ANALYTICAL BIOCHEMISTRY 238, 117–128 (1996)
ARTICLE NO. 0264
Analysis of Carotenoids and Carotenol Fatty Acid Esters
by Matrix-Assisted Laser Desorption Ionization (MALDI)
and MALDI–Post-Source-Decay Mass Spectrometry
Raimund Kaufmann,*,† Thomas Wingerath,‡ Dieter Kirsch,* Wilhelm Stahl,‡ and Helmut Sies†,‡
*Institut für Lasermedizin, ‡Institut für Physiologische Chemie, and †Biologisch-Medizinisches Forschungszentrum,
Heinrich-Heine-Universität, Postfach 101007, D-40001 Düsseldorf, Germany
Received August 24, 1995
Matrix-assisted laser desorption ionization (MALDI)
time-of-flight mass spectrometry and MALDI post-
source-decay (PSD) fragment ion mass analysis were
employed to examine various model carotenoids and
some of their fatty acid esters. It was demonstrated
that the use of MALDI helps overcome problems re-
sulting from the inherent thermal instability and lack
of solubility which render this group of compounds
rather difficult for conventional ionization tech-
niques. Detection limits were in the subpicomolar
range. Rather abundant metastable fragmentation oc-
curred under conditions of prompt extraction, but was
significantly restricted under conditions of delayed ex-
traction (DE). In DE the quasimolecular ion signals
(mainly as odd electron radical ions) became the base
peak even in spectra of the most delicate fatty acid
esters. PSD fragment ion patterns were similar to
those recorded under linked B/E scanning in conven-
tional fast atom bombardment–MS and contained in-
formation on structural end group substituents such
as hydroxyl, epoxide, or carbonyl functions and on the
extent of double bond conjugations. The mono- and
bis-fatty acid esters fragmented mainly by simple fatty
acid cleavage but also furnished some of the end
group-specific fragments. q 1996 Academic Press, Inc.
Conventional analytical protocols for the identifica-
tion of carotenoids and their fatty acid esters in biologi-
cal samples usually rely on reversed-phase HPLC re-
tention times in conjunction with UV/visible absorption
spectra. Although the unsubstituted compounds are
strong absorbers in the wavelength range 400–500 nm,
where most of them can be identified by their charac-
teristic absorption maxima, the spectral shifts of such
maxima induced by esterfication are usually small (1–
0003-2697/96 $18.00
Copyright q 1996 by Academic Press, Inc.
All rights of reproduction in any form reserved.
AID AB 9549 / 6m18$$$$61 05-31-96 03:42:35
2 nm) and nonspecific. On the other hand, identifica-
tion based on HPLC retention times alone can become
rather cumbersome, and the results obtained can be
ambiguous with complex mixtures where coelution is
fairly common (1, 2).
This situation lends itself to the use of mass spec-
trometry for unambiguous structural elucidation.
There is an abundant literature on the identification
of carotenoids by mass spectrometry (for a recent re-
view, see 3). Although carotenoids are amenable to al-
most all ionization protocols including EI,1 CI, FAB,
and PD, practical MS of carotenoids suffers from their
nonvolatility and thermal lability (EI, CI) and from
their lack of solubility and surface activity in the com-
mon liquid matrices of FAB–MS (4–6). This may ex-
plain why MS studies on the identification of carot-
enoids in biological samples are rather scarce.
With the advent of new ionization techniques such
as electrospray ionization (ESI) (7) and matrix-assisted
laser desorption ionization (MALDI) (8), high-molecu-
lar-mass compounds (especially biopolymers) hitherto
inaccessible have become amenable to mass analysis.
Especially MALDI, originally introduced as an ex-
tremely sensitive tool (femto- to attomolar range) for
mass analysis of very large molecules (proteins up to
300 kDa), has rapidly extended its applicability to other
groups of ‘‘difficult compounds’’ such as oligonucleo-
tides, oligosaccharides, and phospholipids (9–11).
A recent extension of MALDI time-of-flight (TOF)
mass spectrometry also provides structural informa-
tion on such molecules by means of so-called post-
1
Abbreviations used: EI, electron ionization; CI, chemical ioniza-
tion; FAB, fast atom bombardment; ESI, electrospray ionization;
MALDI, matrix-assisted laser desorption ionization; TOF, time-of-
flight; PSD, post-source-decay; DE, delayed extraction; PE, prompt
extraction.
117
abas AP: Anal Bio
118 KAUFMANN ET AL.
source-decay (PSD) fragment ion mass analysis (12,
13). For analytes amenable to MALDI–MS this ap-
proach can substitute for the more conventional MS/
MS techniques on sector- or triple quadrupol instru-
ments relying on less-sensitive ionization methods.
This study reports results of MALDI–PSD fragment
ion mass analysis performed on model carotenoids and
some of their fatty acid esters. The results demonstrate
that the body of structural information which can be
