Optimization of N-methyl-N-[tert-butyldimethylsilyl]- trifluoroacetamide as a derivatization agent for determining isotopic enrichment of glycerol in very-low density lipoproteins

Article abstract of DOI:10.1002/rcm.4422

Stable isotope kinetic studies play an important role in the study of very-low density lipoprotein (VLDL) metabolism, including basic and clinical research. Today, [1,1,2,3,3-²H₅]glycerol is the most cost-effective alternative to measure glycerol and triglyceride kinetics. Recycling of glycerol from glycolysis and gluconeogenesis may lead to incompletely labelled tracer molecules. Many existing methods for the measurement of glycerol isotopic enrichment involve the production of glycerol derivatives that result in fragmentation of the glycerol molecule after ionization. It would be favourable to measure the intact tracer molecule since incompletely labelled tracer molecules may be measured as fully labelled. The number of methods available to measure the intact tracer in biological samples is limited. The aim of this project was to develop a gas chromatography/mass spectrometry (GC/MS) method for glycerol enrichment that measures the intact glycerol backbone and is suitable for electron ionization (EI), which is widely available. A previously published method for N-methyl-N-[tert-butyldimethylsilyl]trifluoroacetamide (MTBSTFA) derivatization was signifi-cantly improved; we produced a stable derivative and increased recovery 27-fold in standards. We used the optimized MTBSTFA method in VLDL-triglyceride and found that further modification was required to take matrix effects into account. We now have a robust method to measure glycerol isotopic enrichment by GC/EI-MS that can be used to rule out the known problem of tracer recycling in studies of VLDL kinetics. John Wiley & Sons, Ltd.

Full text of DOI:10.1002/rcm.4422

RAPID COMMUNICATIONS IN MASS SPECTROMETRY
Rapid Commun. Mass Spectrom. 2010; 24: 586–592
Published online in Wiley InterScience (www.interscience.wiley.com) DOI: 10.1002/rcm.4422
Optimization of N-methyl-N-[tert-butyldimethylsilyl]-
trifluoroacetamide as a derivatization agent for
determining isotopic enrichment of glycerol in
very-low density lipoproteins
1
Martin Adiels , Thomas Larsson , Pauline Sutton , Marja-Riitta Taskinen , Jan Boren
1
2
3
1
*
´
and Barbara A. Fielding2
1
¨
¨
Wallenberg Laboratory/Sahlgrenska Center for Cardiovascular and Metabolic Research, Goteborg University, 413 45 Goteborg, Sweden
2Oxford Centre for Diabetes, Endocrinology and Metabolism, University of Oxford, Churchill Hospital, Oxford OX3 7LJ, UK
3Division of Cardiology, Department of Medicine, University of Helsinki, Helsinki, Finland
Received 15 October 2009; Revised 11 December 2009; Accepted 11 December 2009
Stable isotope kinetic studies play an important role in the study of very-low density lipoprotein
(VLDL) metabolism, including basic and clinical research. Today, [1,1,2,3,3-2H5]glycerol is the most
cost-effective alternative to measure glycerol and triglyceride kinetics. Recycling of glycerol from
glycolysis and gluconeogenesis may lead to incompletely labelled tracer molecules. Many existing
methods for the measurement of glycerol isotopic enrichment involve the production of glycerol
derivatives that result in fragmentation of the glycerol molecule after ionization. It would be
favourable to measure the intact tracer molecule since incompletely labelled tracer molecules
may be measured as fully labelled. The number of methods available to measure the intact tracer
in biological samples is limited. The aim of this project was to develop a gas chromatography/mass
spectrometry (GC/MS) method for glycerol enrichment that measures the intact glycerol backbone
and is suitable for electron ionization (EI), which is widely available. A previously published method
for N-methyl-N-[tert-butyldimethylsilyl]trifluoroacetamide (MTBSTFA) derivatization was signifi-
cantly improved; we produced a stable derivative and increased recovery 27-fold in standards. We
used the optimized MTBSTFA method in VLDL-triglyceride and found that further modification was
required to take matrix effects into account. We now have a robust method to measure glycerol
isotopic enrichment by GC/EI-MS that can be used to rule out the known problem of tracer recycling
in studies of VLDL kinetics. Copyright # 2010 John Wiley & Sons, Ltd.
Abnormal concentrations of lipids and apolipoproteins are
markers of several lipid disorders and can result from
changes in the production, conversion or catabolism of
lipoprotein particles. Thus, although static measurements
are important, they do not reveal the underlying mechanisms
involved in the dysregulation of lipid disorders. To infer this
information, it is necessary to perform in vivo tracer/tracee
studies in which the rates of synthesis or catabolism of a
particular lipoprotein or apolipoprotein can be determined.
