Trace detection of glycolic acid by electrophore labeling gas chromatography-electron capture mass spectrometry

Article abstract of DOI:10.1021/ac0304267

As little as 10 pg of standard glycolic acid (glycolate) was detected in a method comprising the following sequence of steps: (1) add glycolate-2,2-d ₂ as an internal standard and exchange the carboxylate oxygens in hot HCl/[¹⁸O]wate

Full text of DOI:10.1021/ac0304267

Anal. Chem. 2004, 76, 3049-3054
Trace Detection of Glycolic Acid by Electrophore
Labeling Gas Chromatography-Electron Capture
Mass Spectrometry
Gang Shao and Roger W. Giese*
Department of Pharmaceutical Sciences in the Bouve College of Pharmacy and Health Professions, Barnett Institute, and
Chemistry Department, Northeastern University, Boston, Massachusetts 02115
Prior assays for glycolate, whether as a standard, an industrial
chemical, or a metabolite, have provided moderate sensitivity,
when handling a detection limit6 is considered. For example, a
detection limit of 12 ng for standard glycolate was achieved
recently using a plant tissue-based chemiluminescence flow
biosensor.7 An amperometric biosensor was used to detect as little
as 90 ng (S/ N ) 6) of glycolate.5 Yao and Porter reported the
detection of spiked glycolate in serum down to 0.5 ng based on
trimethylsilylation/ GC-FID,8 while showing a chromatogram from
a sample containing a spike of 25 ng. Soga and co-workers, using
capillary electrophoresis-electrospray ionization-mass spectrom-
etry, showed an electropherogram obtained by making an injection
from a 20-µL sample containing 152 ng of glycolate and reported
a corresponding detection limit of 9 ng considering S/ N )3.9 In
an HPLC procedure with UV detection, glycolate was measured
in cosmetic products in the low-microgram range.10 Although
previously we reported the detection of a diluted standard of O2-
pivalyl-3′,5′-bis(trifluoromethyl)benzylglycolate by GC-ECMS at
the zeptomole level, anhydrous conditions were used for the
derivatization, and the derivatization was performed at the mil-
ligram level.11
Methods also have been reported for measuring analogues of
glycolate. For example, γ-hydroxybutyric acid has been detected
at the low-nanogram level by gas chromatography-positive chemi-
cal ionization mass spectrometry after conversion to the corre-
sponding lactone,12 and at the mid-nanogram level by HPLC with
UV detection.13 A low-nanogram level detection limit was reported
for lactic acid when measured by trimethylsilylation-GC-FID.14
Previously we reported the synthesis of an electrophoric
reagent, “AMACE1”, which can be coupled onto carboxylic acids
such as glycolate in an aqueous buffer, yielding, after further
derivatization as necessary, products with excellent detection
As little as 1 0 pg of standard glycolic acid (glycolate) was
detected in a method comprising the following sequence
of steps: (1 ) add glycolate-2 ,2 -d 2 as an internal standard
and exchange the carboxylate oxygens in hot HCl/ [1 8 O]-
water; (2 ) form an amide derivative with a water-soluble
carbodiimide and the electrophoric amine, AMACE1 ; (3 )
purify by bypass HP LC; (4) derivatize the residual hydroxy
with butyric anhydride; (5 ) partition with acetonitrile/ 2
M NaCl; and (6 ) detect by GC-ECMS. At an intermediate
stage in method development, 1 pg of glycolate-2 ,2 ,-d 2
could be detected by subjecting it to the above steps 2 -6 ,
forming product in an overall, absolute yield of 7 6 %. Step
1 was added after an effort to fully overcome background
contamination by glycolate was unsuccessful. For ex-
ample, background contamination by glycolate could
increase rather than decrease when the methanol reagent
in the procedure was “carefully purified.” The work
extends the sensitivity for glycolate detection by ∼1 0 0 -
fold and provides high-performance conditions for the
analytical steps employed.
