as DOI: 10.1124/dmd.117.077073
This article has not been copyedited and formatted. The final version may differ from this version.
DMD # 77073
After SPE purification, the Eppendorf tubes containing the dried samples were left under vacuum
for a minimum of one hour. All the samples were reconstituted in 50 µL of high purity deuterated
DMSO (typically “100%” grade DMSO-d6 which is 99.96 % deuterium enriched) solvent. After
addition of solvent, sample tubes were sonicated for 2 min and vortexed on cyclomixer for 10 min.
The resulting clear solution was transferred to the melting point capillary tube with the help of 100
µL micro syringe. After filling capillary tube to two third of its size (50 µl), it was carefully placed
inside the regular 5 mm NMR tube filled with 600 µL of desired locking NMR solvent.
NMR spectra were recorded on Bruker 400-MHz spectrometer (Bruker BioSpin Corporation,
Billerica, MA) controlled by TopSpin software (version 3.2) and equipped with a 5 mm BBFO
probe. One-dimensional proton spectra were recorded with a sweep width of 12,000 Hz, 32K data
point and a relaxation delay of 2 sec. For calibration sample and the metabolite samples, 10K and
15K scans were accumulated respectively. All spectra were referenced using residual DMSO-d6
(1H, δ 2.5ppm). Phasing, baseline correction, and integration was all performed manually. qNMR
was performed using the aSICCO method (Walker, et al., 2011). For aSICCO method, a standard
of known concentration of maleic acid was used as a calibrant. The concentration of the analyte
was calculated as per the formula given below:
mM of Analyte/ Conc of aSSICO signal = mM of IS/ Conc of Std x IA x NS
IS
NA
IA = Integration of analyte protons (aSICCO signal)/ Integral of aSSICO signal
IS = Integration of internal standard protons/ Integral of standard,
NA = Number of protons in analyte/ Number of protons in standard,
NS = Number of protons in internal standard/ Number of protons in sSSICO signal
13

DMD Fast Forward. Published on September 21, 2017 as DOI: 10.1124/dmd.117.077073
This article has not been copyedited and formatted. The final version may differ from this version.
DMD # 77073
Quantitation of 4-hydroxy diclofenac metabolite by LC/MS/MS
All prepared samples were analyzed using an Acquity UPLC system coupled to an API 5400
quadrupole ion trap (Applied Biosystems, Canada).The chromatographic separation of analytes
was achieved on an Acquity BEH C18 column (2.1 X 50 mm, 1.7 µm) using a binary gradient
mobile phase consisting of 0.1% formic acid water (Mobile phase A) and acetonitrile (mobile
phase B) with a flow rate of 800 µL/min. Initially the column was held at 95% of A for 0.8 min
and then linearly ramped to 50% of B over 0.4 min, followed by linear ramping to 90% of B over
0.4 min held at these conditions for next 0.5 min. The system was then returned to in initial
conditions and equilibrated for next 0.5 min. The total chromatographic run time was 3 min. Mass
spectroscopic analysis was performed with an electrospary ionization (ESI) source operated in
negative ion mode with multiple reaction monitoring (MRM). The ESI source temperature was set
at 6000C and ion spray voltage, cell entrance (EP) and exit potentials (CXP) were set at -4500, -
10 and -10 volts, respectively. The curtain gas, nebulizer gas (Gas1) and heater gas (Gas2) were
set at 40, 30 and 40, respectively. The quantification of analyte was achieved by monitoring m/z
312.2→230.2 (DP -70 and CE -35) for 4-OH diclofenac. The Analyst software (version 4.2) by
Applied Biosystems was used for data acquisition and data analysis.
14

