Anal Bioanal Chem (2006) 386: 306312
DOI 10.1007/s00216-006-0664-2
ORIGINA L PA PER
Francisca Vilar . Anna Canela-Xandri .
Ramon Canela
Quantification of abscisic acid in grapevine leaf (Vitis vinifera)
by isotope-dilution liquid chromatographymass spectrometry
Received: 24 March 2006 / Revised: 26 June 2006 / Accepted: 30 June 2006 / Published online: 26 July 2006
# Springer-Verlag 2006
Abstract A specific, sensitive, precise, and accurate
method for the determination of abscisic acid (ABA) in
grapevine leaf tissues is described. The method employs
high-performance liquid chromatography and electrospray
ionizationmass spectrometry (LC-ESI-MS) in selected ion
monitoring mode (SIM) to analyze ABA using a stable
isotope-labeled ABA as an internal standard. Absolute
recoveries ranged from 72% to 79% using methanol/water
pH 5.5 (50:50 v/v) as an extraction solvent. The best
efficiency was obtained when the chromatographic sepa-
ration was carried out by using a porous graphitic carbon
(PGC) column. The statistical evaluation of the method
was satisfactory in the work range. A relative standard
deviation (RDS) of < 5.5% and < 6.0% was obtained for
intra-batch and inter-batch comparisons, respectively. As
for accuracy, the relative error (%Er) was between 2.7 and
4.3%, and the relative recovery ranged from 95% to 107%.
Keywords Abscisic acid . Isotopically labeled internal
standard . LC-ESI-MS . Vitis vinifera . Grapevine leaf
Introduction
Plant hormones are natural products that act at low
concentrations and regulate important processes during a
plants life cycle. Among these, abscisic acid (ABA), a
monocyclic sesquiterpenoid, is involved in the regulation
of many stress-induced gene-expressions and confers the
plant with adaptability towards drought, salinity, cold, and
F. Vilar
Centre UdL-IRTA,
Rovira Roure 191,
25198 Lleida, Spain
A. Canela-Xandri . R. Canela (*)
Chemistry Department, Lleida University,
Rovira Roure 191,
25198 Lleida, Spain
e-mail: canela@quimica.udl.es
Tel.: +34-973-702843
Fax: +34-973-238264
other environmental stresses [1]. In plant cells, ABA is
continually synthesized and degraded. Catabolism can then
occur by several routes involving oxidation, reduction, and
conjugation [2]. The study of the variation of ABA
concentration in different plant tissues can therefore
provide knowledge on the ABA biosynthesis and degra-
dation pathways and the correlation with different physi-
ological and environmental parameters [35].
In recent years, some studies carried out to increase
water efficiency have shown that the rational control of
irrigation in grapevine promotes the biosynthesis of ABA.
The increase in ABA content reduced vegetative vigor and
increased the antocyanin concentration in berries. These
phenomena occurred without a significant reduction in
crop and berry size [6, 7]. These results suggested that the
ABA level in leaf tissue could be used as a quality index for
the wine obtained. An accurate and fast analytical method
is needed to determine ABA contents in plant tissues.
Numerous analytical techniques have been used to detect
and quantify ABA in plant tissues. Although the most
commonly used techniques are based on gas chromatogra-
phymass spectrometry (GC-MS), they involve several
steps of intense purification and a derivatization step prior to
analysis [810]. ELISA tests using anti-ABA monoclonal
antibodies have also been used to determine ABA in plant
tissues [11]; however, this technique also requires a
purification step to avoid cross-reactivity with other plant
compounds.
In recent years, liquid chromatographytandem mass
spectrometry (LC-MS/MS) techniques have been devel-
oped for the analysis of this phytohormone in some plant
tissues, although not in grapevine, that do not require
purification and/or derivatization steps [1215]. In addi-
tion, an easily prepared isotopomer of the analyte has been
described as an internal standard (IS) [12]. The use of a
stable isotope-labeled internal standard for ABA quantifi-
cation would provide the necessary precision and accuracy
by compensating for both ABA losses in the extraction step
and the non-specific matrix effects caused by co-eluting
components [16, 17]. However, it is well accepted that the
sample preparation for the quantification of ABA is very

different depending on the sample matrix and the detection
technique applied [1215]. For example, Ross et al. [13]
showed that extraction and clean-up procedures for two
different vegetal matrices should be different to obtain
accurate results.