The metabolism of lipoprotein particles can be followed by
injecting amino acids labelled with stable isotopes which are
then incorporated into proteins such as apoB100.1 The
triglyceride content can be followed by infusion of labelled
glycerol or free fatty acids.2 Early studies used radioactive
[14C or 3H]glycerol3,4 but all in vivo lipoprotein kinetic studies
are now performed using infusion of stable isotope labelled
compounds such as [1,2,3-13C3]glycerol or [1,1,2,3,3-2H5]gly-
cerol.2 Kinetic studies of very-low density lipoproteins
(VLDL) have been informative especially when sub-divided
according to density (VLDL1 and VLDL2).5,6 Several
approaches to determine the enrichment of glycerol 2H5
have been developed, using gas chromatography/mass
spectrometry (GC/MS) and electron ionization (EI),2,7–9 or
chemical ionization (CI).10,11 These include derivatization to
triacetate glycerol,7,8 or tris-trimethylsilyl glycerol,9 and EI-
MS. A potential weakness using these methods is that
fragmentation of the molecule occurs between the carbon 1
and 2, or 2 and 3 of the glycerol backbone during electron
ionization. Methods that lead to fragmentation of the glycerol
backbone are inherently less sensitive.12 In addition, it has
been found that a considerable amount of the glycerol in
VLDL-triglyceride can be derived from glycolytic cycling
that result in the formation of partially de-deuterated
glycerol.2 It is important to exclude this recycling from the
kinetic modelling. Any such recycling would not be
detectable unless the entire glycerol backbone remained
intact during GC/MS analysis, enabling the Mþ5 ion to be
detected. Particular derivatization techniques that sub-
*Correspondence to: M. Adiels, Wallenberg Laboratory/LAB 5,
Goteborg University, Bruna Straket 16, 413 45 Goteborg, Sweden.
¨
¨
E-mail: martin.adiels@wlab.gu.se
Copyright # 2010 John Wiley & Sons, Ltd.

MTBSTFA method for analyzing glycerol enrichment 587
sequently tend to leave the glycerol backbone intact while in
the mass spectrometer include reaction with heptafluor-
obutyric anhydride (HFBA, both EI and CI)2,11 and reaction
with acetic anhydride in combination with CI.10 In these
cases, the fragmentation preferentially occurs in the glycerol
side chains created by the derivatization reagent.
(No. 27530) from Thermo (Rockford, IL, USA) were
derivatization grade. MTBSTFA (No. 19915-5) and glycerol
(No. 7757) were from Sigma (Stockholm, Sweden). Zeolite
(No. 96096) was from Fluka Biochemika (Buchs, Switzerland).
Methods
The number of methods that use EI while retaining the
structure of glycerol is limited. An EI method is desirable for
GC/MS compared with positive or negative CI as it keeps the
source cleaner. In addition, although CI is becoming more
common, many low-cost bench-top mass spectrometers do
not posses this facility as standard. Although EI has been
used to analyze glycerol derivatized with HFBA, the relevant
ion is of low intensity.11 The reagent N-methyl-N-[tert-
butyldimethylsilyl]trifluoroacetamide (MTBSTFA) is com-
monly used to create TBDMS (tert-butyldimethylsilyl)
derivatives of metabolites13 in order to make them more
suitable for analysis by MS. MTBSTFA reacts with groups
such as hydroxyl, carboxyl, thiol, and primary and secondary
amines and has been used to derivatize glycerol before its
analysis by GC/EI-MS.12 However, here we report signifi-
cant problems using a method that uses MTBSTFA to
derivatize glycerol when analyzed by EI.12 We describe
improvements in the protocol which have resulted in a clean,
robust method. We have also compared results using
derivatization with MTBSTFA (with EI) and acetic anhydride
(with EI) on the kinetic parameters calculated using a multi-
compartmental model of VLDL1 and VLDL2. We also tested
the improved MTBSTFA method on samples prepared by
solvent extraction, which allows simultaneous fatty acid
analysis.
Standards and samples
Stock solutions (10 mmol/L) of unlabelled and [1,1,2,3,3-
2H5]glycerol were prepared in deionized H2O. For isotopic
standard curve samples, the stock solutions were mixed 1:1
for a 50 atom percent excess (APE) solution and sequentially
diluted 1:1 with the unlabelled solution for solutions with 25,
12.5, 6.25, 3.13, 1.56, 0.78, 0.39, 0.20, 0.098, 0.049 APE. These
standard curve samples were evaporated to dryness in a
Speed-Vac1 sample concentrator (Savant Instruments, Farm-
ingdale, NY, USA) from volumes of 100 mL and were kept frozen
at ꢁ208C. Plasma from eight subjects was collected after injection
of [2H3]leucine and [1,1,2,3,3-2H5]glycerol.7 The samples
were analyzed for enrichment of leucine in plasma, VLDL1
and VLDL2-apoB as described7 and for glycerol in plasma
and VLDL1- and VLDL2-triglyceride as described below.