obtained by MALDI–PSD is very similar to that known
from conventional mass spectrometry but with greatly
improved sensitivity. Thus, this approach may be use-
ful for structural determination or confirmation of ca-
rotenoids isolated from natural sources such as plant
or animal tissues. An example of practical application
will be published elsewhere (15).
MATERIALS AND METHODS
Samples and Sample Preparation
Reference samples of cryptoxanthin, zeaxanthin, and
lutein were a gift from Dr. J. Bausch, Hoffmann–La-
Roche (Basel, Switzerland). b-Carotene was purchased
from Fluka (Basel, Switzerland). Partial synthesis of
further carotenoids and fatty acid esters was performed
as described (15). MALDI specimens were prepared by
dissolving 50–200 pmole of analyte in about 20 ml ace-
tone. To this solution 10 ml of acetone saturated with
2,5-dihydroxybenzoic acid was added. A 5- to 10-ml ali-
quot (sample load of about 10–60 pmol) of this mixture
was pipetted on the surface of a slightly preheated sam-
ple holder, on which the solvent evaporated within a
few seconds. Transfer of the sample into the instru-
ment was performed as quickly as possible.
MALDI (PSD) Mass Spectrometry
The investigation was carried out by means of two
in-house-built TOF–MS instruments. The technical
details of the two instruments have been described
elsewhere (13). Their main features are summarized
as follows: Both instruments use a nitrogen laser (VSL
337 ND Laser Sciences, Inc., Newton, MA) for desorp-
tion. In the linear TOF–MS (MS1), standard accelera-
tion voltage is 20 kV applied over a split acceleration
region (7/7 mm). The field free drift path has a length
of 110 cm; ion detection is performed by means of a 75-
mm-diameter dual microchannel plate with its front
face biased to ground. Over the last 25 cm of the field
free drift path, a retarding field can be applied in order
to separate the flight times of charged and neutralized
ion species.
The second instrument (MS2) is a reflectron-type in-
strument employing a gridded two-stage reflectron at
the end of a first field free drift path of 204 cm in length.
AID AB 9549 / 6m18$$$$62 05-31-96 03:42:35
Its second field free drift path (reflectron to detector)
extends over 175 cm. The nominal half angle between
the two drift path center lines is 47. Standard accelera-
tion voltage in this instrument is 10 kV. For ion detec-
tion a 75-mm-diameter dual microchannel plate is
again employed. The ion source in this instrument is
unusual insofar as accelerated ions are fed through a
center bore (6-mm diameter) in a coaxial optical/ion
optical device which simultaneously serves for laser
focusing, sample imaging, and ion optical beam colli-
mation, all in a common axis strictly perpendicular to
the surface plane of the sample. The rationale behind
this arrangement is the peculiar angular distribution
of the MALDI ion plume. It has been shown (16, 17)
that the major fraction of analyte ions formed is con-
tained in a jet-like forward-peaking cone, but that its
take of angle deviates from the instrumental z axis in
a fashion roughly symmetrical to the incident angle of
the laser beam. Thus, it is believed that an incident
laser beam strictly perpendicular to the sample plane is
the most favorable arrangement for achieving ultimate
useful ion transmission and source stability.
The acceleration stage is a split device (2/6 mm) also
in MS2 with the option to apply delayed extraction
(DE) conditions. This means that desorbed ions are ini-
tially (for Ç0.2–0.5 ms) allowed to expand against a
slightly retarding field before an accelerating field is
turned on (fast HV switch). If the delay time and the
ratio of field strengths in the two acceleration stages
are properly chosen (settings dependent on the mass
range of interest), so-called velocity focusing conditions
can be fulfilled, resulting in a dramatic improvement
of mass resolution (18, 19). In the present investigation
delay times between the desorption pulse and the onset
of the extraction field of 0.23–0.44 ms were chosen, and
a ratio of 0.47 for the field strength in the first and
second acceleration stages, respectively, was found to
perform best with respect to mass resolution of the
quasimolecular ion species. Typical values obtained un-
der these conditions were 2500–4000 M/dM (FWHM).
MS2 has a so-called precursor ion selector. This is an
electrostatically switched ion gate located at a distance
of about 220 mm from the sample. It allows a precursor
ion selection with a mass selectivity of about 60 M/dM
and is a mandatory device for PSD fragment ion mass
analysis in nonhomogeneous samples.
PSD fragment ion spectra were sequentially recorded
over 12–15 mass windows by incremental reduction of
the reflectron voltages. In each spectral window 30–
100 single-shot spectra were averaged depending on
sample conditions. For signal recording, the output of