DMD Fast Forward. Published on September 21, 2017 as DOI: 10.1124/dmd.117.077073
This article has not been copyedited and formatted. The final version may differ from this version.
DMD # 77073
Results
Ibuprofen was selected as a model analyte for comparing the linearity and dynamic range of
classical internal standard qNMR method using 3, 5-dinitrobenzene (DNB) as an internal standard
vs the aSICCO signal as internal standard. 3, 5-dinitrobenzene (DNB) was selected as an internal
1
standard due to its easy commercial availability. H NMR spectra of ibuprofen dissolved in
DMSO-d6 were recorded at various concentrations from 0.8 mM (8.24 µg) to 12.85 mM (132.68
µg). For each spectrum, 10k scans were acquired. Each sample contained 2.35 mM of DNB as an
internal standard. For classical qNMR method using DNB as an internal standard, the linearity
calculations were performed using signals at δ7.10 and δ8.90 for the aromatic protons of ibuprofen
and the internal standard DNB, respectively. For the aSICCO method, the signals at δ7.10 and
δ15.0 for aromatic proton of ibuprofen and artificially inserted aSICCO signal were used for
linearity calculations, respectively. In both cases, the response was linear with correlation
coefficients greater than 0.999 and intercepts not significantly different from 0 (Figure 1 and 2).
To demonstrate the utility of the aSICCO approach, quantitative estimation of major metabolites
of Fosphenytoin, Dextromethorphan, and Diclofenac drug were performed using aSICCO as
reference standard with 15 K scans so as to improve the sensitivity for the metabolite signal. Maleic
acid standards used to calibrate the aSICCO signal were also acquired with 15 K scans. Reported
methods of incubations were used to generate major metabolite for each of these drugs and
metabolites were then isolated by SPE and HPLC chromatography and were concentrated for
NMR analysis.
Phenytoin Metabolite
15

DMD Fast Forward. Published on September 21, 2017 as DOI: 10.1124/dmd.117.077073
This article has not been copyedited and formatted. The final version may differ from this version.
DMD # 77073
Fosphenytoin is a phosphate ester derivative of phenytoin that is susceptible to hydrolysis and
forms phenytoin as a major metabolite (Fischer et al., 2003) Fosphenytoin was incubated with
rALP to generate phenytoin as a single metabolite (Figure 3 and Supplemental Figure S1). As per
the UV chromatogram 90 % conversion of parent to phenytoin was observed. The isolated
metabolite was confirmed by matching the retention time (RT) and mass with the known standard.
The overall recovery of phenytoin metabolite was found to be 84% on w/w basis. The phenytoin
metabolite has been previously characterized using 1H NMR, MS and tandem mass spectrometry
techniques (Annesley et al., 2001). Crude material obtained after drying was reconstituted in
1
DMSO-d6 and H-NMR was recorded, simultaneously 0.45 mM of the maleic acid standard in
DMSO-d6 was also acquired to calibrate the aSICCO signal (Figure 4). The aromatic proton of
phenytoin at δ 7.1 with integral of 4.3 and aSICCO signal at δ 14 with integral of 1 was used for
qNMR calculations. The calculated concentration of the isolated metabolite was found to be 1.5
mM by the aSICCO qNMR method.
Dextrorphan metabolite
Dextromethorphan is rapidly converted into the active metabolite dextrorphan by the cytochrome
P450 enzyme CYP2D6 (Schadel et al., 1995). After in vitro incubation using the general procedure
described in experimental section, dextrorphan was isolated using analytical HPLC (Figure 5). As
per the UV chromatogram, 65 % conversion was observed for dextrorphan metabolite from the
parent drug during in vitro incubation (Supplemental Figure S2). The isolated metabolite was
confirmed by matching the RT and mass with the known standard. The overall recovery of pure
1
dextrorphan metabolite was found to be 69.7 % on w/w basis. H NMR was recorded for the
1
isolated metabolite in DMSO-d6. Simultaneously, H NMR of 8.14 mM of maleic acid standard
was also acquired to calibrate the aSICCO signal (Figure 6). The aromatic proton of dextrorphan
16