We are interested in the trace detection of the polar metabolites
that arise from oxidative damage to the sugars of DNA.1 In part,
the measurement of these metabolites is of interest to better
understand how this type of damage is repaired enzymatically.2
More broadly, oxidative damage to DNA is of interest because it
may play a role in aging and some disease processes such as
cancer and heart disease. We selected glycolic acid (glycolate)
as an initial analyte to test. This compound can form as a
secondary metabolite of phosphoglycolate, which in turn arises
from oxidative damage at the 4′ position of the deoxyribose
residues of DNA.3 Due to its water solubility and ease of
occurrence as a background contaminant, trace glycolate is a
challenging analyte. Phosphoglycolate is a product of a reaction
catalyzed by ribulose-1,5-biphosphate carboxylase-oxygenase (rubis-
co), the most abundant protein on earth.4 Glycolate is an important
chemical industrially, environmentally, cosmetically, and clinically
as has been summarized recently.5
(5) Tsiafoulis, C. G.; Prodromidis, M. I.; Karayannis, M. I. Anal. Chem. 2 0 0 2 ,
74, 132-139.
(6) Yang, C.; Shimelis, O.; Zhou, X.; Li, G.; Bayle, C.; Nertz, M.; Lee, H.;
Strekowski, L.; Patonay, G.; Couderc, F.; Giese, R. W. J. Chromatogr., A
2 0 0 2 , 979, 307-314.
(7) Li, B.; Zhang, A.; Jin, Y. Anal. Chem. 2 0 0 1 , 73, 1203-1206.
(8) Yao, H. H.; Porter, W. H. Clin. Chem. 1 9 9 6 , 42, 292-297.
(9) Soga, T.; Ueno, Y.; Naraoka, H.; Ohashi, Y.; Tomita, M.; Nishioka, T. Anal.
Chem. 2 0 0 2 , 74, 2233-2239.
(10) Scalia, S.; Callegari, R.; Villani, S. J. Chromatogr., A 1 9 9 8 , 795A, 219-225.
(11) Wang, P.; Murugaiah, V.; Yeung, B.; Vouros, P.; Giese, R. W. J. Chromatogr.,
A 1 9 9 6 , 721, 289-296.
* To whom correspondence should be addressed. E-mail: r.giese@neu.edu.
(1) Von Sonntag, C. The Chemical Basis of Radiation Biology; Taylor & Franis,
London, 1987.
(2) Demple, B.; Harrison, L. Annu. Rev. Biochem. 1 9 9 4 , 63, 915-948.
(3) Greenberg, M. M. Chem. Res. Toxicol. 1 9 9 8 , 11, 1235-1248.
(4) Conn, E. E.; Stumpf, P. K.; Bruening, G.; Dot, R. H. Outlines of Biochemistry;
Wiley: New York, 1987; p 490.
(12) Frison, G.; Tedeschi, L.; Maietti, S.; Ferrara, S. D. Rapid Commun. Mass
Spectrom. 2 0 0 0 , 24, 2401-2407.
10.1021/ac0304267 CCC: $27.50 © 2004 American Chemical Society
Published on Web 04/23/2004
Analytical Chemistry, Vol. 76, No. 11, June 1, 2004 3049

characteristics by GC-ECMS.15 In part the favorable detection
properties come from a built-in site for dissociative electron
capture in AMACE1 that yields an analyte-characteristic anion
upon electron capture. Here we report the use of this reagent to
detect glycolate at the low-picogram level. This required us to
overcome two general problems: (1) some of the conditions that
worked well at microgram or nanogram levels of glycolate gave
poor recoveries or interferences at lower levels, and (2) contami-
nation by glycolate was persistent at the low-picogram level.
loosely with aluminum foil; dry in 120 °C oven for 1 h; cool in
desiccator and use within 1 day. Procedure B (disposable inserts;
insert holder vials; reagent test tubes for picogram or lower level
analysis): soak in methanol overnight; change and 15-min sonicate
in methanol twice and in dd water once; rinse with dd water again
and put up-side down and air-dry in a clean beaker, then cover
with aluminum foil and dry in 120 °C oven for 1 h, cool to ambient
T, and gas-phase silanize as described below. Procedure C
(Pasteur pipets): sonicate twice in methanol and then gas-phase
silanize the ones for use in the analytical procedure.
EXPERIMENTAL SECTION
Gas-P hase Silanization. A model 1410 vacuum oven from
VWR Scientific (Boston, MA) was used that was modified as
follows: the line from the vent valve was connected to an injection
septum followed by a valve (a) to N2, and the vacuum valve was
connected to an external three-way valve (b) that gave the options
of venting to a hood, to an oven closure, or connection to a high
vacuum. By relying on the vacuum gauge on the oven, glassware
in the oven preset to 200 °C was subjected to the following
conditions: vacuum (to 28 in. of Hg) and N2 fill (to 20 in. of Hg
and wait 10 min before next vacuum) 3×; vacuum (28 in. of Hg),
N2 (25 in. of Hg), inject 0.8 mL of 1,1,1,3,3,3-hexamethyldisilazane
(Aldrich, 99.9%), and stand overnight (vacuum 20 in. of Hg
observed next day); vacuum and N2 purge as above except 30
min of vacuum and N2 to 0 in. of Hg each time) 5×; turn off heat,
open oven, and remove glassware when cool enough to handle.