In this study, we present an analytical method developed
for specific detection and accurate quantification of ABA
in grapevine leaf using LC-MS with negative ion
electrospray ionization (ESI). Hexadeutero-ABA [12]
was used as an internal standard. We demonstrated the
usefulness of the method by quantifying ABA in several
grapevine leaf samples collected under various irrigation
conditions.
Experimental
Chemicals
()-Abscisic acid [CAS Registry N 14375-45-2, 99%
purity, ref. A1049] (see Fig. 1), deuterium oxide (isotopic
purity 99.96%, ref. 151890), and 30% wt. deuterated
sodium hydroxide in deuterium oxide (minimum isotopic
purity 99%, ref. 164488) were acquired from Sigma
Aldrich Qumica S.A (Madrid, Spain). Acetone and aceto-
nitrile (HPLC-grade purity) were purchased from J.T. Baker
(Deventer, The Netherlands). Formic acid (98100% extra
pure quality) was purchased from Merck (Barcelona, Spain).
Methanol, dichloromethane, and diethyl ether (PA-grade
purity) were purchased from Panreac (Barcelona, Spain).
Water was prepared using a Milli-Q reagent water system.
Specifications for columns used for analysis are cited below.
Preparation of deuterated abscisic acid
[2H6]-ABA was synthesized using a base exchange mech-
anism on ()-ABA as described elsewhere [12, 18]. We
evaluated the purity of the [2H6]-ABA obtained by LC-MS
using the conditions indicated below. No unlabeled ABA
was detected in the [2H6]-ABA synthesized.
Preparation of standard solutions
Stock standard solutions of ABA (189 nmol mL1) and
[2H6]-ABA (315 nmol mL1) were prepared by dissolving
appropriate amounts of each compound in 30% acetonitrile
in water containing 0.1% formic acid (standards solvent).
The solutions were stored in the dark at 18 C and were
found to be stable for at least 3 months.
Fig. 1 Chemical structure of H3C CH3 CH3
abscisic acid (ABA)
OH
O CH3 HO O were tested, a 1502.1-mm i.d. Symmetry C18 packed with
307
Intermediate standard solutions of ABA and [2H6]-ABA
were prepared by tenfold dilution of the stock standard
solutions with the standards solvent. These solutions were
freshly prepared every month and stored in the dark at 18 C.
The standard levels used for the calibration curve were
800, 400, 200, 100, and 50 ng mL1. They were prepared
by dilution of intermediate standard solutions with the
standards solvent. [2H6]-ABA was added as an internal
standard to each solution to achieve a concentration of
700 ng mL1 for the LC-MS analysis. These standard
solutions were freshly prepared every month and stored in
the dark at 18 C.
Sample preparation
Grapevine (Vitis vinifera) leaves were collected from
vineyards in Lleida, Spain. Each sample was immediately
frozen in liquid nitrogen, lyophilized, and milled to obtain
a fine powder. The processed samples were stored under
dry conditions at room temperature.
Extracting solvent test
Grapevine leaf samples were spiked with [2H6]-ABA at
2 nmol g1 level and extracted as described below. One of
the following solutions was used in each case: water at pH 3,
water at pH 5.5, 1% sodium bicarbonate (NaHCO3)
solution, or methanol/water pH 5.5 (50:50 v/v). The essays
were carried out in triplicate.
Extraction procedure
One hundred milligrams of powder material was placed in a
50-mL glass beaker and mixed with 1.26 nmol of internal
standard (40 L of 31.5 nmol mL1 intermediate standard
solution) using a magnetic stirrer for 5 min. The sample was
extracted with 10 mL methanol/water pH 5.5 (50:50 v/v)
using a magnetic stirrer for 20 min. The mixture was poured
through a polyethylene frit with 20-m pore size (Supelco,
Sigma-Aldrich S.A., Madrid, Spain). The remaining solid
was extracted by repeating the same procedure. The filtrates
were joined and extracted twice with 10 mL dichloro-
methane. The dichloromethane extracts were joined and
evaporated under vacuum conditions at 40 C. The residue
was dissolved in 100 L acetone and 250 L water/
acetonitrile (70:30 v/v) (0.1% formic acid).