DMD Fast Forward. Published on September 21, 2017 as DOI: 10.1124/dmd.117.077073
This article has not been copyedited and formatted. The final version may differ from this version.
DMD # 77073
at δ 6.8 with integral of 0.58 and aSICCO signal at δ8.5 with integral of 1 was used for qNMR
calculations. The calculated concentration of the isolated metabolite was found to be 3.33 mM by
qNMR using the aSICCO method.
4-Hydroxy diclofenac
Diclofenac was metabolized to 4-hydroxy diclofenac by polymorphic CYP2C9 (Dorado et al.,
2003). After in vitro incubation using general procedure described in experimental section, 4-
hydroxy diclofenac was isolated using analytical HPLC (Figure 7). As per the UV chromatogram
80 % conversion was observed for 4-hydroxy diclofenac from the parent drug during in vitro
incubation (Supplemental Figure S3). The overall recovery of pure 4-hydroxy diclofenac was
found to be 76 % on w/w basis.
1
1H NMR was recorded for the isolated metabolite in DMSO-d6. Simultaneously H NMR was
acquired for 7.28 mM of the maleic acid standard to calibrate the aSICCO signal (Figure 8) The
aromatic proton of 4-hydroxy diclofenac at δ 6.1 with integral of 1 and aSICCO signal at δ14 with
integral of 1.17 was used for qNMR calculations. The calculated concentration of isolated
metabolite was found to be 4.13 mM by qNMR using the aSICCO method.
17

DMD Fast Forward. Published on September 21, 2017 as DOI: 10.1124/dmd.117.077073
This article has not been copyedited and formatted. The final version may differ from this version.
DMD # 77073
Discussion:
The utility of NMR for metabolite structural characterization has been well established (Forseth et
al., 2011; Caceres-Cortes and Reily, 2010). In recent years, with the advent of high field instrument
and availability of cryo probes the ability of NMR to quantify metabolite in low microgram to
nanogram quantities is possible (Spraul et al., 2003; Molinski et al 2010). However, these types of
high end instrumentation are not available in all the laboratories. The work presented here provides
some practical solution to utilizing a 400 mHz NMR with a 5 mm probe to conduct metabolite
characterization and quantitation. A major limitation for the quantitation of drug metabolites with
a 400 MHz NMR instrument is sensitivity, as the amount of metabolite generated from in vitro
incubations is typically in the range of microgram to nanogram quantities.
Sample concentration could be enriched by carrying out multiple incubation reactions but isolating
the required metabolite from large incubation volumes is a challenging task. Another way of
increasing the sensitivity would be the use of microprobe designed for small volume samples, such
as a 1 mm probe, where sensitivity is increased by improved filling factors. Another advantage of
such a probe is that the smaller sample volume (~ 12 µL) reduces the interference noise from the
NMR solvent and residual water, which in turn enhances the analyte signal-to-noise (Schlotterbeck
et al., 2002). However, this option requires dedicated high cost probes which may be a limiting
factor in many laboratories. Earlier reports proved that the use of 1.7 mm id NMR tube for
metabolite analysis was helpful to reduce the interference/noise from NMR solvent and residual
water, yet the sensitivity gain from using a smaller diameter NMR tube in a 5 mm probe was
insignificant since the rf coil was further away from the sample and interference from the air was
compensating all the gain (Johnson et al. 2010; Schlotterbeck et al., 2002). Based on this available
information, a unique hybrid capillary tube set up was developed which uses a 1 mm capillary tube
18