Synthesis. Dimethyl Amide, 2-Tritylamino-N-propionic Acid
Diethylammonium Salt. To a solution of 1.6 g (5 mmol) of
triphenylmethyl bromide in 10 mL of CHCl3/ DMF (2:1), 0.18 g
(2 mmol) of alanine was added, followed by vigorously stirring
until a clear solution resulted. After adding 8.0 mmol of TEA
dropwise in 2 mL of CHCl3/ DMF slowly and waiting 30 min, 10
mL of methanol was added followed by heating at 50 °C for 2 h.
Evaporation under reduced pressure gave a residue that was
dissolved in 20 mL of diethyl ether, and the solution was washed
with 3 × 10 mL each of 10% citric acid and then water. After drying
over anhydrous sodium sulfate, 2 mmol of dimethylamine in 2
mL of diethyl ether was added, followed by evaporation under
reduced pressure and purification by flash column chromatogra-
phy (acetone:hexane ) 2.5:1), giving a white powder (0.69 g,
86%)): 1H NMR (CDCl3, 300 MHz) δ 7.54-7.25 (m, 15H), 3.22-
3.10 (dd, 1H), 3.10-2.95 (dd, 4H), 1.30-1.27 (t, 6H), 0.90-0.86
(d, 3H). Propionamide,2-amino-N-[{(3,5-bis(trifluoromethyl)phenyl}-
methyl]-N-methyl-, Monohydrochloride (AMACE II). By relying on
the same reactions used to synthesize AMACE I,15 0.31 g of the
prior acid was converted into a white powder (0.26 g, 92%): 1H
NMR (DMSO, 300 MHz) δ 8.23 (s, 2H), 8.06-7.95 (m, 3H), 4.80-
4.65 (dd, 2H), 4.18 (s, 1H), 3.11 (s, 3H), 1.29-1.24 (d, 3H).
Reagent P urification. Methanol and acetonitrile were distilled
twice in this order before each procedure (use within same day)
in distillation apparatus 9317-03 (ACE Glass, Vineland, NJ) as
follows: charge a 250-mL round-bottom flask with 200 mL of
solvent (plus 2 g of LiAlH4 for the methanol) and a magnetic
stirring bar, distill at 20 drops/ min and discard first 20 mL, collect
100 mL in 200-mL round-bottom flask, and repeat the distillation
of the latter in the same way (but omitting LiAlH4) to yield 30 mL
in a 50-mL round-bottom flask that was glass-stoppered and kept
in a glass box until use. Between each solvent purification, the
distillation apparatus was air-dried and then oven-dried. Acetoni-
Chemicals and Materials. Glycolic acid (99% pure), triethy-
lamine, 1-[(3-dimethylamino)propyl]-3-ethylcarbodiimide hydro-
chloride (EDC), 2-(morpholino)ethanesulfonic acid (MES), 1-hy-
droxybenzotriazole hydrate (HOBT), 4-(dimethylamino)pyridine
(DMAP), 1,1,1,3,3,3-hexamethyldisilazane, and butyric anhydride
(all used as received) were purchased from Aldrich (Milwaukee,
WI). Glycolic acid-2,2-d2 (glycolate-d2, 98% isotopic purity) was from
Cambridge Isotope Laboratories, Inc. (Cambridge, MA). Aceta-
mide, 2-amino-N-[[3,5-bis(trifluoromethyl)phenyl]methyl]-N-meth-
yl monohydrochloride (AMACE1: Aminoacetamide Electrophore
1) was prepared as described.15 Triethylamine phosphorus pen-
toxide and all organic solvents (including those used for cleaning)
were Optima Grade from Fisher (Pittsburgh, PA). Glass inserts
for derivatization and HPLC purification (250 and 350 µL, types
500-304 and 500-305, respectively) were from Sun International
(Wilmington, NC). Reagent test tubes (14-961-26) for preparation
and dilution of stock solutions were from Fisher and were capped
as needed with aluminum foil followed by caps 540-001 from PGC
Scientific (Frederick, MD). All solutions were v/ v unless indicated
otherwise. “Al-P capping” refers to capping a vial or tube capped
firmly with aluminum foil and then Parafilm. Sonication was done
in a model 9331 apparatus (Elma, Dubuque, IA). Vortexing was
done on a Genie 2 (Fisher Scientific). Evaporations were done in
a SpeedVac SC 100 (Savant Instruments, Inc., Farmingdale, NY)
Automatic shaking at room temperature was done on a Mistral
Multi-Mixer (Lab-Line Instruments, Melrose Park, Ill). Heating/
stirring of the AMACE1 reaction at 40 °C was done in a Genie 2
placed in an oven. Double-distilled (dd) water was prepared by
first passing distilled water through a D3700 series Nanopure II
cartridge deionization system (Barnstead Co., Boston, MA), fitted
with an ion-exchange and a carbon column, and then distilling it
twice in an all-glass apparatus that was dedicated to this purpose,
rinsing the apparatus between the distillations with the water that
was just distilled.