LC-MS conditions
Analyses were performed using an in-line LC-MS system
equipped with an ESI ion source. The LC separations were
carried out at room temperature and in the isocratic mode
using a Waters Alliance 2695 separation module (Waters
Corp.). The injection volume was 20 L. Two columns
308

5-m particles (Waters Corp., Barcelona, Spain) and a
1002.1-mm i.d. Hypercarb PGC packed with 5-m par-
ticles (Hypersil division of TermoQuest, Barcelona, Spain).
The mobile phase was water/acetonitrile (70:30 v/v) (0.1%
formic acid) at a flow rate of 0.2 mL min1.
The MS detector (Micromass ZMD 2000) consisted of a
single-quadrupole mass spectrometer equipped with a
Z-spray API source (Waters Corp.). The MS was operated
in the negative ESI mode. The source was operated with N2
(LCMS quality) at a 400 L h1 gas flow. MS parameter
were optimized by flow injection analysis (FIA) of the
individual solutions of the ABA and [2H6]-ABA. These
compounds gave a response in the ESI interface in the
negative mode but not in the positive mode. The capillary
voltage was varied from 2.0 to 3.5 kV and the cone voltage
from 10 to 80 V. Optimized conditions for MS were:
desolvation temperature, 350 C; source temperature,
120 C; capillary voltage, 3.0 kV; cone voltage, 30 V;
extractor voltage, 5 V.
For the ABA and [2H6]-ABA, the most abundant and
characteristic ion was chosen for quantification ([M
C6H6O2]) and two ions ([MH] and [MCO2]) were
selected for confirmation (Table 1). The fragmentation
pathways for the deprotonated molecules were reported by
Gmez-Cadenas et al. [12]. Analyses were carried out in
SIM mode with a dwell time of 500 ms. Masslynx version
4.0 software (Micromass) was used to process the quanti-
tative data obtained from calibration standards and from
samples.
Limits of detection (LODs) and quantification (LOQs)
were evaluated from the y-intercept standard deviation (Sb)
and the slope (a) of the calibration curve [1921].
The precision was estimated by the evaluation of the
intra-batch precision (measure repeatability) and the inter-
batch precision (preparation process repeatability). The
intra-batch precision was determined by a set of six
replicate analyses of two samples with different concen-
tration levels of ABA. The inter-batch precision was
evaluated by a set of four extracts of four samples with
different concentration levels.
Accuracy and relative recovery were evaluated over the
linear range at three concentrations. The essays were
carried out using five replicates for each concentration.
Statistical analysis
The t-test and F-test were performed by using Microsoft
Excel version 2000 (Microsoft Corp.); one-way analysis of
variance (ANOVA) and the Duncan test were carried out by
using SAS version 8.02 (SAS Institute, Inc.). All statistical
tests applied in this work were employed to validate the
analytical method.
Results and discussion
LC-MS conditions

Development of the analytical procedure involves the

Validation procedures optimization of analytical conditions to obtain the best
results. Using FIA of the individual solutions of the ABA
Specificity, calibration curve estimation, sensitivity, preci- and [2H6]-ABA, maximum sensitivity was obtained using a
sion, accuracy, and relative recovery were investigated to 3.0-kV capillary voltage and a 30-V cone voltage.
evaluate the integrity of the method.
Specificity was based on relative retention time and
relative abundance of masses obtained in the SIM mode. Extraction procedure
Calibration curves were tested for the linearity, the
interday repeatability of the slopes, and the intercept test Effect of extracting solvent on ABA recoveries
significance. The linearity was investigated by calculation of
the regression curve by the method of least squares and Besides affecting the extraction yield of ABA and [2H6]-
expressed by the determination coefficient (R2). Comparison ABA, the extraction solvent can influence the specificity of
of the interday slopes was carried out by the application of the method. Table 2 shows the results obtained when
t-test at significant level =0.05. The intercept test several spiked samples were extracted with water pH 3,
significance was carried out by means of a t-test at a
significance level of =0.05. Table 2 Influence of extraction solvents on ABA recoveries.