the MCP was fed to a 300-MHz digital oscilloscope (Le-
Croy 9450A Chestnut Ridge, NY) and digitized at a
sampling rate of either 10 or 5 ns. Signal averaging
and processing were carried out using a 486 PC with
abas AP: Anal Bio
 SPECTROMETRIC ANALYSIS OF CAROTENOIDS AND CAROTENOL ESTERS
Ulisses software (version 7.31, Chips at Work, Bonn,
Germany). A subroutine provides for easy PSD ion
mass assignment without the need to calibrate the in-
strument by PSD reference standards (for further de-
tails see 13). Accuracy and precision of PSD mass as-
signments reach {0.2–0.3 u in the mass range of m/z
500–1000. Compilation (stitching) of a set of PSD mass
windows was performed by a subroutine contained in
the ‘‘MALDI–PSD Peptide Sequencer’’ software (ver-
sion 3.2, copyright by Frank Lützenkirchen, ILM, Düs-
seldorf, Germany).
RESULTS
Model–Carotenoids in MALDI–TOF MS
Sensitivity. Carotenoids and their fatty acid esters
were detected by MALDI–TOF MS at about the same
sensitivity as other analytes amenable to MALDI (pep-
tides, oligosaccharides). Typical sample loads for PSD
fragment ion analysis used in this study were in the
range 10–100 pmol but only a few femtomoles was
consumed during a full session of PSD fragment ion
mass analysis (200–1000 laser shots with an attomolar
consumption per shot). Under DE conditions quasimo-
lecular ions were detected with sample loads down to
some tens of femtomoles.
Molecular ion species. All carotenoids and fatty acid
esters investigated so far produced mixed populations of
odd- and even-electron molecular ions. In hydrocarbons
the M/ radical ion and the [M / H 0 2H]/ ion were
dominant. [M 0 2H]/ ions were seen only in the case of
b-carotene (see Fig. 1, top left). The presence of an oxygen
FIG. 1. Isotopic pattern of molecular ion regions in the spectra of
cartenoids recorded under delayed extraction (DE) conditions. See
Table 1 and text for further explanation.
AID AB 9549 / 6m18$$$$62 05-31-96 03:42:35
119
and/or a hydroxyl function induced the appearance of
protonated [M / H]/ and of sodiated [M / Na]/ ions,
which were occasionally even more abundant. It appears
that the ratio of [M / H]//M/• tends to increase with the
number of polar groups which provide possible sites for
protonation (see Table 1). These findings are in line with
similar observations in FAB–MS (4).
Although the apparent abundance ratios of M/•/
[M / H]/ seem to indicate that, in the majority of com-
pounds, protonated molecular ions represent only a mi-
nor fraction of the total precursor ion yield, this must
be taken with caution. We have observed in other
MALDI–PSD studies that the degree of metastable
fragmentation depends largely on the type of precursor
ion chosen and that protonated molecular ions tend to
be more prone to subsequent metastable disintegration
than, e.g., radical (or sodiated) precursors. Thus, in the
present case we cannot exclude the possibility that the
apparent yields of molecular ion species possibly under-
estimate the true [M / H]/.
The actual state of the art of our instrumentation did
not allow selection of monoisotopic precursor ion species.
Thus, while sodiated precursors could easily be sup-
pressed by the ion gate, we had to accept that the PSD
fragment ion mass analysis is inevitably carried out on
a mixed precursor ion population consisting of mainly
M/• and minor fractions of [M / H]/ and [M 0 H]/
species. Nonetheless, for the sake of convenience and sim-
plicity we will denote the molecular masses of our parent
ions as the monoisotopic mass of the majority M/• ion.
Also, the mass losses of PSD fragment ions are always
given as differences calculated with respect to the nomi-
nal M/• precursor ion mass.
PSD fragment ion yield. Under the conditions of
prompt extraction (PE), the total fraction of carotenoid
parent ions which underwent metastable fragmenta-
tion in the first field free drift path was remarkably
high. The abundance ratio (Ifr/Ip ) of ions which arrived
at the detector as either fragmented neutrals (Ifr ) or
unfragmented precursors (Ip) could be readily deter-
mined in MS1 by means of the retarding field in front
of the detector. Ifr/Ip ratios varied between 0.15 and
0.25 in the case of hydrocarbons but could reach over
0.9 in the case of some diepoxides or fatty acid esters.
Since in the MS2 instrument flight times and, hence,
time spans available for metastable decay are more
than twice as long as those in MS1, occasionally the
molecular ion signals of the most unstable parent ions
(e.g., some fatty acid esters) barely exceeded the noise
level, although abundant PSD fragment ion signals
could be recorded at the same time. Thus, in MALDI–
PSD MS of carotenoids and their fatty acid esters,
yields of diagnostic fragment ions are higher than those
occurring in B/E-linked tandem MS by more than an
abas AP: Anal Bio
120 KAUFMANN ET AL.