Glassware Cleaning. Glassware was cleaned in one of three
ways. Procedure A (distillation apparatus including flasks for
synthesis or nanogram-level analysis): soak glassware overnight
in KOH-saturated 2-propanol (solution was reused until it became
colored; after ∼1 month); place in a tray and flush with tap water
for 4 h with flask emptying after each half hour; shake water out
and put onto Kimwipes EX-L paper in a clean basket until almost
dry; sonicate in methanol and then water for 15 min and cap
(13) de Vriendt, C. A.; Van Sassenbroeck, D. K.; Rosseel, M. T.; van de Velde,
E. J., Verstraete, A. G.; Vander Heyden, Y. V.; Blepaire, F. M. J. Chromatogr.,
A 2 0 0 1 , 752, 85-90.
(14) Adams, M. A.; Chen, Z., Landman, P.; Colmer, T. D. Anal. Biochem. 1 9 9 9 ,
266, 77-84.
(15) Lu, R. J.; Giese, R. W. Anal. Chem. 2 0 0 0 , 72, 1798-1801.
3050 Analytical Chemistry, Vol. 76, No. 11, June 1, 2004

trile was stirred with 5 g of P2O5 for 1 h at room temperature
before subjecting this mixture to the first distillation and, similarly,
1 g of P2O5 for the second distillation. In a second, equivalent
distillation apparatus, similarly distill triethylamine twice (recover-
ing in a 50-mL flask), apply a glass stopper, and store in a
refrigerator for use within 1 month. A separate stirring bar was
dedicated to each of the eight distillations that had first been
cleaned by sonication in its assigned solvent. The dd water was
prepared on the day of use as described earlier.
Stock Solutions. Glycolate. A 20-mg sample of glycolic acid
was dissolved in 1 mL of methanol, and a 4× dilution of 10 µL
into 1 mL gave 1 pg/ 5 µL for testing. The stock solution was stored
with AL-P capping.
MES, HOBT, and EDC Solutions. MES (213 mg) was dissolved
in 10 mL of methanol/ water (1:1) with stirring, diluted 100× to 1
mM, and the pH adjusted to 5.0 by 1 N NaOH (freshly prepared).
HOBT (6.8 mg) was dissolved in 1 mL of methanol and diluted
100× with pH 5.0 MES buffer to give 2.5 nmol/ 5 µL. EDC was
divided into 1-g amounts and placed in reagent test tubes followed
by nitrogen filling, AL-P capping, and storage in a vacuum
desiccator up to at least 6 months. Each tube was opened only
once, and then 8.4 mg was dissolved and diluted 100× in methanol
to give 2.5 nmol/ 5 µL. These solutions were used within 6 h of
preparation.
AMACE1 and TEA Solutions. AMACE1 (7 mg) was dissolved
in 0.5 mL of methanol and diluted 100× with pH 5.0 MES buffer
to give 2 nmol/ 5 µL. TEA (8.4 µL) was dissolved in 1.0 mL of
methanol and diluted to 3 nmol/ 5 µL for the AMACE1 reaction
and to 10 nmol/ 5 µL for butyrylation. These solutions were used
within 6 h of preparation.
Figure 1. Analytical scheme for detection of glycolic acid.