DMD Fast Forward. Published on September 21, 2017 as DOI: 10.1124/dmd.117.077073
This article has not been copyedited and formatted. The final version may differ from this version.
DMD # 77073
within a 5 mm outer NMR tube, in a 5 mm room temperature probe. With this strategy, capillary
tubes were used as sample reservoir needing only 20-30 l of the sample solution compared to
600-700 l sample in 5 mm tubes with data acquired on a 5 mm probe. Practically this should
increase the actual sample concentration and sensitivity by multiple folds but increased noise was
observed in the spectra collected probably created by the air gap between the rf coil and the sample
solution (Figure 9). To reduce the noise due to the air gap, the capillary tube containing 20-30 L
of the sample solution in dueterated solvent was placed inside a 5 mm tube containing 600 µL of
same NMR solvent (Figure 9). This theoretically should reduce the interference due to air gap
between the rf coil and the sample while maintaining the sensitivity gain due to increased sample
concentration. To confirm this, NMR spectra of Ibuprofen was acquired in DMSO-d6 in capillary
tube inserted in two separate 5 mm NMR tubes- one containing DMSO in the outer tube and other
without solvent in the outer tube. Signal to noise ratio (S/N) was found to be 301 when the outer
tube was filled with DMSO and 126 (Supplemental Figures S4A, S4B and S5) when outer tube
was empty which supported the proposed hypothesis.
Linearity samples were acquired using 2 sec relaxation delay with 10000 scans and it was observed
that the integration values for the all the characteristic signals were linear. For a fully relaxed
proton AQ+D1 should be > 5 x T1 for the resonances being measured (Edwin D, 2000). T1 for the
test analyte Ibuprofen was measured and was found to be 1.3 sec. Acquisition time (AQ) for this
experiment was 1.99 sec and delay time (D1) was 2 sec under the experimental conditions used
which suggested a total of relaxation delay as 3 times of T1. Usually signal amplitude drops as the
recycle time (relaxation + acquisition time) gets shorter but there are two choices to compensate
the loss: (1) either to cut down recycle time and accumulate more scans with a lower signal
recovery for each scan , or (2) use a longer recycle time and fewer scans with each scan recovering
19

DMD Fast Forward. Published on September 21, 2017 as DOI: 10.1124/dmd.117.077073
This article has not been copyedited and formatted. The final version may differ from this version.
DMD # 77073
nearly 100% of the signal. Increasing the relaxation delay would have increased the total run time
significantly so loss of signal amplitude with lower relaxation delay was compensated by
increasing the number of scans to 15000 for metabolite samples and the aSICCO standards with 2
sec relaxation delay and the total acquisition time was 21 hrs. In order to confirm if 2 sec of delay
time would have major impact on the quality of the data we have acquired two data at 2 and 10
sec relaxation delay and the s/n in both cases were found to be 1912 and 1989 respectively
(Supplemental Figure S6 and S7). Even relative integration of all protons were also similar under
both conditions as shown in figures S6 and S7 in supplementary data. Similar parameters were
used for both analytes and the standard used for calibrating the aSICCO signal so effective loss in
signal amplitude due to shorter relaxation delay should be compensated.
To avoid the breaking of capillary tube, the NMR acquisitions were performed without rotation
of the sample tube. With this new hybrid capillary tube appratus, we were able to achieve the
quantitation limit up to low µg level on 400 MHz instrument with room temperature probe. To test
the utility and accuracy of this hybrid capillary tube apparatus quantitation of Ibuprofen was
1
performed with a standard 400 MHz NMR using the aSICCO approach. H NMR spectra of
Ibuprofen at different concentrations were aquired using identical instrumental parameters. In all
cases, the response of the analyte was linear.
Traditionally the incubated samples are being purified using chromatographic techniques prior to
NMR acquisition (Walker et al., 2014). However, prior to introducing the samples on to the column
it need to be concentrated. Initial attempts to concentrate the sample by rotary evaporation,
Genevac or lyophilization led to degradation of metabolites (Supplemental Figure S8). Apparently
the presence of water in the sample took much longer time to concentrate which caused the
degradation. It was confirmed by LC/MS analysis of the samples during multiple stages of
20

