Profiling and characterization by LC-MSn of the chlorogenic acids and hydroxycinnamoylshikimate esters in mate (Ilex paraguariensis)

Article abstract of DOI:10.1021/jf904537z

The chlorogenic acids of mate (Ilex paraguariensis) have been investigated qualitatively by LC-MSⁿ. Forty-two chlorogenic acids were detected and all characterized to regioisomeric level on the basis of their fragmentation pattern in tandem MS spectra, 24 of them for the first time from this source. Both chlorogenic acids based on trans-and cis-cinnamic acid substituents were identified. Assignment to the level of individual regioisomers was possible for eight caffeoylquinic acids (1-8), five dicaffeoylquinic acids (20-24), six feruloylquinic acids (9-14), two diferuloyl quinic acids (25 and 26), five p-coumaroylquinic acids (15-19), four caffeoyl-p-coumaroylquinic acids (34-37), seven caffeoyl-feruloylquinic acids (27-33), three caffeoyl-sinapoylquinic acids (38-40), one tricaffeoylquinic acid (41), and one dicaffeoyl-feruloylquinic acid (42). Furthermore, four caffeoylshikimates (43-46), three dicaffeoylshikimates (47-49), one tricaffeoylshikimate (51), and one feruloylshikimate (50) have been detected and shown to possess characteristic tandem MS spectra and were assigned by comparison to reference standards.

Full text of DOI:10.1021/jf904537z

J. Agric. Food Chem. 2010, 58, 5471–5484 5471
DOI:10.1021/jf904537z
Profiling and Characterization by LC-MSn of the Chlorogenic
´
Acids and Hydroxycinnamoylshikimate Esters in Mate
(Ilex paraguariensis)†
R
AKESH
J
AISWAL, TINA
SOVDAT, FRANCESCO
V
IVAN
,
AND
N
IKOLAI
K
UHNERT
*
School of Engineering and Science, Chemistry, Jacobs University Bremen, 28759 Bremen, Germany
n
´
The chlorogenic acids of mate (Ilex paraguariensis) have been investigated qualitatively by LC-MS .
Forty-two chlorogenic acids were detected and all characterized to regioisomeric level on the basis
of their fragmentation pattern in tandem MS spectra, 24 of them for the first time from this source.
Both chlorogenic acids based on trans- and cis-cinnamic acid substituents were identified. Assign-
ment to the level of individual regioisomers was possible for eight caffeoylquinic acids (1-8), five
dicaffeoylquinic acids (20-24), six feruloylquinic acids (9-14), two diferuloyl quinic acids (25 and
26), five p-coumaroylquinic acids (15-19), four caffeoyl-p-coumaroylquinic acids (34-37), seven
caffeoyl-feruloylquinic acids (27-33), three caffeoyl-sinapoylquinic acids (38-40), one tricaffeoyl-
quinic acid (41), and one dicaffeoyl-feruloylquinic acid (42). Furthermore, four caffeoylshikimates
(43-46), three dicaffeoylshikimates (47-49), one tricaffeoylshikimate (51), and one feruloylshiki-
mate (50) have been detected and shown to possess characteristic tandem MS spectra and were
assigned by comparison to reference standards.
KEYWORDS: Caffeoylquinic acids; feruloylquinic acids; caffeoylshikimic acids; feruloylshikimic acids;
chlorogenic acids; dicaffeoylquinic acids; diferuloylquinic acid; caffeoyl-feruloylquinic acids; caffeoyl-
´
sinapoylquinic acids; tricaffeoylquinic acids; caffeoyl-epi-quinic acids; Ilex paraguariensis; mate;
LC-MS; tandem MS
INTRODUCTION
the shikimate esters in dietary materials been analyzed by tandem
MS and characterized to regioisomeric level. For galloylshiki-
mates in Quercus robur shikimate esters have been characterized
to regioisomeric level using purification followed by NMR
´
characterization (10). In the case of roasted mate it was suggested
that the shikimate derivatives were produced via the roasting
process leading to loss of water at elevated temperatures (5).
Yerba mate is a popular drink prepared by aqueous infusion of
´
the dried leaves of Ilex paraguariensis indigenous to South
American countries such as Paraguay, Uruguay, Argentina,
and Brazil. According to legend, the goddess of the moon and
the cloud gave the Guarani Indians this “drink of friendship” as a
Chlorogenic acids (CGAs) are a family of esters formed
between quinic acid and certain trans-cinnamic acids, most
commonly caffeic, p-coumaric and ferulic acid (1-3). Represen-
tative structures are shown in Figure 1. Using a more general
definition, all esters of quinic acids and their diastereomers can be
considered as chlorogenic acids. In the IUPAC system, used
throughout this paper, (-)-quinic acid is defined as 1-L-1(OH),-
3,4/5-tetrahydroxycyclohexane carboxylic acid, but Eliel and
Ramirez (4) recommend 1R,3R,4R,5R-tetrahydroxycyclohexane
carboxylic acid.
Structurally related, but much less common, are similar esters
between shikimic acid and caffeic or ferulic acid. Formally,
caffeoylshikimates (CSAs) are closely related to chlorogenic acids
and could be chemically obtained from these through dehydra-
tion by loss of the 1-OH functionality. It is worth noting that in
the literature, similar to quinic acid, a nomenclature issue exists
with respect to the numbering of the ring carbon atoms. In this
paper we use the definition of shikimic acid as (3R,4S,5R)-3,4,5-
trihydroxy-1-cyclohexene carboxylic acid throughout. Cinna-
reward for saving her from a jaguar (11). Mate
infused to produce the desired tea beverage or roasted prior to
infusion. Recently, yerba mate has become increasingly popular
´
leaves are directly
´
in other parts of the world with an estimated 5% of all tea
productssoldworldwidecontainingtheleavesofI. paraguariensis(11).
The commercial importance of this commodity is highlighted by
the fact that the annual mate
U.S. $1 billion, growing at an annual estimated rate of 50 t over
the past decade (11). The consumption of mate has been associated
´
trade is currently worth around
moylshikimates have been previously reported in roasted mate
´
,
´
date palms, sweet basil, and carrots (5-9). On no occasion have
with a range of health benefits connected to their high content of
polyphenolic secondary metabolites (12, 13). Additionally, it is
´
worth noting that the organoleptic properties of mate have been
compared to those of coffee, suggesting that the two popular drinks
share some identical sensory components, most likely chlorogenic
acids.
†This paper is dedicated to Professor W. A. Schenk on the
occasion of his 65th birthday.
*Author to whom correspondence should be addressed (telephone
49 421 200 3120; fax 49 421 200 3229; e-mail n.kuhnert@jacobs-
university.de).
© 2010 American Chemical Society
Published on Web 04/20/2010
pubs.acs.org/JAFC

5472 J. Agric. Food Chem., Vol. 58, No. 9, 2010
Jaiswal et al.
Figure 1. Continued

Article
J. Agric. Food Chem., Vol. 58, No. 9, 2010 5473
Figure 1. Continued

5474 J. Agric. Food Chem., Vol. 58, No. 9, 2010
Jaiswal et al.
Figure 1. Structures, numbering, nomenclature, substituents, and chromatographic retention times for CGAs. *, acylation position is uncertain.
Studies investigating the biological activities of mate -based
´
For example, all four regioisomers of caffeoylquinic acids or
six regioisomers of dicaffeoylquinic acid show distinct MSn
spectra that can be rationalized by taking into account the
unique hydrogen-bonding arrays found in each individual
regioisomer. We believe hydrogen bonding to be the key
stereochemical factor determining the fragmentation pattern
of chlorogenic acids inducing one particular fragmentation
pathway (19-26). It is worth noting that this tandem MS
method represents one of the first examples whereby mass
spectrometry allows a reliable and predictive assignment of
isomeric compounds. If tandem MS spectra do not allow
unambiguous structure information, additional parameters
such as retention time, order of elution, and UV spectra can
assist in the structure assignment. This investigation applies
these methods to the qualitative profiling and characterization
of chlorogenic acids in I. paraguriensis.
drinks identified saponins, purine alkaloids, and polyphenols as
possible plant constituents responsible for health benefits (14-18).
Chlorogenic acids, a subgroup of polyphenols, have been reported
in mate on several occasions, with a total of 18 esters identified and
´
reported so far. However, no attempt was made to characterize
caffeoyl and feruloyl quinic acids to individual regioisomeric level
using NMR or tandem mass spectrometry (5, 17).
Plants that synthesize chlorogenic acids commonly produce
many related compounds (over 45 in the coffee bean, for
example) and by using LC-MSn it has been possible to discrimi-
nate and carry out unambiguous structure elucidation between
individual regioisomers of monoacyl and diacyl chlorogenic
acids, without the need to isolate the pure compounds (19-22).
This identification method is based on the fragmentation
pattern observed in tandem mass spectra of chlorogenic acids.