TABLE 1
Relative Intensity Distributionsa of Quasimolecular Ion Species Observed in MALDI–TOF MS of Some Carotenoids

No. of No. of
Compound Mol. mass. monoisotop. {OH |O (M 0 1)/ M/• (M / 1)/ (M / 2)/•

Lycopene 536.4 0 0 15 100 3 2

b-Carotene 536.4 0 0 70 100 3 0
Cryptoxanthin 552.4 1 0 26 100 6 1
Canthaxanthin 564.4 0 2 12 100 26 13
Zeaxanthin 568.4 2 0 30 100 21 1
Astaxanthin 596.4 2 2 10 100 55 2
Zeaxanthin laurate–myristate 960.6 2 0 10 100 2 3
Antheraxanthin palmitate–myristate 1032.6 2 1 — 100 3 8

Note. Data of [M / H]/ and [M / 2H)]/• are derived from averaged (32–40 shots) signal intensities (as recorded under delayed extraction
conditions) by subtracting the expected isotopic (13C) contributions of the M/• species. For comparison with FAB–MS see data in (4).
a
Note. Intensities expressed as percentages of the base peak M/• signal rounded to the nearest integer value.

order of magnitude (3, 4, 6). Under the conditions of
DE, the yield of PSD ions decreased by up to an order
of magnitude (see also Figs. 2 and 5). It appears, how-
ever, that the drop in PSD fragment ion abundances
occurs mainly at the expense of the nondiagnostic low
mass fragments, whereas the intensities of diagnosti-
cally useful ions (middle and higher mass range) are
much less affected (factor of 0.7–0.3). Thus, the
MALDI–PSD spectra shown in this work were re-
corded under DE conditions unless otherwise stated.
PSD Fragment Ion Pattern
Elimination of in-chain fragments. In line with
conventional mass spectrometry, MALDI–PSD spectra
of hydrocarbon compounds with cyclic end groups con-
sisted nearly exclusively of fragments formed by in-
chain elimination of toluene and xylene. In the case of
b-carotene, the dominant PSD-fragment ion signal was
the [M 0 92]/• species (see Figs. 3A–3C). The much
smaller [M 0106]/• signal never exceeded 10% of the
[M 0 92]/• signal. The (IM 0 92/IM 0 106 ) ratio of ú10 is
consistent with corresponding conventional mass spec-
trometric data for carotenoids carrying nine double
bonds in the polyene chain (20, 21).
Formation of [M 0 78]/•, [M 0 79]/, and [M 0 80]/•
was seen occasionally in hydrocarbon compounds and
carotenoids with hydroxylated end groups but never
exceeded 2% of the [M 0 92]/• fragment ion intensity.
Abundant [M 0 80]/• fragments were, however, ob-
served in all investigated cases of 5,6- and 5,8-epoxides
(see Fig. 4) and are an important diagnostic feature for
this group of substances.
Elimination of end groups and end group consti-
tuents. It is well known from conventional mass
spectrometry of carotenoids that most cyclic end
groups are rather stable. Thus, in FAB MS/MS (2, 4,
6), carotenoids with unsubstituted or hydroxylated
ring structures furnish a polymorphic pattern of only
small to middle mass fragments which, from a diag-
nostic point of view, are rather unspecific (the only
exceptions are a-carotene and lutein; see below). If,
however, end groups contain 5,6- or 5,8-epoxide func-
tions, sets of characteristic low and middle mass frag-
ments are produced by breaking of polyene bonds be-
tween positions 8 and 8*. Larger fragments of these
species can be formed by mechanisms involving the
transfer of a (radical) hydrogen from the larger to the
smaller fragment.
The appearance of MALDI–PSD fragment ion spec-
tra obtained from carotenes and xanthophylls was
largely in line with the above rules.
Carotenoids with unsubstituted or hydroxylated
rings exhibited the expected unspecific cleavage pat-
tern in the m/z 70–250 mass range. The only excep-
tions were found for a-carotene and lutein (see Figs. 3
and 4), in which the double bond of the cyclic end group
is not conjugated with the polyene chain. In both cases
elimination of end groups was seen by bond breaking
at position 6/7 (6*/7*). This mechanism furnished rather
prominent diagnostic fragments at m/z 123 and m/z
413 [M 0 123]/ for a-carotene and at m/z 429 [(M 0
138) 0 H]/ for lutein. The elimination of a fragment
by ring opening was found only in lutein, where the
formation of a [M 0 56]/• fragment is a highly diagnos-
tic feature (see Fig. 4B).
The presence of one or two epoxide functions in the
5,6- and/or 5,8-position (antheraxanthin, violaxanthin,
luteoxanthin) very much favored the characteristic
mechanism of cleavage and rearrangement in the poly-
ene chain. Thus, a wealth of diagnostic fragments (see
Fig. 5 and Table 2) have been recorded from such com-
pounds. The most prominent was usually the m/z (204
/ R) signal (R Å H or OH at position 3) which, in the

