Tài liệu Báo cáo khoa học: Characterization of 1H NMR detectable mobile lipids in cells from human adenocarcinomas doc

Characterization of
1
H NMR detectable mobile lipids
in cells from human adenocarcinomas
Anna Maria Luciani
1
, Sveva Grande
1
, Alessandra Palma
1
, Antonella Rosi
1
, Claudio Giovannini
2
,
Orazio Sapora
3
, Vincenza Viti
1
and Laura Guidoni
1
1 Dipartimento di Tecnologie e Salute and INFN Gruppo Collegato Sanita
`
, Istituto Superiore di Sanita
`
, Rome, Italy
2 Dipartimento di Sanita
`
Pubblica Veterinaria e Sicurezza Alimentare, Istituto Superiore di Sanita
`
, Rome, Italy

3 Dipartimento di Ambiente e Connessa Prevenzione Primaria, Istituto Superiore di Sanita
`
, Rome, Italy
Among the different molecules showing intense and
narrow peaks in the
1
H magnetic resonance spectra,
much attention has been devoted to the fatty acid sig-
nals from mobile lipids (MLs), which are characterized
by high mobility and, thus, differently from most cell
lipids, are visible in high resolution magnetic resonance
spectra. High-intensity MLs are often observed in pro-
liferating cells and in tumour cells [1–4]. Many studies
have found that the onset of apoptosis is accompanied
by an increase in ML intensity [5–7], although other
studies have not [8,9].
A number of studies have been performed in view of
the possible use of MLs as spectroscopic markers of
cell fate, although a clear explanation of their behav-
iour has not yet been provided. Essentially, two differ-
ent localizations have been proposed. In mammalian
cells, ML resonances arise from lipids that are either
present as microdomains with high mobility embedded
in the plasma membrane bilayer [9] or exist in cytosolic
lipid droplets, mostly consisting of triglycerides (TGs)
[10,11]. Some studies have found that the concentra-
tion of total cell TGs was consistent with the intensity
of ML signals [11], whereas, more recently, changes in
the size of lipid droplets were suggested [12].
In the present study, we examined the

1
H NMR ML
signals of cultured tumour cells, specifically HeLa and
Keywords
cell cycle; cell metabolism; lipids; magnetic
resonance spectroscopy; tumour cell lines
Correspondence
L. Guidoni, Dipartimento di Tecnologie e
Salute, Istituto Superiore di Sanita
`
, 00161
Rome, Italy
Fax: +39 06 4938 7075
Tel: +39 06 4990 2804
E-mail:
(Received 5 November 2008, revised 15
December 2008, accepted 22 December
2008)
doi:10.1111/j.1742-4658.2009.06869.x
Magnetic resonance spectroscopy studies are often carried out to provide
metabolic information on tumour cell metabolism, aiming for increased
knowledge for use in anti-cancer treatments. Accordingly, the presence of
intense lipid signals in tumour cells has been the subject of many studies
aiming to obtain further insight on the reaction of cancer cells to external
agents that eventually cause cell death. The present study explored the rela-
tionship between changes in neutral lipid signals during cell growth and
after irradiation with gamma rays to provide arrest in cell cycle and cell
death. Two cell lines from human tumours were used that were differently
prone to apoptosis following irradiation. A sub-G1 peak was present only
in the radiosensitive HeLa cells. Different patterns of neutral lipids changes

