Phospholipids (PLs) are membrane lipids of living cells whose considerable role in biological membranes include protein sorting and regulation of biophysical properties and signaling pathways. PLs are classified by their head groups into phosphatidic acid, phosphatidylcholine (PC), phosphatidylglycerol, phosphatidylinositol (PI), phosphatidylserine (PS), and cardiolipin. Since PLs have varying ionization efficiencies, depending on their electron affinity, they can be detected at positive or negative ion modes so that PC and PS are generally detected as positive ions, and phosphatidylethanolamine and PI as negative ions. As a result, metabolite analyses in time-of-flight secondary ion mass spectrometry (ToF-SIMS) should be carried out by performing tandem mass spectrometry measurements at both ion modes to identify unknown PLs. For tandem mass spectrometry measurements in ToF-SIMS, a postsource decay (PSD)-like method was successfully applied to identify several lipids by using cholesterol as a model molecule at the positive ion mode. In our study, the authors adapted 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphor-rac-(1-glycerol) ammonium salt with well-known fragmentation pathways as a model molecule at the negative ion mode to identify PI lipids. By using the PSD-like method at both ion modes, the authors successfully identified PC and PI from MDA-MB-231 breast cancer cell lysates to show that our PSD-like method would be useful in the process of identifying unknown lipids from biological samples in ToF-SIMS.

Phospholipids (PL) are essential components of the cell membrane by contributing to the cell's biochemical and biophysical properties such as protein sorting and cell signaling.1,2 The types of phospholipid and their distributions vary depending on the cell type and tissues. Because of the important role of PL in metabolism and in regulating development, analyses of molecular species and their modifications have been actively performed.3,4 For decades, thin layer chromatography and gas chromatography-mass spectrometry were used for separation and identification of the lipids; however, these methods were time consuming and required large amounts of lipid samples.5–7 In more recent years, soft ionization techniques such as fast atom bombardment (FAB), electrospray ionization (ESI), matrix-assisted laser desorption/ionization time-of-flight (MALDI-TOF), and TOF secondary ion mass spectrometry (ToF-SIMS) have been successfully adopted to detect and analyze intact PL molecules.8–13 

ToF-SIMS in particular possesses advantages such as high sensitivity, tolerance to impurities, and rapid sample preparation, so it is regarded as an efficient method for biological compound analysis.14,15 Its greater sensitivity to lipid analysis, as opposed to other soft ionization sources such as FAB, ESI, and MALDI makes ToF-SIMS an excellent analytical approach for rapid screening of lipid components in biological matrices. However, in this method, identification is largely based on molecular weight, with limited mass accuracy for small molecules. There is a general consensus that additional proof is required to identify ions of higher molecular mass such as biomolecules, so structural analysis is often performed with tandem mass spectrometry. It has been reported that phosphatidylcholines (PC) could be analyzed by MALDI-“postsource decay” (PSD),16 and Brunelle et al. reported a successful PSD analysis on a ToF-SIMS instrument.17 Hence, it is reasonable to expect that structural analysis can be widely used in ToF-SIMS or other ToF based tandem mass instruments. Brunelle's studies were performed on positive ions using cholesterol as a calibrant. Briefly, the relationship between the fragment mass and its detection time is shown in Eq. (1) with the K value dependent on the instrumental parameter. The mass difference between the precursor ion (mp) and fragment ion (mf) are related as follows by K × mp and the time difference between the stable ion (tp) and fragment ion (tf)

mpmf=K×mp×(tptf)withK=2Ep/4d(Brunelle).
(1)

In theory, the principle of PSD that was applied to positive ions could also be adopted for negative ions. However, the K value for the positive ions is slightly different from the K value for the negative ions even under the same experimental conditions, due to the imperfection of the field inside the reflectron. One can estimate mass error when K is measured for fractional mass. Since several parameters are changed for the negative ion analysis, K should be measured for known fragment ions, and the mass error should be estimated.