Determination of 1 0 -1 0 0 pg of Glycolate. Step 1. Capillary
reaction tubes were prepared by flame-sealing one end of a 100-
µL disposable glass pipet (Catalog No. 21-164-2H, Fisher, Pitts-
burgh, PA) with a Microflame Gas Torch (Microflame Inc.,
Plymouth, MA). Ten picograms of glycolic acid-2,2-d2 in water was
added to each tube followed in duplicate by 10, 20, 50, or 100 pg
of glycolic acid in water. The tubes were dried at room temper-
ature in a SpeedVac system and then charged with 10 µL of HCl-
saturated [18O]water. After each tube was flushed through a needle
with argon for 10 s, the closed end was touched to solid CO2, the
other end was flame-sealed, and the tube was placed in a 90 °C
oil bath for 3 h. The contents were transferred into a 250-µL glass
insert tube and dried under nitrogen at room temperature.
Step 2. To the dried insert tube was added 2 nmol of AMACE1,
2.5 nmol of HOBT, and 2.2 nmol of EDC (using the above stock
solutions), giving 12 µL in pH 5.0 methanolic MES. The tube was
Al-P capped, placed in a vial, and vortexed at room temperature
overnight.
Step 3. After evaporation under nitrogen, the residue was
dissolved in 10 µL of acetonitrile/ water, 9:1, vortexed for 20 s,
and autoinjected into the purification HPLC column of a bypass
HPLC system16 (which indirectly monitors retention time for
quality control), and the product in 0.2 mL was collected into a
borosilicate glass tube (12 × 75 mm, Fisher Scientific) followed
by evaporation in a SpeedVac system.
Step 4. The evaporated residue was transferred with 2 × 30
µL of acetonitrile into a new 250-µL insert tube and evaporated. A
total of 120 nmol of TEA was added followed by shaking (5 s)
and addition of 110 nmol of butyric anhydride and 110 nmol of
DMAP from the above stock solutions, yielding a total volume of
Instrumentation. A SPD 111V SpeedVac with a UVS400
universal vacuum system (Thermo Savant, Holbrook, NY) was
used for drying of glycolate and glycolate-d2 solutions. For HPLC
purification of AMACE1 derivatives of glycolate and glycolate-2,2-
d2 after step 2 (see Figure 1), a bypass HPLC method was used
as described.16 Two, nearly equivalent (6-month difference in prior
use) Zorbax SB-C18 columns (4.6 × 150 mm, 5 µm) were used in
a HP 1100 series HPLC system (Agilent Technologies, Wilming-
ton, DE) in contact with a G1313A autosampler and a G1315A
DAD detector. Acetonitrile/ 0.1% trifluroacetic acid (40:60) at a flow
rate of 1.2 mL/ min was used as the mobile phase with a detection
at 265 nm. Gas chromatography electron capture mass spectrom-
etry (GC-ECMS) was performed with a HP-5MS (HP-Ultra 2)
cross-linked 5% phenyl methyl siloxane capillary column (25 m ×
0.25 mm × 0.25 µm) installed in a GC 6890 series gas chromato-
graph connected to a 5973 Network mass-selective detector
controlled by an HP5973 MS ChemStation data system (Agilent).
An Agilent 7683 autosampler was used in the cold, on-column
injection mode with the inlet pressure at 18.5 psi. The oven
temperature was begun at 70 °C, ramped immediately after
injection at 35 °C/ min up to 290 °C, and held for 3 min. The source
was zoned at 230 °C with the MS Quad at 106 °C. Single ions
were monitored at 219, 217 and 215 m/ z with a dwell time of 50
ms. Helium (40 psi) and methane (20 psi; UHP Grade, Medical-
Technical Gases, Inc., Medford, MA) were used as carrier and
reagent gases, respectively.
(16) Shao, G.; Giese, R. W. J. Chromatogr., A 2 0 0 2 , 965, 233-237.
Analytical Chemistry, Vol. 76, No. 11, June 1, 2004 3051

15 µL in acetonitrile. The tube was Al-P capped, shaken at room
temperature overnight, and evaporated.
the nmol level) after 4 h at 40 °C. A 5-h reaction at 40 °C was
selected. Before step 1 was added to the procedure, we tested
alternate reagents in this derivatization reaction (dicydohexylcar-
bodiimide and diisopropylcarbodimide instead of EDC, and
cacodylic acid buffer instead of MES). This was not successful in
reducing glycolate contamination. Substituting AMACE2 for
AMACE1 also failed to reduce this contamination. The latter
experiment was motivated by the concern that the COCH2NH2
moiety of AMACE1 might degrade to form glycolate prior to or
during the procedure.