were observed in spectra from intact cells, either during unperturbed cell
growth in culture or after radiation-induced growth arrest. The intensities
of triglyceride signals in the spectra from extracted total lipids changed
concurrently. The increase in lipid peak intensities did not correlate with
the apoptotic fate. Modelling to fit the experimental data revealed a
dynamic equilibrium between the production and depletion of neutral
lipids. This is observed for the first time in cells that are different from
adipocytes.
Abbreviations
GPC, glycerophosphorylcholine; IR, intensity ratio; ML, mobile lipid; PC, phosporylcholine; PCA, perchloric acid; PL, phospholipids;
TG, triglycerides.
FEBS Journal 276 (2009) 1333–1346 ª 2009 The Authors Journal compilation ª 2009 FEBS 1333
MCF-7 cells from human cancers. In a previous study
[13], we demonstrated that these cells display intensity
modulation of ML signals as a periodic event accom-
panying cell growth. In a recent study [14], we also
showed that these cells are differently prone to apopto-
sis induced by treatment with gamma rays. We associ-
ated the radiation-induced apoptosis of tumour cells
with the level of reduced glutathione, as detected by
1
H NMR [14].
These cells were therefore considered for use in mag-
netic resonance spectroscopy analysis for the detection
of possible different trends in ML spectral features
during cell growth in culture and after radiation-
induced arrest in proliferation. Within this framework,
a similar modulation of ML signals with growth was
observed in both cell lines, whereas radiation-induced
arrest in proliferation capacity resulted in a different

pattern. Effects on cell cycle frequency were also
observed. ML signal intensity modulation was tenta-
tively related to modulation of lipid metabolism.
Results
Cell spectra were first examined for signal quantifica-
tion after spectral assignment. Subsequently, changes
in signal modulation were monitored when cell growth
was arrested by means of treatment with ionizing radi-
ation. Finally, a model to fit signal intensity modula-
tion was proposed.
Analysis of
1
H NMR spectra – spectral
assignments and quantification
Both cell lines displayed very similar spectral features.
Besides other signals, the characteristic peaks from
MLs were observed to be in agreement with the data
available in the literature for cancer cells [1–4]. Under
the present experimental conditions, the intense ML
signals can be attributed to the fatty acid chains of
neutral lipids, mostly TGs, in agreement with the data
available in the literature [1,15] and on the basis of
our previous observations showing that, in these cells,
TG peaks are more intense in the lipid extract spectra
derived from cells with high ML signals compared to
spectra with low ML signals [13]. This point will be
discussed further below.
Figure 1 shows an example of the
1
H NMR spectra

from MCF-7 cells at day 3 (Fig. 1A) and at day 6
(Fig. 1A¢) after seeding. It is worth noting that the
choline-based signals at 3.2 p.p.m. display the same
intensity with very different ML signals (Fig. 1A,A¢).
When the ML signals were intense in the 1D spectra
(p.p.m.)
(p.p.m.)
(p.p.m.)
(p.p.m.)
(p.p.m.)
M1 + ML1
A
A′
B′
B
Fig. 1. Spectra of MCF-7 cells in different proliferative conditions: 1D
1
H NMR spectra from cells at day 3 after seeding (A) and at day 6
after seeding (A¢). ML signals are at 0.89 p.p.m. (ML1), 1.28 p.p.m. (ML2) and 1.55 p.p.m. (ML3); M1 and M2 from macromolecules are also
labelled. 2D
1
H NMR COSY spectra of the same cell samples at day 3 after seeding (B) and at day 6 after seeding (B¢). Cross peak A is from
terminal methyl and bulk methylene coupling in ML. The reference cross peak from Lys is also labelled. Label T refers to the glycerol cross
peak of triglycerides. The insert shows the glycerol cross peak region of the geminal protons of triglycerides.
1
H NMR of mobile lipids in tumour cells A. M. Luciani et al.
1334 FEBS Journal 276 (2009) 1333–1346 ª 2009 The Authors Journal compilation ª 2009 FEBS
(Fig. 1A), the corresponding 2D COSY spectra were
also characterized by prominent cross peaks from fatty
acid chains (Fig. 1B), in agreement with the data avail-