Preparation of VLDL
VLDL1 and VLDL2 were isolated in a total volume of 8.4 mL
EDTA-plasma.14 For further analysis, 1 mL of VLDL1 or
500 mL of VLDL2 was used. The samples were deproteinized
with isopropanol and delipidated twice with ethanol/
diethyl ether.15 The supernatants were combined, and the
volume was increased to 20 mL with isopropanol. The
phospholipids were removed by adding 2 g of activated
zeolite to each tube and mixing for 20 min.7 After centri-
fugation, the supernatants were evaporated under N2 at
808C, dissolved in isopropanol (1 mL), transferred into a
1.5 mL vial, and dried on a heating block at 808C.
EXPERIMENTAL
Subjects
At this point, samples could be treated in one of two ways
(in order to release glycerol from the triglyceride molecule)
depending on which further analyses were to be performed.
Eight subjects with a varied degree of liver fat (4–25%), body
mass index (BMI) (27–40 kgꢀmꢁ2) and waist measurement
(82–133 cm) were recruited for this study. The purpose,
nature, and potential risks of the study were explained to all
subjects before their written consent was obtained. The study
protocols were approved by the Ethical Committee of the
Helsinki University Central Hospital.
1. Saponification of triglycerides with
potassium hydroxide
Potassium hydroxide stock solution was prepared by
dissolving 33 g KOH in 100 mL deionized H2O. On the
day of analysis, 3 mL of the stock solution were dissolved in
50 mL 95% ethanol. After removal of the phospholipids with
zeolite, the sample was saponified by adding 250 mL
saponification solution. After incubation at 608C for 2 h,
the sample was cooled to room temperature. We modified
the protocol by including a purification step; 200 mL chloro-
form was added, followed by 200 mL H2O. After mixing and
centrifuging, 200 mL of the aqueous phase (top layer) were
transferred to a new vial and dried under nitrogen. This was
necessary to achieve effective derivatization with MTBSTFA.
The samples were then derivatized with either MTBSTFA or
acetic anhydride as described in the relevant sections below.
Equipment
The GC/MS system (Agilent, Santa Rosa, CA, USA)
consisted of a gas chromatograph (model 6890N) equipped
with an autosampler injector (model 7683B) and a mass
spectrometer (model 5975B) with an electron ionization (EI)
source. The column was a DB5MS (30 m, 0.25 mm, film
thickness 0.25 mm; 122-5532GL; J&W, Folsom, CA, USA)
equipped with a guard column of approximately 10 m. The
carrier gas was helium.
Materials
[1,1,2,3,3-2H5]Glycerol was obtained from Sigma/Isotec
(Miamisburg, OH, USA). Cation resin (No. 142-1441), AG-
50W-X8 (100–200 mesh), hydrogen form and anion resin (No.
140-1444), AG-1-X8 (100–200 mesh), formate form, were
obtained from Bio-Rad (Sundbyberg, Sweden). Acetonitrile
(No. 20062) from Pierce (Rockford, IL, USA) and pyridine
2. Transesterification and ion-exchange
chromatography
We wished to test the MTBSTFA method using an alternative
preparative procedure that would allow for the simultaneous
Copyright # 2010 John Wiley & Sons, Ltd.
Rapid Commun. Mass Spectrom. 2010; 24: 586–592
DOI: 10.1002/rcm

588 M. Adiels et al.
isolation of VLDL-fatty acids after removal of phospholipids.
Therefore, the dried samples from the ‘VLDL preparation’
were subjected to a transesterification/methylation pro-
cedure. In this procedure,2,16 the fatty acid methyl esters
were extracted into a solvent layer, leaving the glycerol
backbone from the triglyceride molecule in aqueous solution.
The samples were dissolved in 400 mL toluene, mixed and
transferred to 15 mL Pyrex1 tubes. Fatty acid methyl esters
were prepared from triglycerides (plus, in this case, any
small amount of cholesteryl ester present in the VLDL) by
adding 800 mL of newly prepared 1.5% H2SO4 in methanol to
the glass tubes. The tubes were vortexed briefly, incubated at
808C for 1 h, and allowed to cool down to room temperature.
The samples were neutralized using 2 mL neutralizing
solution (25 g KHCO3 (0.125 M) þ 34.6 g K2CO3 (0.125 M))
made up to 500 mL with distilled water and diluted 1:5 for
use, and then phase-separated using 2 mL cyclohexane and
mixing for 15 min on a rotary mixer before centrifuging at
1000 g at room temperature for 10 min. The supernatant was
removed (and could be used for further fatty acid analyses)
and any remaining solvent was evaporated using nitrogen
(35–508C, 5–10 min). The aqueous phase (which contained
the glycerol released from the triglyceride) was purified by a
standard ion-exchange chromatography protocol as follows.