Here, we adopted 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphor-rac-(1-glycerol) ammonium salt (POPG) with well-known fragmentation pathways as a negative calibrant. Using this new calibrant, we successfully identified an unknown lipid extracted from MDA-MB-231 breast cancer cell lysate as phosphatidylinositol (PI) in the negative mode and confirmed the identification with a pure compound of PI. Positive ion PSD was conducted using cholesterol as the calibrant to identify the unknown peak from the same cell line, MDA-MB-231, by comparing the fragmentation patterns of the unknown and pure compound of PC. Based on these results, PSD would be a useful method to identify unknown lipids from biological samples in ToF-SIMS.

ToF-SIMS spectra were obtained with a ToF-SIMS V (ION-TOF GmbH, Germany) equipped with a 25 keV Bi3+ ion gun. For the ToF-SIMS spectrum, the ion gun was operated at 8.3 kHz with an average current of 0.28 pA (Bi3+) at the sample holder. A bunch pulse of 0.7 ns duration resulted in mass resolution (M/△M) > 8000 at m/z 760. A 200 × 200 μm2 area was rastered by primary ions to obtain the SIMS spectra while maintaining the ion dose below 1 × 1012 ions/cm2. For the PSD experiments, we used a deflection plate, which was located at the beginning of the first field-free region that starts after the extraction and focusing optics. The voltages of this deflection plate could be adjusted to deflect all the secondary ions and to select a specific m/z with a selected time window. The time window was adjusted with the start and stop times. These times were defined in percent of the total duration of analysis in each TOF cycle, i.e., 81.7%–85.7% and 75.5%–78.5% of the time window for negative and positive PSD, respectively.

Cholesterol, POPG, and solvents were obtained from Sigma (St. Louis, MO, USA). Phospholipid standards were purchased from Avanti Polar Lipid Inc. (Alabaster, AL, USA). Cholesterol, POPG, and phospholipids were prepared at concentrations of 1 mg/ml in methanol and 1 μl of each sample was deposited on a precleaned silicon wafer by super piranha solution for the analysis.

MDA-MB-231 cells were cultured in RPMI 1640 (Hyclone), 10% FBS media. The cells were harvested by trypsinization and the cell suspension was washed three times with 4 ml phosphate buffered saline, followed by centrifugation at 2000 rpm for 4 min. The 1 ml cell lysate was treated with 3.75 ml chloroform/methanol 2:1 (v/v) for lipid extraction. The mixture was vortexed and incubated on ice for 30 min, and an additional volume of 1.25 ml chloroform was added. Finally, following vigorous vortex, the samples were centrifuged at 1000 rpm for 5 min at room temperature.18 The upper supernatant was collected and the bottom cell pellet was discarded. The supernatant was dried by using a vacuum dry evaporator. For ToF-SIMS analysis, the dried sample extracted from 2 × 106 harvested cells was melted in a 100 μl solvent of chloroform/methanol 2:1 (v/v).

An ideal calibrant for PSD analysis should produce at least three intense and evenly distributed fragment ions in the whole mass range. Since the fragment mass is calculated from its TOF difference from the precursor, a higher precursor mass would be appropriate to determine the K value. It would also be preferable for the calibrant to be commercially available at a modest price.

POPG is a small lipid with a molecular mass of 747.5 Da. Information on its fragments have been previously reported.19 Since the fragment peaks are well resolved and the peaks can be easily assigned, POPG was adopted as a calibrant for our PSD analysis of negative ions. As shown in Fig. 1(a), the negative molecular ion of POPG produces numerous fragment ions as indicated by the dots; the fragmentation patterns are easily distinguished below the 80 μs region. When the effect of the primary ion beams on the PSD spectra was checked for the calibrants (e.g., cholesterol in the positive mode and POPG in the negative mode), we detected an insignificant difference between the Bi3+ ions and Ar2000+ ions as shown in Fig. S1.20 Because of its superior mass resolution, we used the Bi3+ as a primary ion in our study. In the reflectron TOF spectrum by PSD, the fragment ions originally appear at incorrect mass-to-charge ratio values because their velocities differ from the values of the ions generated at the source region. The calibration constant (K) is used to correlate the measured flight times with the true mass-to-charge ratio values at one reflectron voltage (Uref). In general, fragment ions take a longer time to reach the detector as Uref decreases from the maximum. The calibration constant was calculated from the relation between time and mass based on Eq. (1). For practical purposes, the POPG molecular ion was selected by adjusting the start and stop times of the ion gate as mentioned in Sec. II. The POPG precursor ion was fragmented in the field-free region to provide a unique pattern of TOF as shown in Fig. 1(a). The representative fragments of POPG that were used to calculate the K are shown in Table I. The K parameter depends only on instrumental geometry and extraction energy so that the same K value can be used in the PSD analysis of unknown molecules if the instrument setting conditions are the same.