Number of replicates, average recoveries (%), and relative standard
deviation (%RSD) are indicated
Table 1 Selected ions for the LC-MS determination of ABA and
[2H6]-ABA Solvent na Mean %RSD Duncan test
recovery (=0.05)
Compound Quantification First Second
ion confirmation ion confirmation ion Water pH 3 3 50 3.0 A
m/z Relative m/z Relative m/z Relative Water pH 5.5 3 56 3.8 B
intensity intensity intensity 1% NaHCO3aqueous solution 3 53 1.6 AB
(%) (%) (%) Methanol/water pH 5.5 3 75 4.6 C
(50:50 v/v)
ABA 153 100 263 56 219 25 a
Number of replicates
[2H6]-ABA 159 100 269 56 225 25 The spiking level is 2 nmol g1
 309

Fig. 2 LC-ESI-MS composite mass chromatograms and the and with a C18 column (C SIM for ABA and D [2H6]-ABA).
corresponding spectra resulting from the analysis of a grapevine Retention times (RTs) for both compounds are indicated in each
leaf extract with a PGC column (A SIM for ABA and B [2H6]-ABA) chromatogram
310
water pH 5.5, 1% NaHCO3 aqueous solution, or methanol/
water pH 5.5 (50:50 v/v). The extraction yields varied from
50% to 75% with relative standard deviations lower than
5%. The Duncan multiple comparison procedure indicated
that recoveries obtained using methanol/water pH 5.5
(50:50 v/v) were significantly different from the other three
procedures at the =0.05 level (Table 2). Furthermore, this
procedure gave the best absolute recoveries. Consequently,
this solvent was selected for use in all the subsequent
essays.
Effect of LC column on method specificity
Chromatographic separation of the different compounds of
the matrix sample is an important factor for obtaining a
good specificity because this separation influences the
methods efficiency. Figure 2 shows the chromatograms
and their respective spectra obtained from grapevine leaf
extracts using a C18 column (Symmetry C18 1502.1-mm
i.d., 5 m) and a porous graphitic carbon (PGC) column
(Hypercarb PGC 1002.0-mm i.d., 5 m). The relative
intensity obtained with each column and compound
demonstrated which chromatographic column should be
chosen. The PGC column made it possible to obtain a
series of chromatographic peaks presenting a relative
intensity of ions that were closer to the desired pure
standards (Table 1). Consequently, the PGC column was
selected for all subsequent essays. The total time needed for
one LCMS analysis was 15 min.
Method validation
Calibration integrity
Calibration curves were obtained by least-squares linear
regression analysis of the peak area ratio of analyte to
internal standard (y) versus analyte concentration (x). Five
standard solutions were employed to determine the
calibration curve. Two replicate measurements were
made for each standard solution. The interday calibration
behavior was linear with determination coefficients (R2)
between 0.9990 and 0.9997. The calibration curve was
expressed by:
y 2:652
0:015  x 2:835
6:268
Two Students t-test was performed to determine the
calibration curve quality. The interday slopes comparison
was carried out by using three calibration curves obtained on
three different days. For all cases tcalculated value<tcritical value
for 16 degrees of freedom at the =0.05 level (Table 3). The
second t-test (9 degrees of freedom) was used to compare
the y-intercept values of the three calibration curves with
zero. Table 3 shows that for all cases tcalculated value<tcritical value.
Consequently, there were no significant differences between
y-intercept and zero at =0.05. These results suggest that the
Table 3 t-Test values for the comparison of three calibration curves
obtained on three different days at =0.05
Interday slope comparison y-intercept comparison with
(16 dfa) zero (9 df)
Curve tcalculated value tcritical value tcalculated value tcritical value
1 0.06 1.43
2 0.40 2.12 1.28 2.26
3 0.29 0.43
a
Degrees of freedom
calibration curve presented good linearity and interday
stability.
Lower limits
The sensitivity was evaluated by determining the LODs
and LOQs. Both parameters were respectively calculated
by using the equations LOD=3.3Sb/a and LOQ=10Sb/a,
where a is the slope and Sb is the standard deviation of the
y-intercept [17]. Values of 0.2 nmol g1 and 0.6 nmol g1
dry grapevine leaves were calculated for LODs and LOQs,
respectively.
Precision
The intra-batch precision of the method based on mea-
surement repeatability was assessed by replicate injections
(n=6) of two sample extracts with different concentration
levels, i.e., 3.89 and 6.24 nmol g1 dry sample. The
ANOVA test was used to estimate significant differences at
=0.05 among the different instrumental measures. The
results demonstrated that the instrument did not influenced
the response obtained (Fcalculated value (0.03) < Fcritical value
(4.39)). Table 4 shows that the intra-batch precision,
calculated by relative standard deviation, did not exceed
5.5%.