Methanol is a cosolvent in step 2. When we began to focus on
the problem of glycolate contamination, methanol was not one of
the first reagents we worried about. This is because we already
were doubly distilling HPLC-grade methanol on the day of use.
Further, methanol is a single-carbon reagent, which might obviate
its oxidative conversion to glycolate. But when the problem of
glycolate contamination persisted after we purified or substituted
the other reagents in the procedure, we turned our attention to
the methanol.
To investigate methanol more carefully as a potential source
of contamination by glycolate in step 2 of our method, and also
the water in this reaction at the same time, we conducted the
reaction in methanol-d4 (99.8 atom % D, Aldrich) and water-d2 (99.9
atom % D, Aldrich). (These solvents were used as received
because of their cost.) However, the contamination level of
glycolate increased 100-fold. Instead, more rigorous conditions
were studied for purifying the methanol. We redistilled the
methanol as separate batches in four different ways: from NaBH4,
metallic sodium spheres, KMnO4/ HCl, and LiAlH4. Relative to our
usual technique of purifying the methanol twice by distillation
without additives, all of these techniques, except the use of LiAlH4,
increased rather than decreased the level of contamination by
glycolate. Subsequently we purified the methanol by distilling it
from LiAlH4.
Apparently glycolate contamination is both ubiquitous and
dynamic: it can be constantly and irreproducibility produced by
oxidation of ubiquitous organic matter at the trace level in many
reagents and under many conditions, even purification conditions.
(We were reminded of our observation that a highly purified
reagentsessentially a single peak by capillary electrophoresis with
laser-induced fluorescence detectionswas actually contaminated
with many oxidative degradation products as seen by injecting
an off-scale amount, observing many nearby contamination peaks,
and then discovering that most of the peaks became intensified
upon addition of H2O2.19) At this point in our work, we added step
1, in which glycolate analyte is converted to 1,1-[18O]2-glycolate
as a way to minimize contamination by adventitious glycolate.
Step 3. Purify by HPLC. We studied several techniques for
removal of residual AMACE1 from intermediate VI (see Figure
1) before we selected HPLC for this purpose. (At this stage in
method development, only VI was formed.) These techniques fell
into the two categories of extractions (both liquid and solid
phases) and reactions (e.g., react residual AMACE1 with a liquid-
or solid-phase reagent such as poly(aspartic acid) active ester,
sulfosuccinic acid active ester, activated agaroses (aldehyde-,
Step 5. The residue was dissolved with 0.5 min of manual
shaking in 30 µL of acetonitrile and transferred into a 350-µL insert
tube followed by two 20 µL of similar transfers. A 30-µL aliquot of
2 M NaCl was added. After 2 min of vortexing, both phases were
pulled into a silanized Pasteur pipet. A sharp boundary formed,
and only the upper phase was transferred to a new 250-µL conical
insert followed by evaporation.
Step 6. The residue of one tube at a time was dissolved in 10
µL of hexane and placed in the GC autosampler containing two
washing solvents (methanol and ethyl acetate) in separate vials.
After washing the injector to waste with 3 × 4 µL of methanol
and 2 × 4 µL of ethyl acetate, 1 µL of the sample was injected.
RESULTS AND DISCUSSION
Our method for determining glycolate is summarized in Figure
1. In this method, the carboxyl oxygens of glycolate are exchanged
with 18O at the outset to distinguish analyte glycolate from
background (method-derived) glycolate. The following discussion
is organized according to the six steps of this procedure.
A. Analytical P rocedure. Step 1. Add Internal Standard
(HOCD2CO2H) and React with HCl/ [18O]Water. This step was
introduced into our procedure after we were unsuccessful in
overcoming glycolate contamination at the low-picogram level by
purifying reagents, using clean conditions, and testing alternative
reagents and conditions (see below). The conditions that we
adopted (90 °C for 3 h in [18O]water saturated with gaseous HCl)
were selected after we found this exchange reaction to proceed
much slower when conducted at room temperature or with 2 M
HCl at 60 °C. The reaction was done in a small volume (10 µL) in
a flame-sealed glass tube, since [18O]water is expensive, and for
the purpose of minimizing contamination. The carboxyl oxygens
of amino acids have been exchanged in 1 N HCl in [18O]water at
60 °C for 2-3 days,17 and of valproic acid (9.8% single labeled
and 90.2% double labeld) in 0.4 M HCl at 75 °C after 4 days.18
Step 2. Derivatization with AMACE1. In this step, 18O-
exchanged glycolate and glycolate-2,2-d2 in MES buffer are
derivatized with AMACE1 in the presence of HOBT and a water-
soluble carbodiimide (EDC) to form products VI-IX (Figure 1).