AID AB 9549 / 6m18$$$$63 05-31-96 03:42:35 abas AP: Anal Bio

 SPECTROMETRIC ANALYSIS OF CAROTENOIDS AND CAROTENOL ESTERS 121

FIG. 2. Comparison of MALDI–PSD fragment ion spectra of cryptoxanthin recorded under conditions of delayed extraction (top) and
prompt extraction (bottom). See text for further explanation.

case of violaxanthin and luteoxanthin, largely exceeded
the amplitude of any other fragment.
The presence of carbonyl groups as in canthaxanthin or
astaxanthin also facilitated polyene bond fragmentation.
The patterns, however, were qualitatively and quantita-
tively different from those of the epoxide species. The type
of fragment which appears to be diagnostic for this group
is the cleavage of the 8/7 (8*/7*) bond(s).
The data shown in Table 2 summarize the major
observations. Many of the diagnostic signals in this
group reached 10–40% precursor ion intensity.
Hydroxylated carotenoids such as cryptoxanthin or
zeaxanthin formed [M 0 17]/ and [M 0 18]/• ions; the
presence of an oxygen function induced the formation
of an [M 0 16]/• fragment. Analog satellite signals
were also seen with the [M 0 92]/• fragment. Those
signals, however, were difficult to entangle due to the
mixed precursor ion population (see above). It is note-

AID AB 9549 / 6m18$$$$63 05-31-96 03:42:35 abas AP: Anal Bio

122 KAUFMANN ET AL.

FIG. 3. PSD fragment ion spectra displaying diagnostic ions in the middle and higher mass range of (A) b-carotene, (B) a-carotene, and
(C) lycopene.

worthy that under DE conditions the abundance of the
[M 0 18]/• fragment was significantly increased up to
50% precursor intensity.
The only acyclic carotenoid included in this study was
lycopene (see Fig. 3). Here we observed the formation of
an abundant [M 0 69]/ ion which occasionally exceeded
the intensity of the [M 0 92]/• signal. Additional promi-
nent signals occurred at m/z 398 [M 0 2 1 69]/• and m/
z 375 [(M 0 92) 0 69]/. This pattern, although qualita-
tively consistent with B/E linked-scan spectra of lycopene
(2, 4), differs quantitatively insofar as in the MALDI–
PSD spectra the m/z 398 and 375 signals seem to be