able in the literature [15] and previous observations
[13]. The cross peak at 4.07–4.24 p.p.m., generated by
protons in the glycerol backbone of TG, was also visi-
ble only in ML rich spectra (T). Details with respect to
this signal are provided in Fig. 1B (insert), which
shows the characteristic cross peak from the geminal
protons of carbons 1 and 3 of glycerol in TG [1].
Cross peaks of lipids, including the T signal, were
absent in cells characterized by low ML signals in 1D
spectra (Fig. 1B¢). Very similar behaviour was
observed in HeLa cells according to previously
reported data [4,13].
Signal assignments in cell spectra were performed
after comparison with the spectra from lipid and per-
chloric acid (PCA) extracts, derived from cell samples
grown and harvested under similar conditions, and
with compound spectra. Assignments from the litera-
ture were also taken into account [1–4,16]. Figure 2
shows typical spectra (1D and 2D COSY) from lipid
and PCA extracts both relative to a MCF-7 cell
sample with high MLs. It is worth noting that, in the
2D COSY spectra from extracted lipids, the glycerol
geminal cross peaks of 1,3-glycerol protons of TG and
of 1-glycerol protons for phospholipids (PL) are clearly
separated, in agreement with the data available in the
literature [1].
For peak assignments and intensity quantification,
deconvolution of 1D spectra and integration of 2D
cross peaks was performed as described in the Experi-
mental procedures. The signal intensity refers to the

peak area in the 1D spectra and to the cross peak inte-
gral in the 2D spectra. Internal reference signals were
used. Our study compared samples where cell volumes
and cell packing change, which hinders the use of an
external reference.
The signal at 0.96 p.p.m., present in cells (peak M2
in Fig. 1) and PCA extract spectra, derives from poly-
peptide chains and was used as the intensity reference
for 1D spectra because it is indicative of cell mass. As
far as the intensity reference for 2D spectra is con-
cerned, the sum of peaks of Lys at 1.70–3.00 p.p.m.
and Ala at 1.48–3.77 p.p.m. (Fig. 1B) was chosen as
the area reference in the 2D COSY spectra.
(p.p.m.)
(p.p.m.)
(p.p.m.)
(p.p.m.)
(p.p.m.)
(p.p.m.)
A
A′
B
B′
Fig. 2. 1D
1
H NMR spectra of extracted lipids (A) and PCA extracts (A¢) from one representative sample of MCF-7 cells; 2D
1
H NMR spectra
of extracted lipids (B) and PCA extracts (B¢) from one representative sample of MCF-7 cells. The insert shows the glycerol cross peak region
of the geminal protons of TGs and PLs.

A. M. Luciani et al.
1
H NMR of mobile lipids in tumour cells
FEBS Journal 276 (2009) 1333–1346 ª 2009 The Authors Journal compilation ª 2009 FEBS 1335
For 1D spectra, deconvolution of spectra of the type
shown in Fig. 1A¢, with ML signals of a very low
intensity, was performed as a first step. Deconvolution
of spectra of the type shown in Fig. 1A, with intense
ML signals, was then performed starting from the lines
and parameters previously found, by adding the signals
from MLs. A typical deconvolution pattern with the
used resonances is shown in Fig. 3A. The correspond-
ing parameters are given in Table 1. Experiments con-
ducted on different samples derived from the same
culture (at least three samples) demonstrated that the
SD from this procedure did not exceed 0.01 p.p.m. for
chemical shifts and 10% for linewidths and intensity
ratios (IRs). On the other hand, the variability of
signal intensities, especially for MLs, exceeded the
measurement error in spectra from samples derived
from different culture, even when cells were harvested
under similar growth conditions. For this reason, the
spectral behaviour over time was compared among
samples obtained by cells harvested at different days
from the same seeding.
Cholesterol peaks were present in lipid extracts at
0.70 and 1.03 p.p.m. in the 1D spectra (Fig. 2A) and
with the typical cross peak at 0.87–1.50 p.p.m. in the
2D spectra (Fig. 2B). In spectra from cell samples, we
could therefore assign the peak at 0.71 p.p.m. (Fig. 3A