Conditions for this step were optimized by testing nanogram
amounts of glycolate where the yield of product could be
monitored by HPLC. At this level of analyte, the final conditions
selected gave 90% conversion of glycolate to product. Initially
N-hydroxysulfosuccinimide was tested instead of HOBT, but
HOBT was easier to purify. As reported before, the low pKa (8.2)
of AMACE1, by design, enables it to react efficiently with active
ester derivatives of carboxylic acids at pH 6.0 in MES buffer.15
Here we found that the yield is slightly higher at pH 5.0, but it
drops to 40-50% when the reaction is conducted outside the pH
range of 4-7. Initially, 100 mM MES was employed when
nanomole levels of glycolate were tested, but later the concentra-
tion of this buffer was lowered to 1 mM in order to yield a smaller
residue of it after evaporation prior to the next step. While this
step also can be performed at room temperature for 24 h, we
observed that the yield similarly went to completion (tested at
(19) Li, W.; Moussa, A.; Giese, R. W. J. Chromatogr., A 1 9 9 3 , 633, 315-319.
(20) Abushamaa, A.; Naim, A.; Giese, R. W. J. Chromatogr., A 1 9 9 6 , 742, 295-
297.
(17) Murphy, R.; Clay, K. Methods Enzymol. 1 9 9 0 , 193, 338-348.
(18) Leis, H. J.; Windischhofer, W.; Rechberger, G. N.; Fauler, G. J. Chromatogr.,
B 2 0 0 3 , 784, 69-75.
(21) Saha, M.; Giese, R. W. J. Chromatogr., A 1 9 9 3 , 631, 161-163.
3052 Analytical Chemistry, Vol. 76, No. 11, June 1, 2004

Figure 2. GC-ECMS chromatograms obtained by subjecting 1 pg of glycolate-2,2-d2 (a) and a blank (b) to the steps 2-6 of the procedure
shown in Figure 1. Of the final sample volume of 10 uL, 1 uL was injected.
N-hydroxysuccimidyl- and 1,1′-carbonyldiimidazole-activated aga-
rose from Sigma, and epoxy-activated agarose from Pierce) and
a 4-benzyloxybenzaldehyde polystyrene resin from NovaBiochem
(La Jolla, CA). Use of these techniques was fairly successful at
the nanomole level (e.g., removal of 70-90% AMACE1 with a 80-
90% recovery of VI by using 4-benzyloxybenzaldehyde resin or
aldehyde-agarose from Sigma), but 100% removal of AMACE1 was
never observed, and higher amounts of these reagents lowered
the recovery of VI or added interferences, especially when smaller
amounts of VI were tested. HPLC vastly outperformed the other
strategies that we tested (complete removal of AMACE1, no added
interferences, quantitative recovery of VI), and so was adopted.
Step 4. Derivatization with Butyric Anhydride. The butyrylation
reaction was initially performed in pyridine in anhydrous aceto-
nitrile at 60 °C overnight. While the reaction proceeded quanti-
tatively and cleanly for micromole amounts of glycolate, interfer-
ences showed up at lower levels of glycolate, and the yield dropped
markedly, for example, to 83% and then 19% when 320 and 40
pmol, respectively, of glycolate were derivatized. High yield and
absence of interferences were maintained at this level by using
triethylamine (TEA) instead of pyridine and conducting the
reaction at room temperature. Lowering the reaction temperature
is a well-known technique for minimizing interferences in trace
derivatization.
M NaCl and inject into HPLC. This led to GC-ECMS chromato-
grams in which the peak for X progressively tailed as samples
were injected. We postulated that either residual triethylamine
or NaCl in the HPLC fraction for X was responsible for this
behavior. A subsequent experiment involving the presence and
absence of these ingredients demonstrated that triethylamine was
the origin of this problem, which, in part, led to the final conditions
for step 5 (removal of triethylamine by evaporation) that avoid
this problem.