Article
J. Agric. Food Chem., Vol. 58, No. 9, 2010 5475
MATERIALS AND METHODS
vacuo, and the gummy residue was dissolved in ethyl acetate (10 mL). This
mixture was washed with water (20 mL), and water was back-extracted
with ethyl acetate (2 ꢀ 15 mL). The organic layers were dried over sodium
sulfate, and solvents were removed in vacuo. The resulting pale yellow,
semisolid residue was purified by silica gel column chromatography
(EtOAc/petroleum ether, 1:2), which yielded methyl-3,4-isopropylidene-
shikimate 55 as white crystals (515 mg, 2.25 mmol, 85%): 1H NMR (400
MHz, CDCl3) δ 6.91 (1H, t, J = 4.0, 2.6 Hz, -CdCH), 4.73 (1H, t, J =
9.6, 9.4 Hz, -C;H), 4.10 (1H, t, J = 13.7 Hz, 6.4, -C;H), 3.90 (1H, m,
-C;H), 3.76 (3H, s, -(CO)OCH3), 2.79 (1H, dd, J = 17.4, 4.6 Hz, H-
6eq), 2.55 (1H, s, OH), 2.23 (1H, ddt, J = 17.4, 8.2, 2.3 Hz, H-6 ax), 1.43
(3H, s, Me), 1.39 (3H, s, Me); 13C NMR (100 MHz, CDCl3) δ 171.0, 133.7,
129.1, 111.1, 71.8, 69.3, 51.7, 30.9, 29.2, 26.2, 21.1, 21.1.
All of the chemicals (analytical grade) were purchased from Sigma-
Aldrich (Bremen, Germany). Green dry yerba mate
were purchased from two supermarkets in Bremen, Germany.
Sample Preparation. Green dry yerba mate leaves (5 g) were roasted
in an oven (110 °C for 5 h), extracted with methanol (100 mL), homo-
genized with a blender, and ultrasonicated for 5 min. This extract was
filtered through a Whatman no. 1 filter paper. The methanol was removed
by vacuum evaporation, and the extract was stored at -20 °C. When
required, it was thawed at room temperature, dissolved in methanol
(120 mg/10 mL of methanol), filtered through a membrane filter, and
used directly for LC-MS. Similar extraction and sample preparation
´
leaves (Argentina)
´
methods have been used for the green dry yerba mate
roasting.
UV Irradiation. The prepared samples of the green and roasted mate
´
leaves without
Synthesis of 5-(3,4-Diacetyl)-3,4-isopropylidene-caffeoylshikimic Acid
(56). To a solution of 3,4-isopropylidene shikimic acid 55 (100 mg, 0.46
mmol) and DMAP (5 mg, 0.04 mmol) in CH2Cl2 (15 mL) were added
pyridine (4 mL) and 3,4-diacetylcaffeic acid chloride 54 (129 mg, 0.46
mmol) at room temperature. The reaction mixture was stirred for 12 h and
acidified with 1 M HCl (pH ≈3). The layers were separated, and the
aqueous phase was re-extracted with CH2Cl2 (2 ꢀ 15 mL). The combined
organic layers were dried over Na2SO4 and filtered, and the solvents were
dried in vacuo. The crude product was purified by column chromatogra-
phy (petroleum ether/EtOAc 3:1) to give the title compound 56 as a yellow
´
(1 mL of each) were placed in a photoreactor (Luzchem LZC-4 V, Ottawa,
Canada) under a shortwave UV lamp and irradiated at 245 nm for 40 min.
LC-MSn. The LC equipment (Agilent 1100 series, Bremen, Germany)
comprised a binary pump, an autosampler with a 100 μL loop, and a diode
array detector with a light-pipe flow cell (recording at 320 and 254 nm and
scanning from 200 to 600 nm). This was interfaced with an ion-trap mass
spectrometer fitted with an ESI source (Bruker Daltonics HCT Ultra,
Bremen, Germany) operating in Auto-MSn mode to obtain fragment ion
m/z. As necessary, MS2, MS3, and MS4 fragment-targeted experiments
were performed to focus only on compounds producing a parent ion at m/z
335.1, 337.1, 349.1, 353.1, 367.1, 497.2, 499.2, 515.2, 529.2, 543.2, 559.2,
659.3, or 677.3. Tandem mass spectra were acquired in Auto-MSn mode
(smart fragmentation) using a ramping of the collision energy. Maximum
fragmentation amplitude was set to 1 V, starting at 30% and ending at
200%. MS operating conditions (negative mode) had been optimized
using 5-caffeoylquinic acid (3) with a capillary temperature of 365 °C, a dry
gas flow rate of 10 L/min, and a nebulizer pressure of 10 psi.
High-resolution LC-MS was carried out using the same HPLC
equipped with a MicrOTOF Focus mass spectrometer (Bruker Daltonics)
fitted with an ESI source, and internal calibration was achieved with 10mL
of 0.1 M sodium formate solution injected through a six-port valve prior to
each chromatographic run. Calibration was carried out using the en-
hanced quadratic calibration mode. It should be noted that in TOF
calibration the intensities of the measured peaks have a significant
influence on the magnitude of the mass error with high-intensity peaks
resulting in detector saturation displaying larger mass errors. Usually this
problem can be overcome by using spectra averaging on the side flanks of
a chromatographic peak orusing a moredilute sampleas carried out in this
case. All MS measurements were carried out in the negative ion mode.
HPLC. Separation was achieved on a 150 ꢀ 3 mm i.d. column con-
taining diphenyl 5 μm with a 4 ꢀ 3 mm i.d. guard column of the same
material (Varian, Darmstadt, Germany). Solvent A was water/formic acid
(1000:0.05 v/v), and solvent B was methanol. Solvents were delivered at a
total flow rate of 500 μL/min. The gradient profile was from 10% B to
70% B linearly in 60 min followed by 10 min isocratic and a return to 10%
B at 80 min and 10 min isocratic to re-equilibrate.
1
oil (33.3%): H NMR (400 MHz, CDCl3) δ 7.58 (1H, d, J = 15.8 Hz,
-CdCH), 7.36 (1H, dd, J = 7.5, 1.8 Hz, Ar-H), 7.32 (1H, d, J = 1.6 Hz,
Ar-H), 6.90 (1H, t, J 2.5, 1.4, Ar-H), 6.34 (1H, d, J 16.1, -CdCH), 5.26
(1H, dd, J 11.0, 6.19, -C;H), 4.74 (1H, m, -C;H), 4.28 (1H, t, J = 6.19
Hz, -C;H), 3.75 (3H, s, -(CO)OCH3), 2.82 (1H, dd, J = 17.8, 4.5 Hz,
-C;H), 2.41 (1H, dd, J = 17.5, 6.2 Hz, -C;H), 2.3 (6H, s, -(CO)CH3),
1.39 (3H, s, -CH3), 1.37 (3H, s, -CH3); 13C NMR (100 MHz, CDCl3) δ
168.1, 168.1, 166.5, 165.8, 143.6, 142.5, 134.3, 133.0, 129.5, 126.3, 123.9,
122.7, 118.9, 110.2, 74.0, 71.9, 70.1, 52.2, 27.9, 26.5, 25.9, 20.7, 20.7.
For the deprotection of 55, a suspension of the methyl-3,4-isopropy-
lideneshikimate ester 56 (515 mg, 2.25 mmol) in KOH solution (10 mL,
1M) was stirred at room temperature for 30 min, and the reaction mixture
was neutralized by the addition of acidic ion-exchange resin (Amberlite IR
400, 2 g). The aqueous layer was extracted with ethyl acetate (2 ꢀ 20 mL),
combined organic layers were dried over sodium sulfate, and solvent was
evaporated in vacuo. The resulting ester was dissolved in a mixture of
20 mL of trifluoroacetic acid and water (7:3) at room temperature and
stirred for 30 min. The solvents were removed in vacuo, and the resulting
yellowish crude product that was analyzed by 1H NMR and HPLC-MS
and showed a single caffeoyl shikimic acid derivative, 52, at m/z 335.1.
Synthesis of Bisacetal Protected Shikimic Acid 57. To a suspension of
shikimic acid (500 mg, 2.87 mmol) in methanol (10 mL) were added 2,3-
butanedione (0.48 mL, 5.5 mmol), trimethylorthoformate (1.1 mL,
10 mmol), and D-camphorsulfonic acid (50 mg, 0.01 mmol). The mixture
was reflux for 16 h, then cooled to room temperature, and treated with
sodium bicarbonate (80 mg, 0.80 mmol). The solvent was removed in
vacuo to give a paste that was dissolved in ethyl acetate. Activatedcharcoal
(2 g) was added, and the mixture was refluxed for 2 h and then left to cool
to room temperature. The mixture was filtered over a thick pad of silica
gel, which was further washed using ethyl acetate/methanol (9:1), and the
resulting colorless filtrate was evaporated in vacuo to give a white solid.
The crude was recrystallized from ethyl acetate to form the white shiny
needles of bisacetal-protected methyl shikimate 57 (693 mg, 2.29 mmol,
80%): 1H NMR (400 MHz, CDCl3) δ 6.86 (1H, dd, J = 5.5, 2.7 Hz, H-2),
4.34, (1H, t, J = 4.6 Hz, H-4), 4.07 (1H, ddd, J = 10.4, 5.9, 5.5 Hz, H-5),
3.72 (3H, s, OMe), 3.57 (1H, dd, J = 10.5, 4.6 Hz, H-4), 3.28 (3H, s, OMe),
3.24 (3H, s, OMe), 2.84 (1H, dd, J = 17.4, 5.9 Hz, H-6eq), 2.25 (1H, ddd,
J = 17.4, 10.4, 2.7 Hz, H-6ax), 1.38 (3H, s, Me), 1.31 (3H, s, Me); 13C
NMR (100 MHz, CDCl3) δ 166.6 (CdO), 135.1 (CdC), 132.2 (CdC),
100.1 (OCOMe), 99.3 (OCOMe), 70.6 (C3), 65.1 (C;O), 62.5 (C;O),
52.3 (COOMe), 48.1 (OMe), 48.0 (OMe), 30.1 (C6), 17.9 (Me), 17.7 (Me).
(a) Synthesis of Bisacetal-Protected 3-(3,4-Diacetyl-caffeoyl)-
5-shikimic Acid Methyl Ester 58. To a solution of bisacetal-protected
shikimic acid 57 (132 mg, 0.46 mmol) and DMAP (5 mg, 0.04 mmol) in
CH2Cl2 (15 mL) were added pyridine (4 mL) and 3,4-diacetylcaffeic acid
chloride 54 (129 mg, 0.46 mmol) at room temperature. The reaction
mixture was stirred for 12 h and acidified with 1 M HCl (pH ≈3). The
layers were separated, and the aqueous phase was re-extracted with
Synthesis of Mixture of Regioisomers of Caffeoylshikimic Acid (44, 45,
and 52). To a solution of shikimic acid (80 mg, 0.46 mmol) and DMAP
(5 mg, 0.04 mmol) in CH2Cl2 (10 mL) were added pyridine (3 mL) and 3,4-
diacetylcaffeic acid chloride (129 mg, 0.46 mmol) at room temperature.
The reaction mixture was stirred for 12 h and acidified with 1 M HCl
(pH ≈3). The layers were separated, and the aqueous phase was re-
extracted with CH2Cl2 (2 ꢀ 15 mL). The combined organic layers were
dried over Na2SO4 and filtered, and the solvents were dried in vacuo. The
resulting ester was dissolved in a mixture of 20 mL of trifluoroacetic acid
and water (7:3) at room temperature and stirred for 30 min. The solvents
were removed in vacuo, and the resulting yellowish product was analyzed
by HPLC-MS (see conditions above) and shown to contain a mixture of
45, 44, and 52.
Synthesis of 3,4-Isopropylidene Shikimic Acid. A mixture of methyl
shikimate 53 (500 mg, 2.66 mmol), p-toluenesulfonic acid monohydrate
(5 mg, 0.029 mmol), 2,2-dimethoxypropane (2 g, 19.2 mmol), and acetone
(10 g) was heated at reflux for 4 h. The reaction was cooled to room
temperature and neutralized by the addition of sodium methoxide (16 mg,
0.30 mmol) in methanol (2 mL). Most of the solvent was removed in