Fig. 1.

(Color online) (a) Time-of-flight spectrum of the selected ion at m/z 747.5 (POPG) in the negative mode. (b) Reconstructed m/z spectrum of the selected ion at m/z 747.5 using the calculated K value.

Fig. 1.

(Color online) (a) Time-of-flight spectrum of the selected ion at m/z 747.5 (POPG) in the negative mode. (b) Reconstructed m/z spectrum of the selected ion at m/z 747.5 using the calculated K value.

Close modal
Table I.

Calculated mass-to-charge ratios of product ions, experimental time-of-flight values of the product ions and corresponding values of K for the precursor ion at 747.5 (POPG, [M–H]). The mean value of K is 5.7607 × 10−4 amu1/2 ns−1 (standard deviation: 1.33 × 10−6 amu1/2 ns−1).

Precursor ion (m/z)Time of flight (ns)K (amu1/2 ns−1)
747.5 92005.8  
Calculated products (m/z
491.3 75773.8 5.7730 × 10−4 
483.3 75263.7 5.7719 × 10−4 
465.3 74121.7 5.7714 × 10−4 
417.3 71065.1 5.7674 × 10−4 
391.3 69396.1 5.7623 × 10−4 
281.3 62346.2 5.7491 × 10−4 
255.2 60664.7 5.7453 × 10−4 
Precursor ion (m/z)Time of flight (ns)K (amu1/2 ns−1)
747.5 92005.8  
Calculated products (m/z
491.3 75773.8 5.7730 × 10−4 
483.3 75263.7 5.7719 × 10−4 
465.3 74121.7 5.7714 × 10−4 
417.3 71065.1 5.7674 × 10−4 
391.3 69396.1 5.7623 × 10−4 
281.3 62346.2 5.7491 × 10−4 
255.2 60664.7 5.7453 × 10−4 

By applying the relation between time and mass in Eq. (1) into the reflectron TOF spectrum of POPG, we were able to obtain 5.7607 × 10−4 as the K value in a negative ion mode, as shown in Table I. Using the calculated K value, the TOF spectrum was converted to a m/z spectrum. The reconstructed m/z spectrum is shown in Fig. 1(b). The K for positive mode was 5.7161 × 10−4 using the same relation and cholesterol as a calibrant molecule. The difference between the K values would likely be due to the differences in analyzer setting conditions. It should also be noted that each mode needs its own K value because the calculated mass difference (Δm = m/zcalc+ − m/zcalc, ex., Δm = 5 Da for phosphocholine, m/z 184) of the same fragment peak is too large to be tolerable when the mean K value (5.7384 × 10−4) obtained from the positive (5.7161 × 10−4) and negative (5.7607 × 10−4) values is used for both positive and negative PSD experiments. Considering the above results, it is clear that K should be adopted in each mode depending on its reflectron condition.