and Table 1) to the methyl group in C18 of choles-
terol. This signal was present in all cell spectra,
although it was broader and more intense when ML
signals were present, and its IR changed from 0.07 to
0.36 as obtained by 1D spectra deconvolution. In par-
allel, peaks at 1.03 and 1.50 p.p.m., as obtained from
1D spectra deconvolution (Fig. 3A and Table 1),
increased, and the cross peak at 0.87–1.50 p.p.m.
appeared in the cell spectra. This latter feature is
evident in Fig. 1A¢,B¢. The peak at 1.03 p.p.m. could
therefore be attributed to the methyl group in C19 and
(p.p.m.)
(p.p.m.)
(p.p.m.)
A
B
Fig. 3. Example of the deconvolution pattern of the methylene
region in the 1D
1
H NMR spectra of MCF-7 cells (A); integration
regions for selected cross peaks in the 2D
1
H NMR spectra of
MCF-7 cells (B) (Lys1, lysine; A, bulk methylene and terminal
methyl group in fatty acids; E, C4 methylene and C3 methylene in
fatty acids; Lino, linolenic acid). Other ML related cross peaks are
visible as peaks B and F; for assignments, see [2]. Rectangles indi-
cate the area used for signal integration.
Table 1. Mean values of parameters d (p.p.m.), Dm (Hz) and IR,
after deconvolution (1D) and integration (2D COSY) of spectra, such

as those presented in Fig. 3. Values were obtained from the spec-
tra of three different samples derived from the same culture. The
standard deviation was 0.01 p.p.m. for chemical shift values and
10% for linewidths and IR. Chemical shifts are referring to lactate
methyl; IR are calculated with respect to M2 in 1D and to
Lys1 + Ala in the 2D spectra.
1D d (p.p.m.) Dm (Hz) IR
Chol 0.71 40 0.15
ML1 + M1 + Chol 0.89 24 0.16
M2 0.95 28 1.00
Chol 1.03 20 0.05
M3 1.22 46 0.14
ML2 1.28 17 3.26
ML3 1.31 23 2.70
LAC 1 1.32 8.3 0.58
LAC 2 1.34 7.1 0.67
M5 1.40 36 1.30
M6 + Chol 1.48 22 0.13
ML4 1.58 38 0.89
M7 1.70 39 0.55
Broad 2.20 870 3.10
2D d (p.p.m.) IR
Lys1 1.70–3.00 1.00
Lys2 1.46–1.67 0.28
A 0.89–1.27 4.50
E 1.31–1.55 2.20
B 1.34–2.00 2.70
F 1.58–2.22 2.47
Lino 0.96–2.03 0.72
Chol 0.87–1.51 0.24

1
H NMR of mobile lipids in tumour cells A. M. Luciani et al.
1336 FEBS Journal 276 (2009) 1333–1346 ª 2009 The Authors Journal compilation ª 2009 FEBS
the peak at 1.50 p.p.m. to the other bulk protons of
cholesterol.
As far as MLs are concerned, spectra with intense
ML signals could be fitted by adding a peak at
1.28 p.p.m., a second peak at 1.31 p.p.m. and a peak
at 1.58 p.p.m. (Table 1 and Fig. 3A) to the spectral
fitting of samples without ML signals. The presence of
these peaks was paralleled by the existence of lipid
cross peaks in the 2D spectra (Table 1 and Fig. 3B).
The cross peak A, due to the interaction of terminal
methyl group peak at 0.89 p.p.m. and the proximal
methylene at 1.28 p.p.m., was used to quantify MLs
because it is representative of the corresponding bulk
fatty acids chains. This excludes the contribution from
x-3 fatty acids, where methyl protons at 0.98 p.p.m.
are coupled to the allylic methylene at 2.09 p.p.m. [16].
The signal at 0.88 p.p.m. is mostly from the terminal
CH
3
of ML chains, with a minor contribution from
the cholesterol C25 and C26 methyl groups, and from
an unidentified macromolecule methyl group (M1).
This signal is present: (a) in PCA extracts (Fig. 2A¢)
and (B) in cells when ML signals are absent
(Fig. 1A,B). The same considerations hold for the
peaks at 1.22 and 1.40 p.p.m. (Table 1 and Figs 2 and
3). These signals most likely arise from the aggregation