Anion and cation resins were mixed separately with
deionized H2O (1:1 v/v) in a beaker and stirred at room
temperature for 5 min. A polypropylene frit was placed into a
3 mL column and, after stirring the resins, 1 mL cation resin
was added with a pipette and the wall and resin were
washed with 1 mL deionized H2O. Then 1 mL anion resin
was added on top of the cation with a pipette and the walls
and resin were washed with 3 ꢂ 1 mL deionized H2O. The
columns could be stored at room temperature for 3 days
before use. The samples were carefully added to the
prepared ion-exchange columns and carefully eluted into
15 mL tubes with 3 ꢂ 1 mL deionized H2O. The samples were
then frozen at ꢁ808C for at least 2 h before being placed in a
freeze-drier over night (approximately 15 h). Since glycerol is
volatile care must be taken not to dry samples excessively.
Thus, the samples were removed when a few hundred mL of
the samples remained and they were then put under nitrogen
at 358C to evaporate until dry.
time 18 min). The MSD interface temperature was 3208C. A
volume of 1 mL of the sample was injected at a split ratio 50:1.
Tri-[tert-butyldimethylsilyl]glycerol has a molecular mass of
434 and the major EI peak corresponds to a fragmentation
where a butyl group is removed and results in a fragment ion
with a monoisotopic mass-to-charge ratio (m/z) of 377.3. The
peaks of m/z 377.3 and 382.3 were then measured using
selective ion monitoring. If necessary, samples were rerun
with a variable split ratio to ensure that the 377.3 peak areas
were of similar size (ꢃ20%).
Derivatization with acetic anhydride
The triacetate protocol was performed as previously
described.7 Glycerol was derivatized to its 1,2,3-triacetate
ester by adding equal volumes (20 mL) of pyridine and acetic
anhydride and was left at room temperature for 30 min. An
additional purification step was included; 100 mL chloroform
and 50 mL water were added and, after mixing and
centrifuging, 95 mL of the chloroform phase were transferred
to the autosampler vials, dried under nitrogen and re-
dissolved in 30 mL pyridine.
GC/MS analysis of glycerol triacetate
The initial oven temperature was 1708C and this was held
constant for 5 min before being ramped at 108C/min to 2108C
and held constant for 6 min. The injector and interface
temperatures were both 3208C and the column flow rate was
1.2 mL/min. A volume of 1 mL of the sample was injected at a
split ratio of 50:1. The peaks areas corresponding to m/z 145,
147 and 148 were measured. The tracer/tracee ratio (TTR)
was calculated either as the ratio of m/z 148 to 145 using a
fixed m/z 147-to-145 ratio determined from the baseline
samples, which were run in duplicates,7 or by a direct ratio of
the m/z 145 and 148 peaks.
Preparation of plasma samples (free glycerol)
The plasma samples were de-proteinized by adding 2 mL ice-
cold acetone to 0.5 mL of plasma. The samples were mixed by
vortex and incubated for 1 h on ice. After centrifuging at
3200 g at room temperature for 10 min, the supernatant was
transferred to a new tube. The acetone was separated from
the water phase by adding 2 mL chloroform, centrifuging at
3200g at 48C for 10 min and transferring the aqueous phase
(upper 0.5 mL) to a new tube. This procedure was repeated
once more after the addition of 0.5 mL deionized H2O. To
remove any remaining acetone, the samples were incubated
under nitrogen for 10–15 min at 45–508C. The aqueous phase
was then passed through an ion-exchange column as
described above, and samples derivatized with MTBSTFA.
Optimized derivatization with MTBSTFA
The samples that had been through the transesterification
procedure were derivatized by adding 100 mL pyridine and
100 mL MTBSTFA, and samples that had been passed
through saponification were derivatized in 50 mL pyridine
and 50 mL MTBSTFA. The samples were incubated for 4 h at
708C, cooled down, and transferred to autosampler vials
before being evaporated under nitrogen at 358C and re-
dissolved in 50 mL pyridine. The vials were tightly closed
with crimp lids.
Data analysis
Isotopic standard curve samples that had been derivatized
with the same reagent (MTBSTFA or acetic anhydride) and
were run before and after each sample sequence. The mean
ratio of the measured to expected APE, expressed as a
percentage, was calculated for the standard curves and
sample enrichments were corrected using a linear interp-
olation between the ‘before’ and ‘after’ standard curves. The
atom percent excess (APE) was calculated for each sample by
calculating the tracer/tracee ratio (TTR), followed by
GC/MS analysis of tri-[tert-butyldimethylsilyl]glycerol
For GC/MS analysis, the injector temperature on the gas
chromatograph was 3208C and column flow rate was
1.1 mL/min. The initial oven temperature was 1108C and
it was ramped at 108C/min to 2108C and then at 208C/min to
3108C, and was held at this temperature for 3 min (total run
Copyright # 2010 John Wiley & Sons, Ltd.