5476 J. Agric. Food Chem., Vol. 58, No. 9, 2010
Jaiswal et al.
´
Figure 2. TICs of roasted mate extract: (A) full view; (B, inset) using ion trap MS in negative ion mode.
CH2Cl2 (2 ꢀ 15 mL). The combined organic layers were dried over
Na2SO4 and filtered, the solvents were dried in vacuo, and the crude
product was purified by column chromatography (petroleum ether/EtOAc
3:1) to give the title product 58 as a yellow oil (37.2%): 1H NMR (400
MHz, CDCl3) δ 6.86 (1H, dd, J = 5.5, 2.7 Hz, H-2), 4.34, (1H, t, J = 4.6
Hz, H-4), 4.07 (1H, ddd, J = 10.4, 5.9, 5.5 Hz, H-5), 3.72 (3H, s, OMe),
3.57 (1H, dd, J = 10.5, 4.6 Hz, H-4), 3.28 (3H, s, OMe), 3.24 (3H, s, OMe),
2.84 (1H, dd, J = 17.4, 5.9 Hz, H-6eq), 2.25 (1H, ddd, J = 17.4, 10.4, 2.7
Hz, H-6ax), 1.38 (3H, s, Me), 1.31 (3H, s, Me); 13C NMR (100 MHz,
CDCl3) δ 171.2, 168.0, 166.4, 165.9, 143.6, 143.5, 142.5, 133.5, 133.4,
131.8, 126.6, 123.9, 122.7, 119.1, 99.9, 99.3, 68.9, 66.6, 63.1, 60.4, 52.1, 48.1,
29.7, 22.7, 21.1, 20.7, 14.3.
Preliminary Assessment of Data. All data for CGA presented
use the recommended IUPAC numbering system (1), and speci-
men structures are presented in Figure 1. When necessary,
previously published data have been amended to ensure consis-
´
tency and avoid ambiguity. The roasted mate extract (sample I)
and dry green leaf mate extract (sample II) were examined at least
´
on five separate occasions. All other samples III-V were mea-
sured in duplicate. No appreciable differences in the LC-MS
chromatograms between the different samples or different chro-
matographic runs were observed independent of whether they
were obtained from dried leaves or roasted material. Addition-
ally, specimen total ion chromatograms (TICs) for a representa-
(b) Deprotection of 58. A suspension of the methyl-3,4-isopropy-
lideneshikimate ester 56 (515 mg, 2.25 mmol) in KOH solution (10 mL,
1M) was stirred at room temperature for 30 min, and the reaction mixture
was neutralized by the addition of acidic ion-exchange resin (Amberlite IR
400, 2 g). The aqueous layer was extracted with ethyl acetate (2 ꢀ 20 mL),
the combined organic layers were dried over sodium sulfate, and solvent
was evaporated in vacuo. The resulting ester was dissolved in a mixture of
20 mL of trifluoroacetic acid and water (7:3) at room temperature and
stirred for 30 min. The solvents were removed in vacuo, and the resulting
yellowish crude product that was analyzed by 1H NMR and HPLC-MS
(high-resolution TOF and tandem MS) showed a single caffeoyl shikimic
acid derivative 45 at m/z 335.1
tive mate extract are shown in Figure 2.
´
For the chromatographic analysis used in this contribution the
HPLC method used was slightly modified from that reported
previously (19). The phenylhexyl packing was replaced by a
diphenyl packing, and acetonitrile was replaced by methanol as
an eluent, resulting overall in a more economical method with
slightly improved resolution.
When commercial standards were not available, peak identities
were assigned on the basis of the structure-diagnostic hierarchical
keys previously developed, supported by means of their parent
ion, UV spectra, and sequence ofelution/retention timerelativeto
5-CQA using methods validated in our laboratory (23-25).
RESULTS AND DISCUSSION
In a first set of experiments the mate extracts were analyzed by
´
A total of five different mate samples (samples I-V) were
´
high-resolution mass spectrometry using an ESI-TOF instrument
in the negative ion mode. Following the strategy employed in
previous studies (2, 3, 23), selected ion monitoring (SIM) located
eight CQAs (1-8), six FQAs (9-14), five p-CoQAs (15-19), five
diCQAs (20-24), two diFQAs (25 and 26), seven CFQAs
(27-33), four CSAs (43-46), one FSA (50), three diCSAs
(47-49), three CSiQAs (38-40), four p-CoCQAs (34-37), one
diCFQA (42), one triCQA (41), and one triCSA (51). IUPAC
names, abbreviations, numbering, and retention times are given in
Figure 1 including unassigned derivatives. All structures are
analyzed, three dried leaves samples (II-IV) and two roasted
samples (I and V). One roasted sample was obtained as a
commercial product (I), and the second sample was roasted under
defined conditions(V) witha roasting timeof 5 h at 110 °C, longer
than that typically used in commercial products, to study whether
dehydration of CGA derivatives to shikimates takes place. Mate
´
leaves dried or roasted were extracted using methanol, and
the extract ws directly used for LC-MS analysis. In comparison
to isolation of CGAs from green coffee beans, no removal of
proteins and peptides using Carrez reagent was required (21, 22).