of large molecules because they are characterized by
large linewidths (Table 1 and Figs 2 and 3). Work is in
progress to clarify the nature of these structures.
Signal intensities were also measured in the 2D spec-
tra. A typical 2D COSY spectrum is shown in Fig. 3b,
including the details of the cross peaks examined. The
peaks chosen for evaluation are framed with rectangles
that denote the areas used for volume integration.
When very intense ML signals are present in the spec-
tra, the signals related to unsaturated fatty acids also
are evident, and are more intense in MCF-7 than in
HeLa cell samples. Besides the cross peaks resulting
from the connectivity of the vinyl protons (at
5.35 p.p.m.) to the allylic protons (at 2.05 p.p.m.) in
monounsaturated chains and to the bis-allylic protons
(at 2.80 p.p.m.) in polyunsaturated fatty acids (not
shown), the cross peaks at 1.64–2.09 p.p.m. and 1.68–
2.24 p.p.m., attributed to arachidonic acid chains, and
at 0.93–2.04 p.p.m., attributed to linolenic acid chains
on the basis of a comparison with lipid extracts and
from the data available in the literature, are also
clearly visible in Fig. 1B,B¢.
Experiments on different samples derived from the
same culture (at least three samples) were also exam-
ined to assess measurement errors in the 2D cross peak
integration. Under these conditions, errors on cross
peak volumes did not exceed 10%, whereas the vari-
ability of integral values exceeded this error in the
spectra from samples derived from different cultures,
even when cells were harvested under similar growth

conditions. For this reason, the behaviour over time of
2D COSY cell spectra were compared in samples
obtained by cells harvested at different days from the
same seeding.
In the following, ML quantification derives from
intensity measurements of the peak at 1.28 p.p.m. in
1D spectra and from the integral of cross peak A at
0.89–1.28 p.p.m Cross peaks from PL glycerols
(Fig. 2B, insert) were never observed in the 2D COSY
spectra of cells, even in the presence of very high ML
signals in the 1D spectra. Cross peaks from glycerol
protons of TG in cells were not routinely used for TG
quantification because the intensities were much smal-
ler and the errors were larger. ML signals were then
used to monitor TG levels in cells.
Biological changes and ML signal modulation
with growth and in growth-arrested cells
Changes in ML signals were monitored in parallel
with cell growth and after cell cycle arrest due to
irradiation.
Cell proliferation and cell cycle
Cell growth behaviour was examined in HeLa and
MCF-7 cells. Cells were routinely grown as described
in the Experimental procedures. Cells were sampled at
different days after seeding for both NMR experiments
and cell cycle measurements. Under the chosen experi-
mental conditions, cells were kept in the exponential
growth phase (up to 3 days from seeding). Cells were
then irradiated with a single dose of 20 Gy (gamma
irradiation) to provide growth arrest and cell death

with different characteristics in the two cell lines,
according to previous observations [14]. Figure 4
shows the cell counts as a function of time for one
representative experiment in HeLa and MCF-7 cells.
Compared to control samples, the differences in cell
counts were larger in HeLa than in MCF-7 cells.
Similar behaviour was observed in at least three
independent experiments.
To assess cellular transcriptional responses to radia-
tion-induced DNA damage, we examined cell cycle
arrest in MCF-7 and HeLa cells at 1, 2 and 3 days
after treatment. Both cell lines underwent cycle arrest
upon irradiation, with different characteristics.
Figure 4A¢,B¢ shows the percentage of cell phases of
both cell lines observed after 2 days after irradiation at
20 Gy. Although both cells were blocked in G2 ⁄ M,
HeLa cells displayed a remarkable decrease in the
A. M. Luciani et al.
1
H NMR of mobile lipids in tumour cells
FEBS Journal 276 (2009) 1333–1346 ª 2009 The Authors Journal compilation ª 2009 FEBS 1337
G1 phase, whereas MCF-7 cells showed G1 block and
a decreased percentage in the S phase compared to
control samples. Irradiated HeLa cells showed an
intense sub-G1 peak (> 20%), indicating DNA frag-
mentation and the occurrence of significant apoptosis.
This observation is in agreement with previous data
obtained in the same cells by monitoring apoptosis as
the externalization of phosphatidyserine [14].
1