Mass spectrometry has been widely used to detect and determine lipid structure as a lipidomic methodology tool. Numerous breast cancer studies have shown that different kinds of phospholipids are present in different types of breast cancer cells and that the changes are associated with malignant progression.21 In order to determine all of the types of lipids in the cell lysate, both positive and negative ion modes should be employed to identify the unknown lipids. Phosphatidylethanolamine (PE) and PI are easily ionized as negative ions; negative PSD was considered to be an adequate method. We began with standard PE and PI molecules to confirm the usefulness of PSD in the negative ion mode. The 16:0/18:1–PE and 18:0/20:4–PI standard molecules were used. Under PSD, 16:0/18:1–PE underwent preferential losses of fatty acyl substituents as ketenes ([M−H–Rx′CH=C=O]) over as acids ([M−H–RxCO2H]) and showed characteristic peaks that were m/z 452 ion ([M−H–R2′CH=C=O]), m/z 478 ion ([M−H–R1′CH=C=O]), m/z 434 ion ([M−H–R2′CO2H]), and m/z 460 ion ([M−H–R1′CO2H]).22 The fatty carboxylic acid anions such as m/z 281 ion (R2CO2) and m/z 255 ion (R1CO2) were also detected as shown in Fig. S2(a).20 PSD mass spectrum of the [M−H] 18:0/20:4–PI at m/z 885 [Fig. 2(b)] contained ions at m/z 599 and 581, representing neutral losses of ketene ([M−H–R2′CH=C=O]) and neutral losses of fatty acids ([M−H–R2′CO2H]), respectively.23 Ions at m/z 437 and 419 lost an inositol head group such as [C6H10O5] from m/z 599 and 581 ions. The ions at m/z 303 and 283 were assigned to arachidonate (20:4) [C20H31O2] and stearate (18:0) [C18H35O2], respectively.24 

Fig. 2.

(Color online) (a) Reconstructed PSD spectrum of the selected ion m/z 885, taken from a sample of lipids extracted from MDA-MB-231 cell lysate in the negative ion mode. The spectrum was magnified 100 times under m/z 650. The inset chemical structures are m/z 581 and m/z 419. (b) Reconstructed PSD spectrum of the precursor ion m/z 885 (18:0/20:4 phosphatidic inositol) obtained from a commercial compound as a standard in the negative mode.

Fig. 2.

(Color online) (a) Reconstructed PSD spectrum of the selected ion m/z 885, taken from a sample of lipids extracted from MDA-MB-231 cell lysate in the negative ion mode. The spectrum was magnified 100 times under m/z 650. The inset chemical structures are m/z 581 and m/z 419. (b) Reconstructed PSD spectrum of the precursor ion m/z 885 (18:0/20:4 phosphatidic inositol) obtained from a commercial compound as a standard in the negative mode.

Close modal

As a model system, we selected the MDA-MB-231 breast cancer cell which has an abundance of PE and PI. The lipids were extracted by using a chloroform/methanol mixture solution as explained in Sec. II. The PSD spectrum of the selected ion at m/z 716 from the cell lysate did not yield a similar fragmentation pattern to that of the standard 16:0/18:1–PE as shown in Fig S2(b).20 Based on this comparison, we concluded that this unknown m/z 716 was not the 16:0/18:1–PE molecule. On the other hand, in Fig. 2(a), the precursor ion of m/z 885 was selected from the cell lysate and its PSD spectrum was obtained. There were four distinct peaks from the selected peak at m/z 885, which were assigned to [M−H–R2′CH=C=O] and [M−H–R2′CO2H] at m/z 599 and 581, respectively, and also m/z 437 and 419 due to the loss of the inositol head group from both ions. The m/z 303 and 283 ions are not shown in the lysate sample due to the low quantity of the PI in the sample compared to the pure compound.

For the positive PSD, the K value obtained by using cholesterol as a positive calibrant was 5.7161 × 10−4 amu1/2/ns1 (standard deviation: 2.4327 × 10−7amu1/2 ns−1). As a model system, PC was studied since it is the most common lipid in the cell membrane and is easy to be ionized as a protonated form. The PSD spectrum of 16:0:18:1–PC standard (m/z 760) was first obtained and compared with that of an unknown peak at the same m/z 760 in lipids extracted from the MDA-MB-231 cell lysate.