Article
J. Agric. Food Chem., Vol. 58, No. 9, 2010 5477
Table 1. Negative Ion MS3 Data for Monoacylchlorogenic Acids
MS1
MS2
secondary peak
MS3
parent ion
base peak
base peak
secondary peak
no.
compd
3-CQA
4-CQA
m/z
m/z
inta
m/z
int
m/z
int
m/z
m/z
int
m/z
int
67
1
353.1
353.1
353.2
353.1
353.1
353.2
352.9
352.9
367.2
367.2
367.2
367.2
367.2
367.2
337.1
337.1
337.2
337.1
337.2
335.2
335.1
335.1
335.2
349.1
190.9
172.9
190.0
190.9
172.9
190.0
190.7
178.7
192.9
172.9
190.9
192.9
172.9
190.9
162.9
172.7
190.9
162.9
190.9
160.9
178.9
178.9
160.9
174.7
178.5
178.9
178.5
178.5
178.9
178.5
178.9
190.7
191.5
192.9
172.9
191.5
192.9
172.9
190.0
50
60
5
134.9
135.0
135.0
134.9
135.0
135.0
135.0
134.8
7
9
85.3
93.2
127.0
111.0
126.9
127.0
111.0
126.9
126.7
71
48
66
71
67
70
28
172.9
2
190.8
190.8
20
3
5-CQA
15
12
9
85.2
85.3
172.9
172.9
27
45
4
cis-3-CQA
cis-4-CQA
cis-5-CQA
CeQA
45
74
28
49
51
2
5
90
93.2
85.2
6
13
10
20
172.9
170.7
64
46
7
172.7
160.7
173.2
99
10
2
154.8
134.8
133.9
93.1
8
CeQA
3-FQA
9
148.9
111.5
126.9
148.9
111.5
126.9
23
44
70
24
59
52
10
11
12
13
14
15
16
17
18
19
43
44
45
46
50
4-FQA
5-FQA
16
2
85.2
cis-3-FQA
cis-4-FQA
cis-5-FQA
3-p-CoQA
4-p-CoQA
5-p-CoQA
cis-3-p-CoQA
cis-5-p-CoQA
CSA
2
173.2
2
133.9
93.1
9
50
5
85.2
118.9
93.0
111.0
61
162.9
190.0
162.9
178.9
160.9
160.9
178.9
192.7
5
12
8
85.2
118.9
85.2
12
82
80
12
41
134.9
134.9
135.0
135.0
133.7
50
50
25
18
27
132.9
134.9
134.9
132.9
159.7
4-CSA
3-CSA
CSA
4-FSA
148.7
78
145.8
5
110.9
4
a Intensity.
given in Figure 1 excluding unassigned derivatives. For all
compounds the high-resolution mass data were in good
agreement with the theoretical molecular formulas, all dis-
playing a mass error of below 5 ppm, confirming their ele-
mental composition.
and three of them were assigned using the hierarchical keys
previously developed (19-22) as the well-known 3-O-caffeoyl-
quinic acid (1), 4-O-caffeoylquinic acid (2), and 5-O-caffeoylqui-
nic acid (3) (Table 1). A further three peaks present as minor
components displayed fragmentation patterns identical to those
of 3-, 4-, and 5-caffeoylquinic acid, respectively, and we suspected
that they might be cis isomers of the corresponding caffeoylquinic
acids. For confirmation of these isomers, extracts of roasted and
In a second set of experiments the mate extracts were subjected
´
to LC-MSn measurements in the negative ion mode using an ESI-
ion trap mass spectrometer, allowing assignments of compounds
to the regioisomeric level. Selected ion monitoring was carried out
for all CGAs of known molecular weight already reported in the
literature. Unknown compounds were readily identified by their
characteristic fragment peaks at m/z 173 and 191 in an all MSn
extracted ion chromatogram searching for these fragments.
Table 1 contains the summarized MSn data for the monoacyl
CGAs and monoacyl shikimates, Table 2 the equivalent data for
diacyl CGAs, and Table 3 the equivalent data for the triacyl
CGA.
green mate were irradiated with UV light at 245 nm for 40 min.
´
After irradiation, we observed all three putative cis isomers in the
chromatogram as peaks with significantly increased intensities if
compared to their corresponding trans isomers from the original
plant extract (Figure 3), which confirmed the presence of the three
cis-caffeoylquinic acids (4-6).
We reported the presence of cis derivatives of chlorogenic acids
earlier in coffee leaves (26). It appears as if chlorogenic acids present
in plant tissue exposed to natural UV light undergo trans-cis
isomerization, whereas in tissue unexposed to UV light, such as the
seeds of coffee berries, they remain unchanged. Whether this
trans-cis isomerization has any physiological significance such as
protection from UV light, sensing of UV exposure, or even
harvesting of UV photons must at this stage remain subject to
speculation and requires further detailed investigations.
The remaining two caffeoylquinic acids (7 and 8) were tenta-
tively assigned as diastereomers of CGAs, caffeoyl-epi-quinic
acids, because they have different retention times and signifi-
cantly different fragmentation patterns if compared to 1-, 3-, 4-,
or 5-caffeoylquinic acid (Table 1). It is worth pointing out that
diastereomers of chlorogenic acid display different tandem MS
spectra if compared to the parent compound, allowing potential
for a tandem MS based assignment of stereoisomers.
Similarly, five dicaffeoylquinic acid isomers were identified by
their m/z 515 parent ion and assigned as 1,3-di-O-caffeoylquinic
acid(20), 3,4-di-O-caffeoylquinic acid (21), and 4,5-di-O-caffeoyl-
quinic acid (22), according to their characteristic fragmentation
In general, peak identities were consistent both within and
between analyses. However, when the mass spectrum for a
particular substance included two ions of similar mean intensities,
within-analysis experimental error dictated that in some indivi-
dual MS scans one would be more intense and for other scans
the reverse would be true. This phenomenon was encountered
primarily when the signal intensity was lower, that is, with
quantitatively minor components and/or higher order spectra.
For example, the monoacyl CGA MS3 ions at m/z 85.3 and at m/z
126.9 are essentially coequal in some spectra. However, in this
particular case the lower mass ion has been assigned consistently
as the base peak because in the spectra of several compounds this
was clearly the case. Fragment ions with intensities of <10% of
the base peak have been reported only when they are needed for
comparison.
Characterization of Caffeoylquinic Acids (Mr 354), Dicaffeoyl-
quinic Acids (Mr 516), and Tricaffeoylquinic Acid (Mr 678). Eight
caffeoylquinic acids (1-8) were located in the extract (Figure 2),