Rapid Commun. Mass Spectrom. 2010; 24: 586–592
DOI: 10.1002/rcm

MTBSTFA method for analyzing glycerol enrichment 589
subtracting the background TTR from a baseline sample and
then normalizing by total count (e.g. 1 þ TTR) and converting
into a percentage.17
RESULTS
Incomplete derivatization after 30 min at room
temperature
Un-enriched glycerol standards were first derivatized for
30 min with MTBSTFA at room temperature according to the
method of Flakoll et al.12 We realized that derivatization was
continuing in the GC vial, so we sought to investigate under
what conditions the reaction became complete.
Kinetic modelling
We performed a kinetic study in eight subjects (2 groups of 4)
with infusions as previously described,7 and compared the
results obtained using the optimized MTBSTFA and acetic
anhydride methods. In four subjects we compared the
optimized MTBSTFA method with glycerol isolated during
transesterification/methylation of fatty acids (group G1M)
with the acetic anhydride protocol after saponification
(G1A). In the other four subjects we compared the MTBSTFA
(G2M) with acetic anhydride (G2A) derivatization after
saponification (Fig. 1).
Optimization of MTBSTFA derivatization
conditions
Initial experiments at room temperature showed that the
peak area increased by approximately six-fold from 30 min to
70 h (data not shown). Thus, standard samples were
derivatized under different conditions and we found the
Figure 1. Schematic overview of methods. Samples from eight subjects were
analyzed. After ultracentrifugation and removal of phospholipids, samples were
split in two groups of four. Four samples (group 1) were split into two equal volumes
and either treated with the modified saponification protocol and derivatized with
acetic anhydride and analyzed by GC/EI-MS (G1A), or transesterified using the
method of Flakoll et al.12 and derivatized using MTBSTFA and analyzed by GC/EI-
MS (G1M). The other four samples (group 2) were treated with the modified
saponification protocol and split into two equal volumes after phase separation
and then derivatized with either acetic anhydride (G2A) or MTBSTFA (G2M).
Plasma samples from all eight subjects were analyzed with the same protocol
(P1-2M).
Copyright # 2010 John Wiley & Sons, Ltd.
Rapid Commun. Mass Spectrom. 2010; 24: 586–592
DOI: 10.1002/rcm

590 M. Adiels et al.
Figure 4. Measured and expected APE from sequential
dilution series. The recovery was linear in the range 0–50
APE. Solid line, theoretical 100% recovery; Open circles,
measured APE (n ¼ 22). Coefficient of variation was 1–8%.
Figure 2. Peak areas after reaction at room temperature or
708C and the effect of TBDMCS. Glycerol was derivatized with
MTBSTFA with or without TBDMCS at either room tempera-
ture or at 708C for 4 h and then left at room temperature. At
room temperature there was little difference between the peak
areas with or without TBDMCS at 8 h. Performing the reaction
at 708C for 4 h increased the peak area 7-fold compared with
that at room temperature; however, when TBDMCS was
included the increase was only 3-fold. Open boxes, room
temperature; closed boxes, room temperature with TBDMCS;
open circles, 708C 4 h; closed circles, 708C 4 h with TBDMCS.
enrichment of tri-[tert-butyldimethylsilyl]glycerol in a linear
manner in this range (Fig. 4). The ratio of observed to
expected APE was 1.01 ꢃ 0.01 (mean ꢃ standard error of the
mean (SEM), n ¼ 22) and ranged from 0.94 to 1.10 (n ¼ 22). In
addition, the measured TTRs in the standard curve samples
analyzed immediately after the samples were within 1% of
the TTRs measured in the standard curve samples run prior
to sample measurements.
peak area to increase with the derivatization time and
temperature (Fig. 2).
Including the accelerator tert-butyldimethylchlorosilane
(TBDMCS) in the derivatization reaction mixture yielded no
improvement in recovery, either at room temperature or at
708C. In fact, the peak areas were lower when TBDMCS was
included when incubating at 708C (Fig. 2).
To establish a final derivatization protocol, the standards
were then derivatized at 708C for 30 min, 1, 2, 4, 6 and 8 h.
With increasing derivatization time, the product peak area
increased 27-fold from 30 min to 8 h (Fig. 3). The m/z 382.3
peak area could be measured accurately in un-deuterated
samples as the m/z 377.3 peak area reached 20 ꢂ 106, and the
product was found to be stable for at least 5 days.