AID AB 9549 / 6m18$$$$63 05-31-96 03:42:35 abas AP: Anal Bio

 SPECTROMETRIC ANALYSIS OF CAROTENOIDS AND CAROTENOL ESTERS 123

FIG. 4. PSD fragment ion spectra (DE conditions) of (A) zeaxanthin and (B) lutein. Note that these isomeric xanthophylls can be readily
distinguished by some end group eliminations (e.g., m/z 512 [M 0 56]/•, m/z 429 [M 0 139]/) which, in agreement with MS/MS data, are
diagnostic for lutein. The rather prominent fragment ion peak at m/z 338 in the lutein spectrum has not been observed in conventional
MS. See text for further explanation.

much more intense (up to 50% precursor ion intensity)
than those in the B/E linked-scan spectra.
Carotenoid fatty acid esters. Mass spectrometric
studies on carotenoid fatty acid esters are scarce (22,
23). Molecular ions are rather unstable under EI condi-
tions and reach only 5–10% relative intensity (22). Bet-
ter results have been obtained with ammonia desorp-
tion chemical ionization (23).
While in MALDI MS, at least under conditions of
prompt desorption when the rate of metastable frag-
mentation can be rather high (up to 95% fragmented
ions), delayed desorption can reduce the extent of
metastable fragmention to under 10% so that molecu-
lar ion signals become the base peak in the mass
spectra (Fig. 6).
The MALDI–PSD spectra of carotenoid fatty acid
esters are remarkably simple (see Table 3). The vast
majority of fragment ions belong to only four groups of
diagnostic ions. Three of them are formed by the loss
of either one or both fatty acids: [M 0 R1COOH]/•, [M
0 R2COOH]/•, [M 0 R1COOH 0 R2COOH]/•. De-
pending on the nature of the xanthophyll, all three
species of fragments can be accompanied by either [M
0 92]/• or [M 0 80]/• or both satellites formed by the

AID AB 9549 / 6m18$$$$63 05-31-96 03:42:35 abas AP: Anal Bio

124 KAUFMANN ET AL.

FIG. 5. PSD fragment ion spectra of (A) zeaxanthin, (B) canthaxanthin, and (C) violaxanthin. Note the multiplicity of end-group-specific
diagnostic signals formed by bond breaking in the polyene chain of the epoxidized species (B) and (C).

usual in-chain elimination of toluene or xylene. Finally,
there are also abundant [M 0 92]/• and/or [M 0 80]/•
fragments without loss of a fatty acid. In contrast to
EI ionization, MALDI–PSD does not furnish ionized
fatty acids or fragments thereof.
The tendency of epoxidized xanthophylls to produce
end group eliminations via polyene bond breaking
seems to be preserved in the respective fatty acid es-
ters. Especially under PE conditions we found a charac-
teristic repetitive signal pattern over nearly the whole
middle and high mass range (see also Fig. 6, bottom).
Most of these signals seem to represent fragments
formed by eliminations of an end group with and with-
out cleavage of the respective fatty acid. While these

AID AB 9549 / 6m18$$$$63 05-31-96 03:42:35 abas AP: Anal Bio

AID
TABLE 2
Patterns of Main MALDI – PSD Fragment Ions Recorded in the Spectra of Carotenes and Xanthophylls (Delayed Extraction Conditions)

AB 9549
|O In-chain eliminations M/•- Elimination reactions of terminal groupsa,b (polyene bond cleavage)
{OH

/
{O{ 80 92 106 146 172 9/10 10/11 11/12 12/13 14/15 15/15* 15*/14* 14*13* 13*/12* 12*/11* 11*/10* 10*/9* 9*/8* 8*/7* 7*/6*

Carotenes
0
b-Carotene XXX X
0
M/• Å 536.4 u 444 428
0
0

6m18$$9549
Lycopene XXX
0
M/• Å 536.4 u 444
0
Xanthophylls
0
Cryptoxanthin XXX X
1
M/• Å 522.4 u 460 444
0
0
Zeaxanthin XXX X
2
M/• Å 568.4 u 476 460
0
0

05-31-96 03:42:35
Lutein XX (X) (X) XX
2
M/• Å 568.4 u 476 460 416 429
0
0
Cryptoxanthin X XXX X XXX X X X X X XX
1

abas
5,6-Epoxid 488 476 422 221 231 283 296 336 349 375
1
0
Antheraxanthin X XXX (X) XXX X X X (X) X X
2
M/• Å 584.4 u 504 492 476 221 247 287 299 312 352 365
1
0
Violaxanthin XX X (X) XX (X) XXX X XX XX XX X (X) (X) XX XX XXX
2
M/• Å 600.4 u 520 494 454 428 181 221 234 247 287 299 312 325 352 365 391 419
2
0
Luteoxanthin XX X (X) XX X XXX (X) XX XXX XX XX X X XX XX XXX