H NMR ML signals in intact cells and TG signals in
lipid extracts
To clarify whether intensity changes of ML signals
were mainly related to changes in TG concentration,
to differences in chain mobility due to structural
changes, or to the different size of droplets, as recently
suggested by Quintero et al. [12], we compared the
behaviour of the spectra of cells and the total lipid
extracted from cells. Irradiation was then used to
arrest cell growth, thus providing a modification of
ML signal intensity in cells. Lipids were extracted from
cell samples under identical conditions.
In previous experiments on these cell lines [4], we
observed an intensity modulation of ML with cell
growth. Consequently, the extent of the variation in
intensity with irradiation was monitored after different
time intervals after irradiation. Figure 5A,B shows the
ML signal intensities of 1D spectra from the two
cell lines run at different days. Samples of cells grown
under the same conditions were irradiated and the two
sets of spectra compared (control and treated samples).
Similar data were obtained for 2D measurements
(Fig. 5A¢,B¢), where the data are from one representa-
tive experiment. Error bars indicate measurement
errors. By comparing the results of at least seven inde-
pendent experiments, the differences between control
and irradiated samples were significant for MCF-7
cells at all time intervals examined. On the other hand,
statistical significance for HeLa samples was
observed at days 2 and 3 after irradiation. Figure 5C

shows these differences for samples examined 2 days
after irradiation.
Figure 6 shows the glycerol region from two repre-
sentative 1D spectra of total lipids extracted from
control and irradiated HeLa (Fig. 6A,A¢) and MCF-7
(Fig. 6B,B¢) cells. A significant change in relative inten-
sity of TGs (glycerol proton centred at 4.32 p.p.m.)
versus PLs (glycerol proton centred at 4.42 p.p.m.)
was found in irradiated samples compared to controls.
Particularly, TG signals were depressed in HeLa cells
(Fig. 6A,A¢), whereas an increase was evident in
MCF-7 cells (Fig. 6B,B¢). Deconvolution of 1D spectra
was performed to provide the relative intensities of TG
versus PL, which were calculated on sn-1 and sn-3
glycerol signals centred at 4.32 p.p.m. for TG and sn-1
glycerol signals at 4.42 p.p.m. for PL (Fig. 6C). In this
experiment, the calculated relative concentration
of TG versus TG + PL was 15% and 12% in MCF-7
A
A′
B
B′
Fig. 4. Number of HeLa (A) and MCF-7 (B) cells (N) as a function of time after irradiation for both control (h) and irradiated ( ) samples.
The solid black line is the fit with an exponential function. Percentage of MCF-7 (A¢) and HeLa (B¢) cells in the different cell cycle phases,
measured 2 days after irradiation. One representative experiment is reported for both control and irradiated samples.
1
H NMR of mobile lipids in tumour cells A. M. Luciani et al.
1338 FEBS Journal 276 (2009) 1333–1346 ª 2009 The Authors Journal compilation ª 2009 FEBS
and HeLa cells, becoming 20% and 8%, respectively,
after irradiation. The standard deviation in repeated