Step 6. GC-ECMS. This is the first time that we have
conducted trace GC-ECMS with a GC autoinjector. By redis-
solving our samples one at a time in hexane for injection in this
way, we were able to handle redissolved volumes as small as 3
µL, where 1 µL was injected. A comparison of the splitless, pulsed
splitless, and cold on-column injection modes with the autoinjector,
conducted at the 10-fg level, showed that the latter technique gave
about a 2-fold higher response than the others and therefore was
selected. Loss in peak efficiency (after ∼150 injections) was
overcome by removing the first 20 cm of the GC column.
B. Detection of 1 pg of Glycolate-2 ,2 -d 2 . We first encoun-
tered the contamination problem of background glycolate when
we began to develop our method at the subnanogram level, after
conditions had been set up that performed well at the microgram
and nanogram levels. Prior to bringing step 1 (18O exchange
reaction) into the procedure, we tested glycolate-2,2-d2 as an
analyte in order to focus on recovery. For this analyte, contamina-
tion was not a problem. As shown in Figure 2, 1 pg of glycolate-
2,2-d2 can be detected by subjecting it to steps 2-6 of Figure 1.
The absolute yield throughout the overall procedure is 76% (SD
) 17%, n ) 3). The precision for this measurement is low due to
the absence of a stable isotope internal standard.
Step 5. Partition with Acetonitrile/ 2 M NaCl. Initially HPLC
was used successfully for step 5, but later work showed that
significant purification could be achieved by organic/ aqueous
partitioning. We chose a nonsaturating concentration (2 M) of
NaCl in the aqueous phase to minimize NaCl fines in the
acetonitrile phase. When 1 pg of standard X was extracted using
these conditions, a quantitative recovery was observed.
At one stage in method development (prior to adding step 1),
step 5 was done as follows: extract the reaction solution with 2
Detection of 1 0 pg of Glycolic Acid. As explained above,
much work remained after 1 pg of glycolate-2,2-d2 was detected
Analytical Chemistry, Vol. 76, No. 11, June 1, 2004 3053

Figure 3. Representative GC-ECMS chromatograms for the detection of 10 pg of glycolate according to the scheme shown in Figure 1.
Blank peak heights: X, 294,570; XI + XII, 119,048; XIII, 24,390. Sample peak heights: X, 271,818, XI + XII, 127,065: XIII, 31, 956.
before we were able to manage glycolate contamination (by
introducing step 1), so that glycolate itself could be detected at a
trace level (10 pg). Representative GC-ECMS chromatograms
for the blank and 10-pg samples are shown in Figure 3. As seen
in Figure 3A, even the blank sample gives a peak for glycolate,
apparently because of glycolate contamination in step 1, step 2,
or both, of the procedure. Since glycolate-d2 is added to the blank,
and the exact percent of [18O] exchange at the trace-analyte level
is unknown, then it is unclear how much the peak at m/ z 217 is
due to [18O]glycolate versus residual glycolate-2,2-d2. Only the
percentage of the glycolate peak in Figure 3B (2.3%) due to the
presence of natural isotopic forms of glycolate (1.1% of the
glycolate peak in Figure 3A) is known. Because of this uncertainty,
one can only set up an empirical calibration plot, based on the
amounts of authentic glycolate present in a series of standard
samples. Plotting the ratio of peak height at m/ z 217 to that of
m/ z 219 against added amounts of glycolate (5 data points, each
from duplicates) from 0 to 100 pg gave a linear curve (y ) 7.8329X
+ 0.1246 with r2 ) 0.9993; data not shown). In the calculation for
this plot, the peak height of m/ z 217 first was corrected by
removing the isotopic contribution (1.1%) from m/ z 215 to 217.
CONCLUSION
A sensitive method is reported for detecting standard glycolate
in water. Application of the method to biological and other real
samples remains to be done. Highly optimized conditions for
sample handling including derivatization explain the trace detec-
tion limits achieved. These high-performance conditions should
be useful not only in applying the method to related analytes but
also more generally in setting up procedures of this type. If the
other fractions of the HPLC separations were collected from real
samples and subjected to GC-ECMS, many analytes of this type
no doubt would be detected.
ACKNOWLEDGMENT
This work was supported by NIH Grant CA 71993 received as
a subcontract from Harvard Medical School. Contribution 835
from the Barnett Institute.
Received for review December 30, 2003. Accepted March
13, 2004.
AC0304267
3054 Analytical Chemistry, Vol. 76, No. 11, June 1, 2004