DMD Fast Forward. Published on September 21, 2017 as DOI: 10.1124/dmd.117.077073
This article has not been copyedited and formatted. The final version may differ from this version.
DMD # 77073
purification. To avoid this degradation, after incubation sample was passed through SPE cartridges
to remove the biological matrix and the water from the sample. A similar SPE concentration step
was introduced after the collection of HPLC fraction to remove the water and buffer. Eluent from
SPE was dried down under nitrogen and final residue was reconstituted in 50 µl of NMR solvent
for acquisition (Figure 10). Overall sample recovery in this process was found to be between 76-
86%. SPE steps were not completely optimized for this study as we used the commercial samples
and it is well known that recovery from SPE is compound dependent and can vary from sample to
sample. So there is ample opportunity to further improve the sensitivity for actual study samples
by optimizing the SPE step and thereby improve the overall recovery.
Three drugs fosphenytoin, dextromethorphan, and diclofenac were incubated to generate the
specified metabolites for quantification as per the protocol specified in experimental section. The
corresponding metabolites phenytoin, dextrorphan and 4-Hydroxy diclofenac, were isolated with
optimized protocol described above. qNMR was performed with the aSSICO method for
metabolite quantitation. The calculated concentration of 4-hydroxydiclofenac obtained by aSICCO
approach (4.59 mM) was comparable with quantification by LCMS (4.61 mM) technique which
supports the accuracy of this method.
It was observed that due to a larger volume of solvent (DMSO-d6) in the outer tube (~600µL)
compared to the quantity of the sample in the capillary tube , DMSO-d6 and residual water peaks
were suppressing the peak of interest from analyte in the aliphatic region. This could be a limitation
of this method if a proton in aliphatic region is needed for quantification of desired metabolite. To
address this concern other dueterated solvents like benzene, pyridine or chloroform, which lack
aliphatic protons, could be used for sample preparation instead of DMSO but these solvents may
not be ideal if the recovered sample has to be further used as reference standard. With the hybrid
21

DMD Fast Forward. Published on September 21, 2017 as DOI: 10.1124/dmd.117.077073
This article has not been copyedited and formatted. The final version may differ from this version.
DMD # 77073
capillary tube set up, however, these deuterated solvents could be used in the outer 5 mm tube
without contaminating the actual sample in capillary tube. This could help to isolate the aliphatic
peaks without contaminating the actual sample with undesired solvents in the outer tube. To
confirm this ibuprofen was selected as model analyte and the NMR sample was prepared in
DMSO-d6. Different solvents like benzene-d6, pyridine-d5 and CDCl3 were used in the outer 5 mm
tube and the NMR spectra of all these samples were acquired using the same parameters (Figure
11). Spectra acquired with benzene-d6 and pyridine-d5 in outer NMR tube showed better
separation of protons 7 and 9 of Ibuprofen which were merging with water peak in DMSO-d6 data.
Spectra acquired in deuterated chloroform showed clearly separated signals for H7 and H9 which
could be utilized for the quantitation if needed. This clearly indicated the effect of NMR solvent
on the separation of signals from solvent peaks and suggested that this strategy could be
successfully applied when there is a need for the separation of signals in the aliphatic region
without contaminating the actual sample with these uncommon solvents. This could also be used
if data is needed for the structural confirmation of the small quantity of crucial isolated metabolite.
In conclusion, we have developed a novel hybrid capillary tube apparatus which can be used to
quantitate metabolites in the µg level concentration with a 5 mm room temperature NMR probe
on a 400 MHz instrument. With the use of an aSICCO signal as the internal standard the recovered
NMR sample could be further utilized as reference standard for biological quantitation. At the
same time screening of different NMR solvents can be done in the outer tube to avoid
contaminating the actual sample in the capillary tube or compromising the data quality. This novel
set up gives the opportunity to perform metabolite characterization and quantifications with regular
room temperature probes in 400 MHz frequency spectrometers without adding any extra resources
for increasing the sensitivity of the instruments.
22

DMD Fast Forward. Published on September 21, 2017 as DOI: 10.1124/dmd.117.077073
This article has not been copyedited and formatted. The final version may differ from this version.
DMD # 77073
Acknowledgements:
We would like to thank DAS purification team members, Venkatesh T K, Vinay Kempegowda
and Sudam Mane for their support in collecting MS data for the isolated material. We would also
thank Michael Reily (BMS, LVL, USA) for review of this manuscript and sharing his suggestions.
We would also like to thank Syngene International Limited for proving all infrastructure support
to do this work.
23