5478 J. Agric. Food Chem., Vol. 58, No. 9, 2010
Table 2. Negative Ion MS4 Data for Diacyl Chlorogenic Acids
Jaiswal et al.
MS1
MS2
MS3
MS4
parent ion base peak
secondary peak
base peak
secondary peak
base peak
secondary peak
no.
compd
m/z
m/z inta m/z int m/z int
m/z
m/z int m/z int m/z int
m/z
m/z int m/z int
20 1,3-diCQA
21 3,4-diCQA
22 4,5-diCQA
23 A cis-4,5-diCQA
24 A cis-4,5-diCQA
25 1,4-diFQA
26 4,5-diFQA
27 C,FQA
515.2
515.2
515.2
515.3
515.4
543.2
543.2
529.1
529.1
529.2
529.1
529.2
529.1
529.1
499.0
499.1
499.2
499.3
559.2
559.2
559.2
497.1
497.1
497.1
353.1
353.1
353.1
353.1
353.1
349.1
367.1
367.1
367.1
367.1
367.1
367.1
353.1
367.1
353.1
337.1
337.1
353.0
397.1
397.1
397.1
335.1
335.1
335.1
335.1
335.1
335.1
335.1
335.1
2
4
2
5
4
173.0
172.9 20
172.9
172.9 10
172.9
4
190.9
172.9
172.9
172.9
172.9
192.9
172.9
160.8
192.7
192.7
172.9
172.9
172.9
192.7
190.7
172.9
172.9
172.9
172.9
222.9
172.9
160.8
172.9
178.7
179.0 60
178.9 68 191.0 32 135.1
178.9 76 190.9
178.9 60 190.9 20 135.0
135.1
6
9
85.1
93.2
93.1
93.1
93.1
111.1 86 172.9 60
111.1 30
111.0 20
6
9
135.0 19
8
111.0 36
111.0 22
9
178.9 76 190.9
172.9 31 268.8
9
9
135.0 21
133.8 22
367.1 25 172.9 17 268.8
349.1 35
9
178.9 60 190.8 20 135.0
9
93.1
132.7
133.7
133.7
93.2
111.1 40
192.7 14
134.8 18
133.8 18
133.8 36
28 1C,3FQA
29 3F,5CQA
353.1 15
353.1 60 349.0 32 335.0 32
172.6 13 178.6
172.6 36
192.9 62
5
149.0 16 127.0
149.0 19
111.1 22 127.0 14
6
30 A cis-4F,5CQA
31 4F,5CQA
32 4C,5FQA
335.0
335.0
3
4
172.7 14
172.7 21
133.8
133.8
7
8
192.9 71
93.2
93.2
367.1 25
334.8
337.0 15
178.9 49 190.8 35 134.7 10
172.6 52 133.8 15
127.0 nd
149.0 22
33 A cis-3F,5CQA
34 3C,5-p-CoQA
35 4-p-Co,5CQA
36 A cis-4-p-Co,5CQA
37 4C,5-p-CoQA
38 3C,4SiQA
4
133.7
85.2
93.0 70 126.9 99
111.1 98
111.1 90
335.1
335.1
3
3
172.7 59
172.7 30
172.8 15
162.6 8
162.6 21
93.2
93.2
178.7 66 190.6 29 134.8 12
222.9 76
172.9 31
93.2
93.1
490.7 nd 233.0 10 172.9 14
490.7 16 222.9 11 172.9
490.8 23 222.9 nd 172.9 14
111.0 81 71.4 25
192.8 54 148.8 31
39 3Si,5CQA
40 4Si,5CQA
5
163.8
71.4
222.9
178.9 12
5
47 3,4-diCSA
48 3,4-diCSA
49 3,5-diCSA
135.0 30
133.0
111.0
134.7
154.8 nd
154.8 45
178.9 75 160.9 47 135.0 47
160.6 25 134.8 58
449.1 95
a Intensity.
Table 3. Negative Ion MS4 Data for Triacyl Chlorogenic Acids
MS1
MS2
MS3
MS4
parent ion base peak
secondary peak
m/z inta m/z int
base peak
secondary peak
base peak
secondary peak
m/z int m/z int
no.
compd
m/z
m/z
m/z int m/z int m/z int
m/z
41 3,4,5-triCQA
42 3,4-diC, 5-FQA
51 3,4,5-triCSA
677.3
691.3
659.3
515.1
529.2
497.1
497.3
353.1 12
453.2 25 211.0 13
4
353.1 22
353.1
353.1
335.0
335.1 14 190.8 14 172.9 27
367.2 47 335.1 36 192.9 48
317.0 99
172.9
172.8
178.8
178.8 73 190.8 59
178.8 73 191.0 48
a Intensity.
pattern as discussed previously (20). Two other isomers were
assigned as A cis-4,5-di-O-caffeoylquinic acid (23 and 24) on the
basis of their fragmentation pattern and increased intensity after UV
irradiation (Figure 3). In these isomers only one caffeoyl part has cis
geometry (A cis), whereby distinction between 4-cis,5-trans-diCQA
and 4-trans,5-cis-diCQA was not possible.
Characterization of Feruloylquinic Acids (Mr 368) and p-Cou-
maroylquinic Acids (Mr 338). A targeted MSn experiment at m/z
337 applied to the sample located five minor components, and
three of them were identified by their fragmentation (19, 22) as
the well-characterized 3-O-p-coumaroylquinic acid (15), 4-O-p-
coumaroylquinic acid (16), and 5-O-p-coumaroylquinic acid (17)
(Table 1). Another two isomers were identified as the cis-3-O-p-
coumaroylquinic acid (18) and cis-5-O-p-coumaroylquinic acid
(19) on the basis of their fragmentation pattern (identical to
corresponding trans isomers) and comparison with the UV-
irradiated samples showing increased concentration of the cis
isomers (Figure 3).
The analogous experiment at m/z 367 located 3-O-feruloylqui-
nic acid (9), 5-O-feruloylquinic acid (11), 4-O-feruloylquinic acid
(10), cis-3-O-feruloylquinic acid (12), cis-5-O-feruloylquinic acid
(14), and cis-4-O-feruloylquinic acid (13) (Table 1 and Figure 3).
One p-coumaroylquinic acid was recently reported in mate with-
´
out any further structure elucidation and assignment of stereo-
chemistry (5, 33).
A search for tricaffeoylquinic acids as previously reported in
´
various Asteraceae plants (27-32) and mate (5) at m/z 677 in the
extract resulted in the identification of one chromatographic
peak. The pþre-sence of the MS2 base peak at m/z 515 ([M -
caffeoyl - H ] ) and MS3 base peak at m/z 353 ([M - caffeoyl -
caffeoyl - Hþ]-) with a secondary peak at m/z 335.1 (14% of
base peak) and supported by MS4 base peak at m/z 172.9 and a
secondary ion at m/z 178.8 (73% of base peak) clearly suggests a
3,4-disubstitution pattern (Figure 4). A comparison of retention
time and fragmentation pattern with those of 3,4,5-tri-O-caf-
feoylquinic acid present in green coffee extract (unpublished
results) confirms the presence of 3,4,5-tri-O-caffeoylquinic acid
(41) rather than the alternative 1,3,4-substitution, which is also
consistent with the fragmentation data.
Characterization of Caffeoyl-feruloylquinic Acids (Mr 530).
Seven caffeoyl-feruloylquinic acid isomers were readily identified
by their m/z 529 parent ion, and five of them were assigned as
3-O-feruloyl-5-O-caffeoylquinic acid (29), A cis-3-O-caffeoyl-
5-O-feruloylquinic acid (33), 4-O-feruloyl-5-O-caffeoylquinic
acid (30), A cis-4-O-feruloyl-5-O-caffeoylquinic acid (31), and
Prior to this investigation, only 1-, 3-, 4-, and 5-caffeoylquinic
acid (5,17,33), one unassigned caffeoylquinic acid, 1,5-, 3,4-, 3,5-,
and 4,5-dicaffeoylquinic acid, one unassigned dicaffeoylquinic
acid, and one unassigned tricaffeoylquinic acid have been re-
ported in mate (5, 33).
´