Comparison of isotopic enrichments
The ratio between the observed and theoretical APE in the
standard curves was close to 1.0 for tri-[tert-butyldimethyl-
silyl]glycerol (see results above) but only 0.74 ꢃ 0.005(mean
ꢃ SEM, n ¼ 8) for glycerol triacetate, and the sample
enrichments were adjusted accordingly. The isotopic enrich-
ments in glycerol from VLDL-triglycerides were higher (on
average 9–34% for all sample points in VLDL1 and VLDL2)
with the MTBSTFA protocol than with the triacetate method7
(Fig. 5).
Data presented by Patterson et. al.2 in which the relative
proportions of different isotopomers in VLDL glycerol (non-
fragmented) were presented show that only 25% of the
glycerol tracer was mþ5 (the form administered), some 6%
Recovery of glycerol enrichment is linear
between 0 and 50 APE
Running sequential dilution series between 0 and 50 APE
showed that the method made it possible to detect
Figure 5. Enrichment curves in glycerol from VLDL1- and
VLDL2-triglycerides after a bolus injection of [1,1,2,3,3-2H5]
glycerol. Glycerol enrichment (APE) in VLDL1 and VLDL2
using two different derivatization methods. Open symbols,
optimized MTBSTFA (M) method; closed symbols, acetic
anhydride (AA) method.
Figure 3. Reaction speed versus time. We derivatized gly-
cerol for 30 min, 1, 2, 4, 6 and 8 h and found that the
derivatization continued for at least 8 h. We conclude that
4 h at 708C was sufficient for the reaction to be sufficiently
completed. Closed boxes, m/z 377.3 peak area.
Copyright # 2010 John Wiley & Sons, Ltd.
Rapid Commun. Mass Spectrom. 2010; 24: 586–592
DOI: 10.1002/rcm

MTBSTFA method for analyzing glycerol enrichment 591
Groups 1 & 2
Table 1. Kinetic parameters determined using the different methods
Group 1 Group 2
G1A
G1M
pa
G2A
G2M
pa
G1-2A
G1-2M
pb
Triglycerides
VLDL1 FCR
VLDL1 Sec
VLDL2 FCR
VLDL2 DSec
ApoB
11.23
34350
10.64
4444
10.28
32036
10.26
4525
0.625
0.375
0.875
0.875
11.00
21032
11.87
2915
11.96
22189
8.87
0.375
0.625
0.250
0.250
11.12
27691
11.26
3679
11.12
27113
9.56
0.996
0.710
0.124
0.290
2301
3413
VLDL1 FCR
VLDL1 Sec
VLDL2 FCR
VLDL2 DSec
8.40
1170
4.54
363
8.42
1167
4.33
364
0.875
1.000
0.625
0.875
9.44
842
4.25
222
10.03
902
3.89
226
0.250
0.250
0.625
1.000
8.92
1006
4.39
293
9.22
1035
4.11
295
0.383
0.476
0.286
0.806
FCR, fractional catabolic rate [pools/day]; Sec, Secretion [mg/day]; DSec, Direct Secretion [mg/day]; G1A, group 1 saponification þ acetic
anhydride; G1M group 1 transesterification þ MTBSTFA; G2A, group 2 saponification þ acetic anhydride; G2M group 2 transesterifica-
tion þ MTBSTFA.
a p-values attained from paired Wilcoxon paired rank test.
b p-values attained from paired t-test.
was mþ4 and 13% was mþ3. In our samples derivatized to
triacetate glycerol, five-labelled glycerol will occur as mþ3 in
the fragment ion monitored. Four-labelled glycerol will
always occur as mþ2 since it is the product of conversion of
glycerol-3-phospate into dihydroxyacetone phosphate and
back, which replaces the deuterium on the second carbon.18
Further processing of dihydroxyacetone produces glyceral-
dehyde-3-phosphate in which one hydrogen atom on the first
carbon is replaced. This produces a three-labelled glycerol
molecule that can either result in an mþ2 or an mþ1
fragment ion, depending on where the molecule is disrupted
during ionization. Thus, the mþ2 peak may very well be
overestimated and therefore the calculated mþ3/mþ2 ratio
results in an underestimation of the true TTR.
MTBSTFA to derivatize glycerol,12 it was found that the
reaction was far from complete. We therefore investigated
the effect of temperature and time on the reaction, and have
developed a more robust and sensitive method.
Several changes were made to the protocol in order to
maximize the amount of material derivatized, increase the
peak area of the selected ions, and reduce the amount of un-
derivatized material injected onto the column. Using our
revised protocol, we increased the amount of derivatized
material 27-fold. The original protocol stated that 30 min at
room temperature was enough for complete derivatization,12
but we found that the reaction continued for many hours.