DMD Fast Forward. Published on September 21, 2017 as DOI: 10.1124/dmd.117.077073
This article has not been copyedited and formatted. The final version may differ from this version.
DMD # 77073
Authorship Contributions
Participated in research design: Ranjeet Tiwari, Deepak Ahire, Hemantha Kumar, Sarmishtha
Sinha, Murali Subramanian, Ramaswamy Iyer, Ramakanth Sarabu and Lakshmikant Bajpai
Conducted experiments: Ranjeet Tiwari, Deepak Ahire, Hemantha Kumar
Performed data analysis: Ranjeet Tiwari, Deepak Ahire, Hemantha Kumar, Murali
Subramanian, Ramaswamy Iyer, Ramakanth Sarabu and Lakshmikant Bajpai
Wrote or contributed to the writing of the manuscript: Siddheshwar Kisan Chauthe, Ranjeet
Tiwari, Deepak Ahire, Hemantha Kumar, Murali Subramanian, Ramaswamy Iyer, Ramakanth
Sarabu and Lakshmikant Bajpai
24

DMD Fast Forward. Published on September 21, 2017 as DOI: 10.1124/dmd.117.077073
This article has not been copyedited and formatted. The final version may differ from this version.
DMD # 77073
References
Annesley TM, Kurzyniec S, Nordblom GD, Buchanan N, Pool W, Reily M, Talaat R and Roberts WL
(2001) Glucuronidation of prodrug reactive site: isolation and characterization of
oxymethylglucuronide metabolite of fosphenytoin. Clin Chem 47: 910-918.
Barantin L, Pape AL and Akoka S (1997) A new method for absolute quantitation MRS metabolites. Magn
Reson Med 38: 179-182.
Barding GA, Jr., Salditos R and Larive CK (2012) Quantitative NMR for bioanalysis and metabolomics.
Anal Bioanal Chem 404: 1165-1179.
Bharti SK and Roy R (2012) Quantitative 1H NMR spectroscopy. Trends Anal Chem 35:5-26.
Briggs RJ, Nicholson R, Vazvaei F, Busch J, Mabuchi M, Mahesh KS, Brudny-Kloeppel M, Weng N,
Galvinas PA, Duchene P, Hu P and Abbott RW (2014) Method transfer, partial validation, and
cross validation: recommendations for best practices and harmonization from the global bioanalysis
consortium harmonization team. AAPS J 16: 1143-1148.
Caceres-Cortes, J., and Reily, MD. (2010) NMR spectroscopy as a tool to close the gap on metabolite
characterization under MIST. Bioanalysis 2: 1263-1276.
Dorado P, Berecz R, Cáceres MC and Llerena An (2003) Analysis of diclofenac and its metabolites by
high-performance liquid chromatography: relevance of CYP2C9 genotypes in diclofenac urinary
metabolic ratios. J Chromatogr B 789: 437-442.
25

DMD Fast Forward. Published on September 21, 2017 as DOI: 10.1124/dmd.117.077073
This article has not been copyedited and formatted. The final version may differ from this version.
DMD # 77073
Edwin DB (2000) High Resolution NMR- Theory and chemical applications, 3rd ed. Academic Press, New
York.
Esteve V, Celda B and MartÃnez-Bisbal MC (2012) Use of 1H and 31P HRMAS to evaluate the
relationship between quantitative alterations in metabolite concentrations and tissue features in
human brain tumour biopsies. Anal Bioanal Chem 403: 2611-2625.
Farrant RD, Hollerton JC, Lynn SM, Provera S, Sidebottom PJ and Upton RJ (2010) NMR quantification
using an artificial signal. Magn Reson Chem 48: 753-762.
Fischer JH, Patel TV and Fischer PA (2003) Fosphenytoin: clinical pharmacokinetics and comparative
advantages in the acute treatment of seizures. Clin Pharmacokinetics 42: 33-58.
Forseth RR and Schroeder FC (2011) NMR-spectroscopic analysis of mixtures: from structure to function.
Current opinion in chemical biology, 15: 38-47.
Huang Y, Su B-N, Ye Q, Palaniswamy VA, Bolgar MS and Raglione TV (2014) Improving the efficiency
of quantitative 1H NMR: An innovative external standard-internal reference approach. J Pharm
Biomed Anal 88: 1-6.
Huttanus HM, Sheng J and Feng X (2016) Metabolic Engineering for Production of Small Molecule Drugs:
Challenges and Solutions. Fermentation 2: 4.
Johnson KA, Liu X, Huang S, Roongta V, Humphreys WG and Shu Y-Z (2010) Rapid structure
determination of microgram-level drug metabolites using HPLC-MS, fraction collection and NMR
spectroscopy. Anal Methods 2: 1542-1549.
26