Article
J. Agric. Food Chem., Vol. 58, No. 9, 2010 5479
Figure 3. Extracted ion chromatograms (EICs) in negative ion mode before and after UV treatment: (A) at m/z 353 for caffeoylquinic acids (1-6); (B) at
m/z515fordicaffeoylquinicacids(20-24);(C) atm/z 337for p-coumaroylquinicacids(15-19);(D) at m/z 367for feruloylquinicacids(9-14);(E) at m/z529
for caffeoyl-feruloylquinic acids (27-33).
Figure 4. MS2, MS3, and MS4 spectra of 3,4,5-tri-O-caffeoylquinic acid (41) (parent ion at m/z 677 in negative ion mode).

5480 J. Agric. Food Chem., Vol. 58, No. 9, 2010
Jaiswal et al.
Figure 5. MS2, MS3, and MS4 spectra of caffeoyl-feruloylquinic acid (27) (parent ion at m/z 529 in negative ion mode).
Figure 6. MS2 MS3, and MS4 spectrainnegative ion mode:(A) 1,4-diferuloylquinicacid(25) (parent ion at m/z 543); (B) 3,5,-di-O-caffeoyl-4-O-feruloylquinic
acid (42) (parent ion at m/z 691); (C) 3-O-caffeoyl-4-O-sinapoylquinic acid (38) (parent ion at m/z 559); (D) 3-O-sinapoyl-5-O-caffeoylquinic acid (39) (parent
ion at m/z 559); (E) 4-O-sinapoyl-5-O-caffeoylquinic acid (40) (parent ion at m/z 559).
4-O-caffeoyl-5-O-feruloylquinic acid (32) on the basis of their
characteristic fragmentation pattern in the MS2 and MS3 spectra
as described previously (19) and comparison with the UV-
irradiated samples showing an increased concentration of the
cis isomers (Figure 3).
overall data it appears as if I. paraguariensis does not pro-
duce quinic acid esters substituted in the 1-position, similar to
coffee.
A further caffeoyl-feruloylquinic acid (27) was tentatively
assigned as an epimer of caffeoyl-feruloylquinic acid because it
has a different fragmentation pattern (Table 2 and Figure 5) if
compared to MS2 and MS3 spectra of all the possible reported
isomers of caffeoylquinic acid and feruloylquinic acids (19).
Characterization of Diferuloylquinic Acids (Mr 544). Two
diferuloylquinic acid isomers were positively identified by their
parent ion at m/z 543, and one of them was assigned as 4,5-di-O-
feruloylquinic acid (26) on the basis of their characteristic
fragmentation pattern in the MS2 and MS3 spectra as described
previously (19). The presence of the MS2 base peak at m/z 348.9
([feruloylquinic acid - H2O - Hþ]-) supported by secondary
peaks at m/z 367 ([feruloylquinic acid - Hþ]-) (23% of base
peak), m/z 268.8 (9% of base peak) and, m/z 172.9 ([quinic acid -
H2O - Hþ]-) (18% of base peak) defines the 4-substitution
(Table 2 and Figure 6). A comparison to MS2 and MS3 spectra
of 3,4-, 3,5-, and 4,5-diferuloylquinic acids (19) suggested the
The sixth caffeoyl-feruloylquinic acid produced a MS2 base
peak at m/z 367 ([feruloylquinic acid - Hþ]-) by the loss of
caffeoyl residue and a secondary ion at m/z 353 (15% of base
peak), and a MS3 base peak at m/z 193 defines the presence of
feruloyl residue at the 3-position (19). It loses its caffeoyl residue
more preferentially than the feruloyl residue in MS2 spectra,
which suggests the 3-feruloyl-5-caffeoylquinic acid or 1-caffeoyl-
3-feruloylquinic acid, and the lower intensity of the MS3 second-
ary peak at m/z 173 (13%) and MS4 secondary peak at m/z 149
(16%), if compared to 3-feruloyl-5-caffeoylquinic acid, that
favors the assignment as 1-O-caffeoyl-3-O-feruloylquinic acid
(28). It is unclear whether this compound is a genuine mate
´
secondary metabolite or an artifact of the workup procedure
resulting in acyl migration as observed for the known 1,5-diCQA
to 1,3-diCQA (cynarin) isomerization in artichoke. From the

Article
J. Agric. Food Chem., Vol. 58, No. 9, 2010 5481
Figure 7. Synthesis of mixture of regioisomers of caffeoylshikimic acids.
presence of 1,4-substitution and assignment as 1,4-di-O-feruloyl-
quinic acid (25).
All chlorogenic acids described in this contribution have been
located in EIC (extracted ion chromatogram) searching for the
known m/z values corresponding to their parent ions. Unknown
chlorogenic acids can be readily identified and searched for in the
chromatogram by searching in all MSn chromatograms for
signals of fragment ions at m/z 191 and 173 (e.g., in MS3 for
diacyl CGAs and in MS4 for triacyl CGAs), which are character-
istic of esters of quinic acid. By taking this approach, the
following derivatives were located.
Characterization of Putative Dicaffeoyl-feruloylquinic Acid (Mr
692). Targeted MS4 experiments located one parention at m/z 691
([M - Hþ]-) with 320 nm λmax. The presence of a MS2 base peak
at m/z 529 ([M - caffeoyl - Hþ]-), a MS3 base peak at m/z 367
([M - caffeoyl - feruloyl - Hþ]-), and a MS4 base peak at m/z
172.9 with secondary peak at m/z 193 (20% of base peak) (Table 2
and Figure 6) clearly defines the presence of a feruloyl group at
position 4 and caffeoyl groups at positions 3 and 5 (19), and a
comparison of retention time and fragmentation pattern with 3,5-
di-O-caffeoyl-4-O-feruloylquinic acid present in green coffee
extract (unpublished results) confirms the presence of 3,5-di-O-
caffeoyl-4-O-feruloylquinic acid (42).
Figure 8. Synthesis of 5-caffeoylshikimic acid.
Characterization of Putative Caffeoyl-sinapoylquinic Acids (Mr
560). Targeted MS4 experiments located three parent ions at m/z
559 (putative caffeoyl-sinapoylquinic acids) at retention times
ranging from 43 to 54 min. These parent ions had UV spectra of
typical chlorogenic acids. Previously, one caffeoyl-sinapoylquinic
acid isomer has been reported in mate.
´
All three compounds show MS2 base peaks of m/z 397.1-397.2
([SiQA - Hþ]-) and a MS3 base peak of m/z 172.9 ([quinic acid -
H2O - Hþ]-) or m/z 222.9 ([sinapic acid - Hþ]-) (Table 2 and
Figure 6). The presence of a MS3 base peak of m/z 172.9 ([quinic
acid - H2O - Hþ]-) and m/z 222.9 ([sinapic acid - Hþ]-) clearly
defines substitutions at positions 4 and 3, respectively (19). The
presence of a MS3 base peak of m/z 172.9 also shows the presence
of vicinal diacyl groups, either 3,4- or 4,5-diacyl, and the absence
of a MS3 base peak of m/z 172.9 represents 3,5-diacyl groups (19).
MS2 and MS3 patterns and orders of the elution for these putative
caffeoyl-sinapoylquinic acids are similar to caffeoyl-feruloylqui-
nic acids (3C,4FQA, 3F,5CQA, and 4F,5CQA), which strongly
suggest the presence of 3C,4SiQA (38), 3Si,5CQA (39), and
4Si,5CQA (40). It should be noted that we have observed the
presence of a further two regioisomers of sinapoyl-caffeoylquinic
acid in another source, green Robusta coffee beans, assisting in
the assignment (unpublished results).
Figure 9. Synthesis of 3-caffeoylshikimic acid.
acid ester (51) were identified by their parent ions at m/z 335,
497, 349, and 659, respectively, and UV absorption at 320 nm
(λmax).
Further evidence for the assignment of the caffeoylshikimic
acids came from an independent synthesis of a mixture of all three
possible regioisomers of caffeoylshikimic acid (Figure 7). Unfor-
tunately, preparative HPLC separation of the three regioisomers
of caffeoylshikimic acid failed due to insufficient resolution and
unfavorable solubilities. Hence, a selective synthesis of 3-caffeoyl-
shikimic acid (3-CSA) and 5-caffeoylshikimic acid (5-CSA) was
Characterization of Caffeoylshikimic Acids (Mr 336) and Putative
Dicaffeoylshikimic Acids (Mr 498), Feruloylshikimic Acids (Mr 350),
and Tricaffeoylshikimic Acid (Mr 660). Four caffeoylshikimic acid
esters (43-46), three dicaffeoylshikimic acid esters (47-49), one
feruloylshikimic acid ester (50), and one tricaffeoylshikimic
ꢀ
carried out (Figures 8 and 9). For 3-CSA, Leys-diacetal pro-
tection was chosen to selectively protect the trans dihydroxy
functionality, whereas for 5-CSA a di-isopropylidene protecting