The inter-batch precision of the method based on
extraction repeatability was assessed by replicate extraction
(n=4) of four samples with different concentration levels,
i.e., 3.17, 3.44, 4.18, and 6.53 nmol g1 dry sample. The
ANOVA test was used to estimate significant differences at
=0.05 among the different extracts of the sample. Results
(1) demonstrated that the extraction procedure did not influence
the concentration value obtained (Fcalculated value (0.00) <
Fcritical value (3.49)). Table 4 shows that the inter-batch
precision, calculated by relative standard deviation, was less
than 6.0%.
Accuracy and relative recovery
These parameters describe the closeness of mean test
results to the true concentration of the analyte obtained by
the method. Although the absolute recovery of the method
was around 75%, the use of an isotopically labeled internal
 311
Table 4 Repeatability results for ABA in grapevine leaf samples
Precision data
Intra-batchc Inter-batchd

Level Low High Low Medium1 Medium2 High

Meana 3.89 6.24 3.17 3.44 4.18 6.53
SDb 0.21 0.22 0.13 0.20 0.13 0.16
%RSD 5.40 3.53 4.10 5.81 3.11 2.45
a
Concentration (nmol g1 lyophilized sample)
b
Standard deviation
c
Six replicate analysis of two real samples
d
Four replicate extractions of four real samples

standard enabled relative recovery values adjusted to the
theoretical value to be obtained. The standard addition
method was used to compare the difference between the
results obtained in the sample with addition and the sample
with no addition (real value). The theoretical value of the
addition was considered as a reference value through the
relative error (%Er) or t test [22, 23]. Spiked grapevine leaf
samples at three different concentrations of ABA (1, 3.5,
and 7.5 nmol g1) were processed (five replicates) and
analyzed (two replicates). The results are shown in Table 5.
%Er lays in the range of 2.7 to 4.3% showing a high level
of accuracy. Moreover, the assayed mean values are not
statistically different from the theoretical values using a
t test at =0.05, suggesting a good level of accuracy in the
method. An additional parameter to validate the analytical
methods is the relative recovery. This parameter ranged
from 95% to 107% at the levels tested, owing to the
compensation of the analyte loss by the isotope-labeled
internal standard.
Application
Twenty five lyophilized grapevine leaf samples were
processed and analyzed according to the optimized meth-
odology in order to demonstrate their feasibility. The
grapevine crops were subjected to six different irrigation
conditions in order to obtain a large range of ABA content
in the leaf samples. A total of 24 samples (three extractions
for sample and two injections for extract), four samples in
each irrigation condition, were investigated for application
of the analytical method. The ABA concentration showed
levels between 2.5 and 10 nmol g1 dry sample between the
calibration range. %RSD levels between 1 and 10
demonstrate that the method is very precise.
A certain tendency towards an increase in ABA content
was observed when the cultivar was submitted to a certain
hydric stress. Nevertheless, the content of this phytohor-
mone also depended on factors like salt, stress damage, and
low temperature adaptations [3, 24], data which were not
available in our study.