DMD Fast Forward. Published on September 21, 2017 as DOI: 10.1124/dmd.117.077073
This article has not been copyedited and formatted. The final version may differ from this version.
DMD # 77073
Jung Y-S, Hyeon J-S and Hwang G-S (2016) Software-assisted serum metabolite quantification using
NMR. Analytica Chimica Acta 934: 194-202.
Krunic A, Orjala J (2015) Application of high-field NMR spectroscopy for characterization and quantitation
of submilligram quantities of isolated natural products. Magn Reson Chem 53: 1043-1050.
Liang Y, Ge-Ying R, Hui-Ru T and Chao-Yang LIU (2010) Human Urinary Metabolite Quantification in
Nuclear Magnetic Resonance-based Metabonomics Using Electronic Reference. Chinese J Anal
Chem 38: 789-794.
Molinski TF. (2010) NMR of natural products at the ‘nanomole-scale’. Nat Prod Rep 27: 321-329.
Nagana Gowda GA and Raftery D (2015) Can NMR solve some significant challenges in metabolomics?
J Magn Reson 260: 144-160.
Pauli GF, Chen S-N, Simmler C, Lankin DC, Gödecke T, Jaki BU, Friesen JB, McAlpine JB and
Napolitano JG (2014) Importance of Purity Evaluation and the Potential of Quantitative 1H NMR
as a Purity Assay. J Med Chem 57: 9220-9231.
Picà Y and Barcelà D (2008) The expanding role of LC-MS in analyzing metabolites and degradation
products of food contaminants. Trends Anal Chem 27: 821-835.
RundlÃf T, McEwen I, Johansson M and Arvidsson T (2014) Use and qualification of primary and
secondary standards employed in quantitative 1H NMR spectroscopy of pharmaceuticals. J Pharm
Biomed Anal 93:111-117.
27

DMD Fast Forward. Published on September 21, 2017 as DOI: 10.1124/dmd.117.077073
This article has not been copyedited and formatted. The final version may differ from this version.
DMD # 77073
Saito T, Ihara T, Koike M, Kinugasa S, Fujimine Y, Nose K and Hirai T (2009) A new traceability scheme
for the development of international system-traceable persistent organic pollutant reference
materials by quantitative nuclear magnetic resonance. Accredit Qual Assur 14: 79-86.
Scalbert A, Brennan L, Fiehn O, Hankemeier T, Kristal BS, van Ommen B, Pujos-Guillot E, Verheij E,
Wishart D and Wopereis S (2009) Mass-spectrometry-based metabolomics: limitations and
recommendations for future progress with particular focus on nutrition research. Metabolomics 5:
435-458.
Schadel M, Wu D, Otton SV, Kalow W and Sellers EM (1995) Pharmacokinetics of dextromethorphan and
metabolites in humans: influence of the CYP2D6 phenotype and quinidine inhibition. J Clin
Psychopharmacol 15: 263.
Schlotterbeck Gt, Ross A, Hochstrasser R, Senn H, Kühn T, Marek D and Schett O (2002) High-
Resolution Capillary Tube NMR. A Miniaturized 5-μL High-Sensitivity TXI Probe for Mass-
Limited Samples, Off-Line LC NMR, and HT NMR. Anal Chem 74: 4464-4471.
Siddiqui MR, AlOthman ZA and Rahman N (2013) Analytical techniques in pharmaceutical analysis: A
review. Arab J Chem 10: S1409–S1421.
Simmler C, Napolitano JG, McAlpine JB, Chen S-N and Pauli GF (2014) Universal quantitative NMR
analysis of complex natural samples. Curr Opin Biotechnol 25: 51-59.
28