5482 J. Agric. Food Chem., Vol. 58, No. 9, 2010
Jaiswal et al.
group was chosen to selectively protect the cis dihydroxy moiety.
Following coupling with an acetate-protected acid chloride of
caffeic acid and final deprotection, both reference compounds
were obtained in reasonable yield and purity.
5-CSA (52) nor a cis-cinnamate derivative of CSA (no identity of
signals at m/z 335 after UV irradiation). Therefore, we tentatively
assign them as diastereoisomers of caffeoylshikimic acid. All
shikimic acid esters were found in both dried leaf mate
roasted mate, clearly indicating that they are not products
obtained from the leaf processing but are genuine mate secondary
metabolites.
´
and
Comparison of the chromatogram of the synthetic mixture of
three regioisomers and the chromatogram of the two standards
allows unambiguous assignment of all three regioisomers by
comparison of retention time and tandem MS spectra (the third
regioisomer 4-CSA follows automatically in this assignment).
The order of elution of caffeoylshikimic acid regioisomers (5 >
4 > 3) was completely different compared to that reported in
the literature (7) (3 > 4 > 5). The HPLC chromatogram of this
synthetic mixture showed two regioisomers (44 and 45) with
retention times and fragmentation patterns identical to the
naturally occurring counterparts (Figures 10 and 11), which are
assigned as 4-CSA (44) and 3-CSA (45). The two additional
´
´
Interestingly, the three regioisomeric shikimic acid esters show
reproducible differences in their MSn spectra, allowing their
identification by tandem MS. Therefore, following our work on
chlorogenic acids a second class of compounds, for which
regiochemistry can be unambiguously identified using tandem
MS, has been identified.
All three regioisomers show in their MS2 spectra a base peak at
m/z 178.9 corresponding to a caffeic acid fragment A (C9H7O4) in
Figure 12. In all MS3 spectra a fragment of the transition m/z
335f178.8 shows a fragment F at m/z 134.9 corresponding to a
decarboxylation of caffeic acid (C8H7O2). 4-CSA (44) shows a
strong fragment ion C at m/z 160.9 (C9H5O3), which is absent for
3-CSA (45) and 5-CSA (52) and allows unambiguous identifica-
tion of4-CSA (44). This observation can be rationalized bytaking
into account the Hammond postulate. Any fragmentation is an
endothermic process characterized by a product-like late transi-
tion state. For 3-CSA (45) and 5-CSA (52) the products of
fragmentation are cyclohexadiene carboxylic acids B and D
appearing as neutral losses with a conjugated diene moiety
(Figure 12). For 4-CSA (44) no stabilization by forming a
conjugated diene is possible, therefore allowing an alternative
competing fragmentation pathway with a p-quino-methine
ketene C as a fragment ion. It is worth noting that in comparison
to quinic acid esters, where fragment spectra were rationalized in
terms of hydrogen bonding and activation of leaving groups
by the 1-COOH group, this pathway is absent in shikimic acid
derivatives due to the stereochemical restriction of the sp2
hybridized carbons.
caffeoylshikimic ester derivatives found in the mate samples are
´
by comparison of retention time and tandem MS spectra neither
Figure 10. Extracted ion chromatogram (EIC) at m/z 335 (negative ion
mode) for synthetic caffeoylshikimic acid regioisomers.
Figure 11. MS2 and MS3 spectra for synthetic 3-, 4-, and 5-caffeoylshikimic acid (parent ion at m/z 335 in negative ion mode).

Article
J. Agric. Food Chem., Vol. 58, No. 9, 2010 5483
Figure 12. Fragmentation pathway of 3-, 4-, and 5-caffeoylshikimic acid.
Figure 13. MS2 and MS3 spectra for 3,4,5-tri-O-caffeoylshikimic acid (51) (parent ion at m/z 659 in negative ion mode).
5-CSA (52) shows in MS2 a fragment ion E at m/z 316.9 (M -
H - H2O, C16H14O7), which is absent in 3-CSA (45), allowing
differentiation between these two regioisomers. This observation
can be rationalized by assuming a 1,2-acyl migration with loss of
water, similar to 4-CQA (44) derivatives, only possible for the
trans stereochemical relationship in the 4,5-disubstituted moiety
(Figure 12).
source (34). In our extracts no catechin derivatives were observed,
reported by other groups (34). Interestingly, at least 12 isomeric
monocaffeoyl-hexoses were observed additionally, on which we
will report in due course.
ACKNOWLEDGMENT
Excellent technical support by Anja Mueller is acknowledged.
All further shikimic acid esters were tentatively identified, due
to the lack of authentic standards, due to similarities of tandem
MS spectra if compared to the monocaffeoyl esters. Compounds
47 and 48 are 3,4-dicaffeoyl derivatives due to the absence of
the fragment ion at m/z 316.9, suggesting 3-substitution and the
appearance of the fragment ion at m/z 160.9 in MS3, suggesting
4-substitution. Compound 49 is most likely 3,5-dicaffeoylshiki-
mic acid and compound 50 4-FSA following the same arguments.
Tricaffeoylshikimic acid was assigned as 3,4,5-tri-O-caffeoyl-
shikimic acid (51) (Figure 13) because there is only one regi-
oisomer. This compound was reported previously from this
Supporting Information Available: Additional MS2, MS3,
and MS4 spectra and high-resolution mass data. This material is
available free of charge via the Internet at http://pubs.acs.org.
LITERATURE CITED
(1) IUPAC. Nomenclature of cyclitols. Biochem. J. 1976, 153, 23-31.
(2) Clifford, M. N. Chlorogenic acids and other cinnamates - nature,
occurrence, dietary burden, absorption and metabolism. J. Sci. Food
Agric. 2000, 80, 1033–1043.