As shown in Fig. 3(a), an unknown precursor ion m/z 760 produced a peak at m/z 184.0, which corresponds to the phosphocholine. This ion likely originated from the fragmentation process involving the participation of the α-hydrogens of the fatty acyl chain,25 which suggests that the precursor ion is a phosphatidylcholine molecule. To examine its exact mass and structure, a phosphatidylcholine having the same mass as that of the unknown precursor ion, PC 16:0/18:1 as a standard, was also investigated in positive PSD. A PSD of the standard lipid, PC 16:0/18:1 was conducted and the generated fragment ions are presented in Fig. 3(b). The fragment ions at m/z 496 ([M−H–R2′CH=C=O]) and m/z 522 ([M−H–R1′CH=C=O]) arose from losses of the fatty acyl at sn–2 and sn–1 as a ketene, respectively. In addition, the fragment ions at m/z 478 and m/z 504 arose from the elimination of fatty acid moieties to result in [M−H–R2′CO2H] and [M−H–R1′CO2H].25 As expected, these numerous fragment ions were the result of the energetic nature of the SIMS process since the SIMS-PSD spectrum resembled a MALDI-PSD spectrum obtained with increased collisional energy and laser fluence.26 

Fig. 3.

(Color online) (a) Reconstructed PSD spectrum of the selected ion m/z 760, taken from a sample of lipids extracted from the MDA-MB-231 cell lysate in the positive ion mode. The inset image has an adjusted y-scale to represent fragment ions from m/z 400 to 540. The inset chemical structures are m/z 522, m/z 496 and m/z 184. (b) Reflectron TOF spectra of the precursor ion of m/z 760 (phosphatidic choline 16:0/18:1) obtained from a commercial compound as a standard in the positive mode.

Fig. 3.

(Color online) (a) Reconstructed PSD spectrum of the selected ion m/z 760, taken from a sample of lipids extracted from the MDA-MB-231 cell lysate in the positive ion mode. The inset image has an adjusted y-scale to represent fragment ions from m/z 400 to 540. The inset chemical structures are m/z 522, m/z 496 and m/z 184. (b) Reflectron TOF spectra of the precursor ion of m/z 760 (phosphatidic choline 16:0/18:1) obtained from a commercial compound as a standard in the positive mode.

Close modal

We show that a negative metastable ion generated in the source region of a ToF-SIMS instrument can be measured to determine its structure based on its fragmentation pattern without any modifications in the TOF analyzer. PSD was used to identify unknown lipid molecules extracted from breast cancer cell lysate, MDA-MB-231. The unknown molecules were identified as PI 18:0/20:4 and PC 16:0/18:1 at negative and positive PSD, respectively. These analysis results make it possible to identify unknown molecular structures at both ion modes. Thus, PSD could be a useful method to identify unknown lipids from biological samples by using the ToF-SIMS technique.

This study was supported by the Development of Platform Technology for Innovative Medical Measurements Program (GP2016-0022) from the Korea Research Institute of Standards and Science, the Pioneer Research Center Program (NRF-2012-0009541), the Bio and Medical Technology Development Program (NRF-2015M3A9D7029894) and Global Frontier Project (H-GUARD_2013M3A6B2078962) of the National Research Foundation (NRF) funded by the Ministry of Science, ICT and Future Planning.