Moreover, we found that increasing the temperature to 708C
further increased the amount of derivatized material several-
fold. For practical reasons, 4 h at 708C was chosen as the
reaction time and temperature. Under these conditions, the
reaction was nearly complete. However, it was possible to
incubate for up to a further 24 h at room temperature. We
included an evaporation step and re-dissolved the deriva-
tized sample in pyridine to ensure a stable solution and to
avoid continuation of the reaction in the autosampler vials.
TBDMCS is often used as an accelerator for MTBSTFA in
derivatization protocols. However, we found that adding 1%
TBDMCS to the MTBSTFA neither enhanced the reaction
speed nor stabilized the product, and therefore the accel-
erator was not necessary. We found that the standard curves
were linear from 0 to 50 APE, with a linear detection range
using our GC/MS instrumentation. A wide linear range is
necessary since plasma glycerol enrichment may reach 50
APE immediately after tracer injection, whereas the typical
enrichment of VLDL1 and VLDL2 samples is below 1.0 APE.
We measured higher isotopic enrichment in the described
MTBSTFA methods than in the triacetate method. This can be
attributed to the fact that the triacetate method measures the
mþ2 and mþ3 peaks instead of the mþ0 and mþ3 peaks of
the fragment ion. The reason that this method was originally
employed is that the ratio of the peak areas of mþ3 and mþ0
is only 0.58 ꢂ 10ꢁ3 in the natural molecule and it is thus
difficult to measure low abundances and the background
abundance with good accuracy. As it has been shown that
Impact of choice of method for glycerol
enrichment measurements on kinetic
modelling
The VLDL1 and VLDL2 kinetic parameters calculated using a
multi-compartmental model7 are summarized in Table 1.
There were no significant differences when comparing
results from the same subjects using different methods, as
explained in Fig. 4 (G1A vs. G1M and G2A vs. G2M; NS
Wilcoxon paired rank test), nor was there any significant
difference when grouping subjects based on derivatization
method (G1-2A vs. G1-2M).
To test if the amplitude of the APE curve affected the
model outcome we scaled the VLDL1 and VLDL2 enrich-
ments curves by factors of 0.5, 0.75, 1.5 and 2.0 and re-
analyzed the results using the mathematical model. The
change in kinetic parameters after scaling was less than 1%.
DISCUSSION
In this paper we present a method for preparing glycerol
from triglycerides in VLDL1 and VLDL2 and measuring the
enrichment of the intact glycerol backbone using MTBSTFA
and EI-MS. The starting point for this analysis was to
compare the protocol described by Flakoll et al.12 with our
previously published protocol. However, when using
Copyright # 2010 John Wiley & Sons, Ltd.
Rapid Commun. Mass Spectrom. 2010; 24: 586–592
DOI: 10.1002/rcm

592 M. Adiels et al.
partially deuterated products are formed during tracer
recycling,2 it is a potential drawback of the method to have to
measure the mþ2 peak. However, this does not seem to
impact on the results of the modelling since the differences in
estimated parameters were small and non-systematic
(Table 1). We also tested the impact on the amplitude itself
of the enrichment curves by scaling the amplitude (between
0.5- and 2-fold) and found a difference less than 1% in all
estimated parameters. Thus, we suspect that even if the two
methods give different amplitudes, the small observed
difference in results is not due to the difference in amplitude
(which was 9–34% in the analyzed time series). Amplitude
variation (within the given range) does not influence the
results of the model. This is an inherent effect of the design
and implementation of the model and reflects the fact that the
production rate of VLDL is invariant of the contribution of
different glycerol sources. On the other hand, the relative
contribution from different glycerol sources cannot be
determined using this experimental setup.
current circumstances, and this allows for comparison
between results derived using these methods. However,
the possibility that there are circumstances where it is
important to measure the intact molecule cannot be ruled
out, and here we present such a method suitable for use with
EI-MS. In practice, the saponification protocol with deriva-
tization with MTBSTFA is the recommended protocol
whereas the transesterification protocol is applicable when
the free fatty acids are also to be measured.
Acknowledgements
The authors thank Mike Stolinski for helpful discussions and
Hannele Hilden, Helina Perttunen-Nio, Anne Salo, Elin
Stenfeldt and Maria Arnell for excellent technical assistance.
This work was supported by the the Swedish Foundation for
Strategic Research, European Atherosclerosis Society, the
Novo Nordisk Foundation, Emelle fond, Sigrid Juselius
Foundation Finland, the project ‘‘Hepatic and adipose tissue
and functions in the metabolic syndrome’’ (HEPADIP,
http://www.hepadip.org/), which is supported by the
European Commission as an Integrated Project under the
6th Framework Programme (Contract LSHM-CT-2005-
018734), and by ETHERPATHS (Contract FP7-KBBE-222639).