AP: Anal Bio

2
M/• Å 600.4 u 520 494 454 428 181 221 234 247 287 299 312 325 352 365 391 419
2
2
Canthaxanthin XXX XX X (X) (X) (X)
0
M/• Å 564.4 u 472 203 269 281 284 413
0
SPECTROMETRIC ANALYSIS OF CAROTENOIDS AND CAROTENOL ESTERS

2
Astaxanthin XXX (X) (X) X XX X X XX
2
M/• Å 596.4 u 504 488 219 (285) 298 311 363 429
0

Note. Signal intensities: XXX, ú40% (bold, leading signal); XX, 10 – 40%; X, 5 – 10%; (X); õ5% precursor ion intensity.
a
Analogous eliminations from the right-hand side of the molecule are ignored.
b
Larger fragments formed by bond breakings between position 15 and 8* can occur with and without transfer of a hydrogen to the smaller fragment. The mass numbers
given in italic indicate the respective fragments.
125
126 KAUFMANN ET AL.

FIG. 6. PSD fragment ion spectra of bis-fatty acid esters of antheraxanthin recorded under conditions of delayed extraction of palmitate-
myristate (top) and prompt extraction of dipalmitate (bottom).

species of fragment ions are rather equally distributed
over the middle and high mass range and, thus, cannot
be identified for diagnostic purposes, there are at least
a few of the xanthophyll-specific end group fragments
at the low mass end which help to identify the nature
of the carotenoid involved. This is particularly true for
epoxidized end groups where the prominent signal at
m/z 221 of, e.g., antheraxanthin, violaxanthin, and lu-
teoxanthin (bond breaking at position 10/11) transform
into a signal at m/z 203 (1 H2O has been lost by the
cleavage of the respective fatty acid).
In conclusion, the present study has demonstrated
that MALDI produces molecular ions of nonpolar car-
otenes of polar xanthophylls and of their fatty acid
esters equally well. Metastable PSD is abundant es-
pecially with the fatty acid esters where, under PE,
only a few percent of the molecular ions reach the
detector as intact molecules. It is generally assumed
that ions produced by MALDI acquire much internal
energy by collisions during the early phase of acceler-
ation through the expanding plume of laser-induced
particles. This can be minimized, however, by the

AID AB 9549 / 6m18$$$$63 05-31-96 03:42:35 abas AP: Anal Bio

 SPECTROMETRIC ANALYSIS OF CAROTENOIDS AND CAROTENOL ESTERS 127

TABLE 3
Pattern of MALDI–PSD Fragment Ions Recorded in the Spectra of Fatty Acid Esters of Xanthophylls

Cleavages of fatty acids

Elimination of
In-chain [M- [M-R2- [M-R1COOH- terminal
cleavages R1COOH]/• COOH]/• R2COOH]/• groups: polyene
M/• chain cleavages
092 092 092
092 080 080 080 080 leading

Cryptoxanthin monopalmitate XX XX XX
—
Zeaxanthin monomyristate XX XX XXX
—
Zeaxanthin dipalmitate XX XX XX — —
—
Zeaxanthin laurate–myristate XX XX XX XX XX XXX
— —
Zeaxanthin myristate–palmitate XX XX XX XX XX
— —
Antheraxanthin dipalmitate XX XX XX XXX — — XX XX (X) 203
(X)
Antheraxanthin palmitate–myristate XX X XX X XX XX X (X) (X) 203
(X) (X) —
Violaxanthin myristate–palmitate XX (X) XX X XX X X (X) 203
(X) (X)

Note. Delayed extraction (DE) mode. Code of signal intensities as in note to Table 2.