1.
P. V.
Escribá
 et al,
J. Cell Mol. Med.
12
,
829
(
2008
).
2.
Y. A.
Hannun
and
L. M.
Obeid
,
Nat. Rev. Mol. Cell Biol.
9
,
139
(
2008
).
3.
G.
Vereb
,
J.
Szöllösi
,
J.
Matkó
,
P.
Nagy
,
T.
Farkas
,
L.
Vígh
,
L.
Mátyus
, and
T. A.
Waldmann
,
Proc. Natl. Acad. Sci. U. S. A.
100
,
8053
(
2003
).
4.
R.
Zechner
,
R.
Zimmermann
,
T. O.
Eichmann
,
S. D.
Kohlwein
,
G.
Haemmerle
,
A.
Lass
, and
F.
Maedeo
,
Cell Metab.
15
,
279
(
2012
).
5.
B.
Fuchs
,
R.
Süß
,
K.
Teuber
,
M.
Eibisch
, and
J.
Schiller
,
J. Chromatogr. A
1218
,
2754
(
2011
).
6.
G.
Rouser
,
S.
Fleischer
, and
A.
Yamamoto
,
Lipids
5
,
494
(
1970
).
7.
J. J.
Myher
and
A.
Kulsis
,
J. Chromatogr. B
671
,
3
(
1995
).
8.
S.
Chen
,
G.
Kirschner
, and
P.
Traldi
,
Anal. Biochem.
191
,
100
(
1990
).
9.
S.
Milne
,
P.
Ivanova
,
J.
Forrester
, and
H. A.
Brown
,
Methods
39
,
92
(
2006
).
10.
M. R.
Wenk
 et al,
Nat. Biotechnol.
21
,
813
(
2003
).
11.
J.
Schiller
 et al,
Prog. Lipid Res.
43
,
449
(
2004
).
12.
B.
Fuchs
,
R.
Süß
, and
J.
Schiller
,
Prog. Lipid Res.
49
,
450
(
2010
).
13.
I.
Lanekoff
,
P.
Sjövall
, and
A. G.
Ewing
,
Anal. Chem.
83
,
5337
(
2011
).
14.
J.
Brison
,
M. A.
Robinson
,
D. S. W.
Benoit
,
S.
Muramoto
,
R. S.
Stayton
, and
D. G.
Castner
,
Anal. Chem.
85
,
10869
(
2013
).
15.
J. W.
Park
 et al,
Sci. Rep.
5
,
11077
(
2015
).
16.
K. A.
Al-Saad
,
W. F.
Siems
,
H. H.
Hill
,
V.
Zabrouskov
, and
N. R.
Knowles
,
J. Am. Soc. Mass Spectrom.
14
,
373
(
2003
).
17.
D.
Touboul
,
A.
Brunelle
, and
O.
Laprevote
,
Rapid Commun. Mass Spectrom.
20
,
703
(
2006
).
18.
M. L.
Doria
,
C. Z.
Cotrim
,
C.
Simoes
,
B.
Macedo
,
P.
Domingues
,
M. R.
Domingues
, and
L. A.
Helguero
,
J. Cell. Physiol.
228
,
457
(
2013
).
19.
J. I.
Zhang
,
N.
Talaty
,
A. B.
Costa
,
Y.
Xia
,
W. A.
Tao
,
R.
Bell
,
J. H.
Callahan
, and
R. G.
Cooks
,
Int. J. Mass Spectrom.
301
,
37
(
2011
).
20.
See supplementary material at http://dx.doi.org/10.1116/1.4944955 for checking the effect of the primary ion beams on the PSD spectra (Supplementary Materials 1) and the negative PSD in 16:0/18:1-PE model system and MDA-MB-231 cell extraction sample (Supplementary Materials 2).
21.
M. L.
Doria
,
W.
Cotrim
,
B.
Macedo
,
C.
Simoes
,
P.
Domingues
,
L.
Helguero
, and
M. R.
Domingues
,
Breast Cancer Res. Treat.
133
,
635
(
2012
).
22.
F.-F.
Hsu
and
J.
Turk
,
J. Chromatogr. B
877
,
2673
(
2009
).
23.
F.-F.
Hsu
and
J.
Turk
,
J. Am. Soc. Mass Spectrom.
11
,
986
(
2000
).
24.
A. S.
Woods
,
H.-Y. J.
Wang
, and
S. N.
Jackson
,
Curr. Pharm. Des.
13
,
3344
(
2007
).
25.
F.-F.
Hsu
and
J.
Turk
,
J. Am. Soc. Mass Spectrom.
14
,
352
(
2003
).
26.
C. D.
Calvano
and
C. G.
Zambonin
,
Adv. Dairy Res.
1
,
101
(
2013
).

Supplementary Material