DMD Fast Forward. Published on September 21, 2017 as DOI: 10.1124/dmd.117.077073
This article has not been copyedited and formatted. The final version may differ from this version.
DMD # 77073
Spraul, M, Freund AS, Nast RE, Withers RS, Maas WE and Corcoran O. (2003) Advancing NMR
sensitivity for LC-NMR-MS using a cryoflow probe: application to the analysis of acetaminophen
metabolites in urine. Anal Chem 75: 1536-1541.
Subramanian M, Paruchury S, Singh S, Singh SP, Arla R, Pahwa S, Jana S, Katnapally P, Yoganand V,
Lakshmaiah B, Mazumder D, Ghosh K, Marathe PH and Mandlekar S (2017) Characterization of
Recombinantly Expressed Rat and Monkey Intestinal Alkaline Phosphatases: In Vitro Studies and
In Vivo Correlations. Drug Metab Dispos 45: 1425-32.
Timmerman P, Blech S, White S, Green M, Delatour C, McDougall S, Mannens G, Smeraglia J, Williams
S and Young G (2016) Best practices for metabolite quantification in drug development: updated
recommendation from the European Bioanalysis Forum. Bioanalysis 8: 1297-1305.
Timmerman P, Lowes S, Fast DM and Garofolo F (2010) Request for global harmonization of the guidance
for Bioanalytical Method Validation and sample analysis. Bioanalysis 2: 683.
Tao JA, Lin GQ and Liese A (2009) Biocatalysis for the pharmaceutical industry: discovery, development,
and
manufacturing,
pp
324,
John
Wiley
&
Sons
(Asia)
Singapore, Singapore
Walker GS, Bauman JN, Ryder TF, Smith EB, Spracklin DK and Obach RS (2014) Biosynthesis of drug
metabolites and quantitation using NMR spectroscopy for use in pharmacologic and drug
metabolism studies. Drug Metab Dispos 42: 1627-39.
29

DMD Fast Forward. Published on September 21, 2017 as DOI: 10.1124/dmd.117.077073
This article has not been copyedited and formatted. The final version may differ from this version.
DMD # 77073
Walker GS, Ryder TF, Sharma R, Smith EB and Freund A (2011) Validation of isolated metabolites from
drug metabolism studies as analytical standards by quantitative NMR. Drug Metab Dispos 39: 433-
440.
Weber M, Hellriegel C, Rueck A, Wuethrich J and Jenks P (2014) Using high-performance 1H NMR (HP-
qNMR®) for the certification of organic reference materials under accreditation guidelines--
describing the overall process with focus on homogeneity and stability assessment. J Pharm
Biomed Anal 93: 102-110.
Wieruszeski JM, Fritzinger B, Hanoulle X, Martins JC, Lippens G (2008) Sandwich-ELISE NMR:
Reducing the sample volume of NMR samples. J Magn Reso, 193: 37-40.
Wishart DS (2008) Quantitative metabolomics using NMR. Trends Anal Chem 27: 228-237.
Zhou J and Yin Y (2016) Strategies for large-scale targeted metabolomics quantification by liquid
chromatography-mass spectrometry. Analyst 141: 6362-6373.
Zhang X, Iraneta PC, Marszalkowski FJ, Jr. Fountain KJ, A simplified solid phase extraction (SPE) protocol
for
bioanalysis
using
Oasis
HLB,
Waters
Corporation,
Milford,
MA,
USA,
http://www.waters.com/webassets/cms/library/docs/720005140en.pdf
30