protocol of DE, which indeed restricted the extent ACKNOWLEDGMENT

of (unspecific) metastable fragmentation by up to an
order of magnitude. Remarkably enough, fragment
ion formation of diagnostic ions was far less sup-
pressed under DE conditions, which, thus, appear to
be the modality of choice for the investigation of ca-
rotenoids with MALDI – PSD.
The sensitivity limits have not been systemically
explored. For the PSD studies, standard sample loads
employed were in the range of 10 – 60 pmol; molecular
ion detection could, however, be followed down to the
range of 10 fmol where S/N ratios for molecular ions
were still at about 10 or higher. It is believed that
the MALDI – PSD approach can overcome many of the
difficulties encountered in the conventional MS anal-
ysis of carotenoids and thus, from a practical point
of view, may supplement and facilitate the structural
confirmation of this group of compounds.
It might be suspected that carotenoids, being strong
absorbers in the blue, would not need a matrix for as-
sisted desorption but rather could be directly desorbed
by lasers (such as the 337-nm N2 laser). In a few pilot
experiments, however, we found that laser desorption
of unprotected carotenoids leads to excessive (mainly
unspecific) prompt fragmentation of the analytes and,
most probably, also to photofragmentation of sample
in the solid phase.
Technical support for this study was provided by Thermo Bioanaly-
sis, Paradise, Hemel, UK.
REFERENCES
1. Schmitz, H. H., van Breemen, R. B., and Schwartz, S. J. (1992)
Methods Enzymol. 213, 322–336.
2. van Breemen, R. B., Schmitz, H. H., and Schwartz, S. J. (1993)
Anal. Chem. 65, 965–969.
3. Enzell, C. R., and Back, S. (1995) in Carotenoids (Britton, G.,
Liaaen-Jensen, S., and Pfander, H., Eds.), Vol. 1B, Spectroscopy,
pp. 261–317, Birkhäuser Verlag, Basel.
4. Caccamese, S., and Garozzo, D. (1990) Org. Mass Spectrom. 25,
137–140.
5. Lusby, W. R., Khachik, F., Beecher, G. R., and Lau, J. (1992)
Methods Enzymol. 213, 111–129.
6. van Breemen, R. B., Schmitz, H. H., and Schwartz, S. J. (1995)
J. Agric. Food Chem. 43, 384–389.
7. Mann, M., Shen, S., and Fenn, J. B. (1992) in Mass Spectrometry
in the Biological Sciences (Gross, M. L., Ed.), pp. 145–163,
Kluwer, Dordrecht.
8. Karas, M., and Hillenkamp, F. (1988) Anal. Chem. 60, 2299–
2301.
9. Stemmler, E. A., Buchanan, M. V., Hurst, G. B., and Hettich,
R. L. (1995) Anal. Chem. 67, 2924–2930.
10. Spengler, B., Kirsch, D., and Kaufmann, R. (1995) J. Mass Spec-
trom. 30, 782–787.
11. Harvey, D. J. (1995) J. Mass Spectrom. 30, 1333–1346.

AID AB 9549 / 6m18$$$$63 05-31-96 03:42:35 abas AP: Anal Bio

128 KAUFMANN ET AL.

12. Kaufmann, R., Spengler, B., and Lützenkirchen, F. (1993) Rapid
Commun. Mass Spectrom. 7, 902–910.
13. Kaufmann, R., Kirsch, D., and Spengler, B. (1994) Int. J. Mass
Spectrom. Ion Processes 131, 355–385.
14. Spengler, B., Kirsch, D., Kaufmann, R., and Jaeger, E. (1992)
Rapid Commun. Mass Spectrom. 6, 105–108.
15. Wingerath, T., Stahl, W., Kirsch, D., Kaufmann, R., and Sies,
H. (1996) Submitted for publication.
16. Bökelmann, V., Spengler, B., and Kaufmann, R. (1995) Eur.
Mass Spectrom. 1, 81–93.
17. Aksouh, F., Chaurand, P., Deprun, C., Della-Negra, S., Hoyes,
J., Le Beyec, Y., and Rosas Pinho, R. (1995) Rapid Commun.
Mass Spectrom. 9, 515–518.
18. Brown, R. S., and Lennon, J. J. (1995) Anal. Chem. 67, 3990–
3999.
19. Vestec, M. L., Juhasz, P., and Martin, S. A. (1995) Rapid Com-
mun. Mass Spectrom. 9, 1044–1050.
20. Enzell, C., Franzis, G., and Liaaen-Jensen, S. (1968) Acta Chem.
Scand. 22, 1054–1058.
21. Budzikiewicz, H., Brzezinka, H., and Johannes, B. (1970) Mo-
natsh. Chem. 101, 579.
22. Khachik, F., and Beecher, G. R. (1988a) J. Agric. Food Chem.
36, 929–937.
23. Khachik, F., Beecher, G. R., and Lusby, W. R. (1988) J. Agric.
Food Chem. 36, 938.

AID AB 9549 / 6m18$$$$64 05-31-96 03:42:35 abas AP: Anal Bio