Conclusions
This work shows that isotope-dilution LC-ESI-MS under
negative ionization could be a valuable tool for quantitative
determination of ABA in grapevine leaf. Although most of
the LC-MS ABA analytical procedures reported use
tandem mass spectrometry (MS/MS) in multiple reaction
monitoring (MRM) mode, the use of LC-MS in ABA
analysis presents advantages in terms of instrumental cost
and availability. The method developed is fast and easy to
handle, and the instrumental equipment needed is present
in most routine laboratories nowadays. The method
includes an extraction step with mean recovery around
the 75% and is specific because it permits a good
separation of the matrix compounds. Moreover, the
calibration curve presents good linearity (R2>0.9990) and
interday stability. The method has been validated and has

Table 5 Method accuracy and ABA relative recoveries in grapevine leaf samples (nmol g1 lyophilized sample)
Accuracy Recovery

Level Spiked Na nb Means of SD %RSD Theoretical tcalculated Test result %Erd Mean SD %RSD
ABA measured conc conc (tcritical=2.78)
Blank 0.0 5 2 3.17 0.09 2.84
Low 1.0 5 2 4.14 0.16 3.86 4.17 0.47 NSDc 0.7 100.3 7.19 7.17
Medium 3.5 5 2 6.49 0.19 2.93 6.67 2.23 NSD 2.7 95.2 5.43 5.71
High 7.5 5 2 11.13 0.44 3.95 10.67 2.29 NSD 4.3 106.6 6.09 5.72
a
Extraction replicates
b
Injection replicates
c
No statistical differences between mean of measured concentration and theoretical concentration (tcalculated< tcritical)
d
Relative error calculated by %Er=[(Mean of measured conc-theoretical conc)/theoretical conc]100
312
been shown to be sensitive (LOD=0.2 nmol g1 dry sample
and LOQ=0.6 nmol g1 dry sample), accurate (%Er
between 2.7 and 4.3%), and intra-batch and inter-batch
precise (%RSD 5.5 and 6.0% respectively). Finally, it
presents excellent relative recoveries (ranging from 95% to
107%). The method is at least as specific, sensitive, precise,
and accurate as the LC-MS/MS methods described for
other plant tissues. Consequently, this analytical method
could be a useful tool for the grapevine growers and wine-
making industry.
Acknowledgement The authors are grateful to the Departament de
Tecnologia Frutcola of the Institut de Recerca i Tecnologa
Agroalimentries (IRTA) for providing grapevine leaf samples.
References
1. Bray EA (2002) Plant Cell Environ 25:153161
2. Cutler AJ, Krochko JE (1999) Trends Plant Sci 4:472478
3. Xiong LM, Zhu JK (2003) Plant Physiol 133:2936
4. Soar CJ, Speirs J, Maffei SM, Loveys BR (2004) Funct Plant
Biol 31:659669
5. Koussa T, Colin L, Broquedis M (2004) J Int Sci Vigne Vin
38:141146
6. McCarthy MG (1997) Aust J Grape Wine Res 3:102108
7. Stoll M, Loveys B, Dry P (2000) J Exp Bot 51:16271634
8. Kamboj JS, Browning G, Blake PS, Quinlan JD, Baker DA
(1999) Plant Growth Regul 28:2127
9. Muller A, Duchting P, Weiler EW (2002) Planta 216:4456
10. Birkemeyer C, Kolasa A, Kopka J (2003) J Chromatogr A
993:89102
11. Gomez-Cadenas A, Tadeo FR, Talon M, Primo-Millo E (1996)
Plant Physiol 112:401408
12. Gomez-Cadenas A, Pozo OJ, Garcia-Augustin P, Sancho JV
(2002) Phytochem Analysis 13:228234
13. Ross ARS, Ambrose SJ, Cutler AJ, Feurtado JA, Kermode AR,
Nelson K, Zhou R, Abrams SR (2004) Anal Biochem 329:
324333
14. Lopez-Carbonell M, Jauregui O (2005) Plant Physiol Bioch
43:407411
15. Zhou R, Squires TM, Ambrose SJ, Abrams SR, Ross ARS,
Cutler AJ (2003) J Chromatogr A 1010:7585
16. Jemal M, Schuster A, Whigan DB (2003) Rapid Commun Mass
Spectrom 17:17231734
17. Liu RH, Lin DL, Chang WT, Liu CR, Tsay WI, Li JH, Kuo TL
(2002) Anal Chem 74:618A626A
18. Abrams SR, Nelson K, Ambrose SJ (2003) J Label Compd
Radiopharm 46:273283
19. Food and Drug Administration (1995) International Conference
on Harmonisation [ICH]. Validation of analytical procedures:
definitions and terminology, Q2A. Federal Register 60:11260
20. Food and Drug Administration (1997) International Conference
on Harmonisation [ICH]. Validation of analytical procedures:
methodology, Q2B. Federal Register 62:27464
21. Baker SE, Olsson AO, Needham LL, Barr DB (2005)
Anal Bioanal Chem 383:963976
22. Garcia I, Ortiz MC, Sarabia L, Vilches C, Gredilla E (2003)
J Chromatogr A 992:1127
23. Feinberg M, Boulanger B, Dewe W, Hubert P (2004) Anal
Bioanal Chem 380:502514
24. Mauch-Mani B, Mauch F (2005) Curr Opin Plant Biol 8:
409414