Even if there was no significant difference in the outcome
of the kinetic model, applications may still exist where it is
important to be able to measure the intact glycerol molecule.
Furthermore, the findings in this study also strengthen the
conclusions in earlier studies using the old protocol for the
measurement of glycerol isotopic enrichment since it was not
known how much these results were influenced by the
fragmentation of the glycerol molecule after ionization.7
Previously, EI-MS has been used with HFBA derivatiza-
tion, but the analytical response is of low intensity.11 Since
fluorine, and other halogens, are highly electronegative,
compounds containing these halogens preferentially form
negative ions and they are often analyzed by negative
chemical ionization. The MTBSTFA method presented here
gives a high signal, suggesting that lower initial amounts of
material could be used and thus reduce the amount of blood
drawn from the subject. HFBA, on the other hand, has the
advantage that it seems to be able to derivatize glycerol in the
presence of salts, such as directly after transesterification of
VLDL-triglycerides or directly in the glycerol-containing
aqueous phase of plasma,2 resulting in less time-consuming
procedures.
REFERENCES
1. Parhofer KG, Barrett PH. J. Lipid Res. 2006; 47: 1620.
2. Patterson BW, Mittendorfer B, Elias N, Satyanarayana R,
Klein S. J. Lipid Res. 2002; 43: 223.
3. Harris WS, Connor WE, Illingworth DR, Rothrock DW,
Foster DM. J. Lipid Res. 1990; 31: 1549.
4. Zech LA, Grundy SM, Steinberg D, Berman M. J. Clin Invest.
1979; 63: 1262.
5. Malmstrom R, Packard CJ, Caslake M, Bedford D, Stewart P,
Yki-Jarvinen H, Shepherd J, Taskinen MR. Diabetologia 1997;
40: 454.
6. Adiels M, Taskinen MR, Packard C, Caslake MJ, Soro-Paa-
vonen A, Westerbacka J, Vehkavaara S, Hakkinen A, Olofs-
son SO, Yki-Jarvinen H, Boren J. Diabetologia 2006; 49: 755.
7. Adiels M, Packard C, Caslake MJ, Stewart P, Soro A, Wester-
backa J, Wennberg B, Olofsson SO, Taskinen MR, Boren J.
J. Lipid Res. 2005; 46: 58.
8. Beylot M, Martin C, Beaufrere B, Riou JP, Mornex R. J. Lipid
Res. 1987; 28: 414.
9. Wolfe RR. Radioactive and Stable Isotope Tracers in Biomedicine.
Wiley-Liss: New York, 1992; 424.
10. Ackermans MT, Ruiter AF, Endert E. Anal. Biochem. 1998;
258: 80.
11. Gilker CD, Pesola GR, Matthews DE. Anal. Biochem. 1992;
205: 172.
CONCLUSIONS
12. Flakoll PJ, Zheng M, Vaughan S, Borel MJ. J. Chromatogr. B
Biomed. Sci. Appl. 2000; 744: 47.
13. Schwenk WF, Berg PJ, Beaufrere B, Miles JM, Haymond
MW. Anal. Biochem. 1984; 141: 101.
14. Lindgren FT, Jensen LC, Hatch FT. In Blood Lipids and
Lipoproteins: Quantitation, Composition, and Metabolism, Nel-
son GJ (ed). Wiley-Interscience: New York, 1972; 181–274.
15. Egusa G, Brady DW, Grundy SM, Howard BV. J. Lipid Res.
1983; 24: 1261.
16. Chong MF, Fielding BA, Frayn KN. Am. J. Clin. Nutr. 2007;
85: 1511.
17. Wolfe RR, Chinkes DL. Isotope Tracers in Metabolic Research:
Principles and Practice of Kinetic Analysis. Wiley-Liss: New
Jersey, 2005.
We have compared three different protocols for the
derivatization of glycerol from triglycerides derived from
lipoproteins. We optimized the derivatization reaction with
MTBSTFA with respect to time, temperature and the
inclusion of TBDMCS. We then compared the results from
kinetic modelling using data derived from our original
protocol using acetic anhydride, the optimized protocol
(optimized from Flakoll et al.12), and a new simpler protocol
for preparing VLDL-triglyceride glycerol, also using
MTBSTFA. We conclude that the results from kinetic
modelling using glycerol enrichment curves derived with
the different methods did not differ significantly, under the
18. McMurry J. Organic Chemistry. Brooks/Cole Publishing
Company: Pacific Grove, CA, 1996.
Copyright # 2010 John Wiley & Sons, Ltd.
Rapid Commun. Mass Spectrom. 2010; 24: 586–592
DOI: 10.1002/rcm