5484 J. Agric. Food Chem., Vol. 58, No. 9, 2010
Jaiswal et al.
(3) Clifford, M. N. Chlorogenic acids and other cinnamates - nature,
occurrence and dietary burden. J. Sci. Food Agric. 1999, 79, 362–372.
(4) Eliel, E. L.; Ramirez, M. B. (-)-quinic acid: configurational
(stereochemical) descriptors. Tetrahedron: Asymmetry 1997, 8,
3551–3554.
(5) Bravo, L.; Goya, L.; Lecumberri, E. LC/MS characterization of
phenolic constituents of mate (Ilex paraguariensis, St. Hil.) and its
antioxidant activity compared to commonly consumed beverages.
Food Res. Int. 2007, 40, 393–405.
(20) Clifford, M. N.; Knight, S.; Kuhnert, N. Discriminating between the
six isomers of dicaffeoylquinic acid by LC-MSn. J. Agric. Food Chem.
2005, 53, 3821–3832.
(21) Clifford, M. N.; Knight, S.; Surucu, B.; Kuhnert, N. Characteriza-
tion by LC-MSn of four new classes of chlorogenic acids in green
coffee beans: Dimethoxycinnamoylquinic acids, diferuloylquinic
acids, caffeoyl-dimethoxycinnamoylquinic acids, and feruloyl-di-
methoxycinnamoylquinic acids. J. Agric. Food Chem. 2006, 54,
1957–1969.
(6) Mansouri, A.; Embarek, G.; Kokkalou, E.; Kefalas, P. Phenolic
profile and antioxidant activity of the Algerian ripe date palm fruit
(Phoenix dactylifera). Food Chem. 2005, 89, 411–420.
(22) Clifford, M. N.; Marks, S.; Knight, S.; Kuhnert, N. Characterization
by LC-MSn of four new classes of p-coumaric acid-containing diacyl
chlorogenic acids in green coffee beans. J. Agric. Food Chem. 2006,
54, 4095–4101.
(23) Clifford, M. N. The analysis and characterisation of chlorogenic
acids and other cinnamates. In Methods in Polyphenol Analysis;
Santos-Buelga, C., Williamson, G., Eds.; Royal Society of Chemistry:
Cambridge, U.K., 2003; Vol. 14, pp 314-337.
(24) Clifford, M. N.; Kellard, B.; Birch, G. G. Characterisation of
caffeoylferuloylquinic acids by simultaneous isomerisation and
transesterification with tetramethylammonmium hydroxide. Food
Chem. 1989, 34, 81–88.
(7) Ziouti, A.; ElModafar, C.; ElMandili, A.; ElBoustani, E.; Macheix,
J. J. Identification of the caffeoylshikimic acids in the roots of the
date palm, principle fungitoxic compounds vis-a-vis Fusarium oxy-
sporum f sp albedinis. J. Phytopathol. 1996, 144, 197–202.
(8) Gang, D. R.; Beuerle, T.; Ullmann, P.; Werck-Reichhart, D.;
Pichersky, E. Differential production of meta hydroxylated phenyl-
propanoids in sweet basil peltate glandular trichomes and leaves is
controlled by the activities of specific acyltransferases and hydro-
xylases. Plant Physiol. 2002, 130, 1536–1544.
(9) Kuhnl, T.; Koch, U.; Heller, W.; Wellmann, E. Chlorogenic acid
biosynthesis - characterization of a light-induced microsomal 5-
(25) Clifford, M. N.; Zheng, W.; Kuhnert, N. Profiling the chloro-
genic acids of aster by HPLC MSn. Phytochem. Anal. 2006, 17,
384–393.
O-(4-coumaoyl)-D
-quinate/shikimate 30-hydroxylase from carrot
(Daucus carota L) cell-suspension cultures. Arch. Biochem. Biophys.
1987, 258, 226–232.
(10) Ishimaru, K.; Nonaka, G.; Nishioka, I. Tannins and related-com-
pounds. 57. Gallic acid-esters of proto-quercitol, quinic acid and (-)-
shikimic acid from Quercus mongolica and Quercus myrsinaefolia.
Phytochemistry 1987, 26 (5), 1501–1504.
(26) Clifford, M. N.; Zheng, W.; Kuhnert, N. Profiling the chlorogenic
acids of aster by HPLC MSn. Phytochem. Anal. 2006, 17, 384–
393.
(27) Agata, I.; Goto, S.; Hatano, T.; Ishibe, S.; Okuda, T. 1,3,5-Tri-O-
caffeoylquinic acid from Xanthium strumarium. Phytochemistry
1997, 33, 508–509.
(11) Folch, C. Stimulating consumption: yerba mate myths, markets and
meanings from conquest to present. Comp. Studies Soc. Hist. 2010,
52, 6–36.
(28) dos Santos, M. D.; Gobbo-Neto, L.; Albarella, L.; de Souza, G. E.;
Lopes, N. P. Analgesic activity of di-caffeoylquinic acids from roots
of Lychnophora ericoides (Arnica da serra). J. Ethnopharmacol. 2005,
96, 545–549.
(29) Ma, Y. T.; Huang, M. C.; Hsu, F. L.; Chang, H. F. Thiazinedione
from Xanthium strumarium. Phytochemistry 1998, 48, 1083–1085.
(30) Merfort, I. Caffeoylquinic acids from flowers of Arnica montana and
Arnica chamissonis. Phytochemistry 1992, 31, 2111–2113.
(31) Peluso, G.; de Feo, V.; de Simone, F.; Bresciano, E.; Vuotto, M. L.
Studies on the inhibitory effects of caffeoylquinic acids on monocyte
migration and superoxide ion production. J. Nat. Prod. 1995, 58,
639–646.
(32) Timmermann, B. N.; Hoffmann, J. J.; Jolad, S. D.; Schram, K. H.;
Klenk, R. E.; Bates, R. B. Constituents of Chrysothamnus pani-
culatus. 3. 3,4,5-Tricaffeoylquinic acid (a new shikimate pre-
aromatic) and 3,4-, 3,5- and 4,5-dicaffeoylquinic acids. J. Nat. Prod.
1983, 46, 365–368.
(33) Dugo, P.; Cacciola, F.; Donato, P.; Jacques, A. J.; Caramao, E. B.;
Mondello, L. High efficiency liquid chromatography techniques
coupled to mass spectrometry for the characterization of mate
extracts. J. Chromatogr., A 2009, 1216, 7213–7221.
(12) Schinella, G. R.; Troiani, G.; Davila, V.; de Buschiazzo, P. M.;
Tournier, H. A. Antioxidant effects of an aqueous extract of Ilex
paraguariensis. Biochem. Biophys. Res. Commun. 2000, 269, 357–360.
(13) Schubert, A.; Pereira, D. F.; Zanin, F. F.; Alves, S. H.; Beck, R. C.
R.; Athayde, M. L. Comparison of antioxidant activities and total
polyphenolic and methylxanthine contents between the unripe fruit
and leaves of llex paraguariensis A. St. Hil. Pharmazie 2007, 62, 876–
880.
(14) Gnoatto, S. C. B.; Schenkel, E. P.; Bassani, V. L. HPLC method to
assay total saponins in Ilex paraguariensis aqueous extract. J. Brazil.
Chem. Soc. 2005, 16 (4), 723–726.
(15) Gosmann, G.; Schenkel, E. P. A new saponin from mate
´
, Ilex
paraguariensis. J. Nat. Prod. 1989, 52, 1367–1370.
(16) Mazzafera, P. Caffeine, theobromine and theophylline distribution
in Ilex paraguariensis. Braz. J. Plant Physiol. 1994, 6, 149–151.
(17) Carini, M.; Facino, R. M.; Aldini, G.; Calloni, M.; Colombo, L.
Characterization of phenolic antioxidants from mate (Ilex para-
´
guayensis) by liquid chromatography mass spectrometry and liquid
chromatography tandem mass spectrometry. Rapid Commun. Mass
Spectrom. 1998, 12, 1813–1819.
(34) Markowicz Bastos, D. H. M.; Saldanha, L. A.; Catharino, R. R.;
Sawaya, A. C. H. F.; Cunha, I. B. S.; Carvalho, P. O.; Eberlin, M. N.
Phenolic antioxidants identified by ESI-MS from yerba mate (Ilex
paraguayensis) and green tea (Camelia sinensis) extracts. Molecules
2007, 12, 423–432.
(18) Chandra, S.; de Mejia, E. G. Polyphenolic compounds, antioxidant
capacity, and quinone reductase activity of an aqueous extract of
Ardisia compressa in comparison to mate (Ilex paraguariensis) and
´
green (Camellia sinensis) teas. J. Agric. Food Chem. 2004, 52, 3583–3589.
(19) Clifford, M. N.; Johnston, K. L.; Knight, S.; Kuhnert, N. Hier-
archical scheme for LC-MSn identification of chlorogenic acids.
J. Agric. Food Chem. 2003, 51, 2900–2911.
Received for review December 24, 2009. Revised manuscript received
March 24, 2010. Accepted March 27, 2010. Financial support from
Jacobs University Bremen is gratefully acknowledged.