calculations was 2%. The relative concentrations of
TG calculated by
1
H NMR were found to be in agree-
ment with that reported in previous studies [17].
Intensity measurements of 1D signals of sn-2
glycerol protons of TG at 5.29 p.p.m. and PL at
5.25 p.p.m., clearly resolved only at 700 MHz,
gave similar results (not shown). This behaviour is
similar to that observed for ML signals in whole
cells (Fig. 5).
Total extracted TG may be more abundant with
respect to NMR visible TG in cells. For this reason,
the spectra from extracted lipids were not used to
calculate NMR visible TG in intact cells.
Lipid metabolites from PCA extracts
To provide further information on lipid metabolism,
the PCA extracts were also analyzed. The region of
choline metabolites around 3.2 p.p.m. is shown in
Fig. 7 for PCA extracts of HeLa. Mean values of
parameters d (p.p.m.), Dm (Hz) and IR were obtained
from deconvolution of the 1D spectra (three different
samples derived from the same culture) of PCA
extracts and are reported in Table 2. The standard
deviation was 0.005 p.p.m. for chemical shift values
and 10% for linewidths and IR.
Signals from the headgroups of glycerophosphoryl-
choline (GPC) at 3.24 p.p.m., phosporylcholine (PC)
at 3.23 p.p.m. and choline at 3.21 p.p.m. showed dif-
ferent behaviour in the two cell lines after irradiation.

In particular, more relevant changes in GPC ⁄ PC ratios
were observed in irradiated MCF-7 cells (Fig. 7B,B¢)
with respect to HeLa cells (Fig. 7A,A¢), indicating the
different equilibrium of catabolism versus anabolism.
Table 3 reports on the intensity of changes of the
choline-based metabolites for the two cell lines after
irradiation resulting from fittings of at least three
different spectra.
Time (day)
Time (day)
Time (day)
Time (day)
A/(Lys + Ala) c
A/(Lys + Ala) i
A
B
A′
B′
C
Fig. 5. Intensity modulation of ML signals
from a representative experiment (1D and
2D
1
H NMR data) for HeLa (A, A¢) and
MCF-7 cells (B, B¢). Spectra were acquired
at different days from seeding for both
control (h) and irradiated (
) samples
(D = 20 Gy). Errors obtained from spectral
fitting (1D) and integration (2D COSY) are

contained within the symbols. (C) Relative
IRs ML ⁄ M (control: white; irradiated: black)
and A ⁄ (Lys + Ala) (controls: dotted white;
irradiated: dotted black) as obtained from 1D
and 2D COSY spectra of HeLa and MCF-7
cell samples. Data are the mean ± SD
values of seven independent experiments.
Spectra were acquired on day 5 after seed-
ing and 2 days after irradiation with a single
dose of 20 Gy. *P < 0.05 (t-test).
A. M. Luciani et al.
1
H NMR of mobile lipids in tumour cells
FEBS Journal 276 (2009) 1333–1346 ª 2009 The Authors Journal compilation ª 2009 FEBS 1339
Model for the ML signals
To find a possible explanation for the experimental
data, a model for ML intensity modulation is pro-
posed.
As previously reported [4,13] growth of MCF-7 and
HeLa cells slows down as cells approach confluence.
This finding is in agreement with the observed increase
of the G1 phase in the final days in culture (Fig. 4
A¢,B¢). It is reasonable to assume that cell metabolism,
including PL synthesis, slows down accordingly. On
the other hand, intensities of ML (in cells) signals,
mainly due to TG according to a previous study [13]
as well as the present study (compare inserts in Figs 1
and 2), are characterized by a nonlinear behaviour
over time in culture (Fig. 5). A mechanism that is
more complicated than a simple decrease of lipid

production over time must be therefore envisaged.
There is a growing body of evidence indicating that
lipid metabolism possesses an articulated role with
respect to maintaining cell equilibrium, which takes
into consideration both PL synthesis ⁄ breakdown and
TG metabolism [18–20]. We may infer that there are
two mechanisms inside the cell: one relative to the pro-
duction of ML (and TG) with a rate constant R
p
and
one relative to the consumption of ML (and TG) with
a rate constant R
c
. The signal that we observe in the
NMR spectra is due to the net accumulation of reserve
lipids and its rate, dML ⁄ dt, is given by the difference
of the rate of lipid production R
p
and the rate of lipid
consumption R
c
:
dML=dt ¼R
p
ÀR
c
ð1Þ
We may assume that both rates R
p
and R

c
are not
constant, but decrease over time in culture as a conse-
quence of cell proliferation slowing down. A linear
dependence of both rates can be assumed:
R
p
¼p
1
Àp
2
t ð2Þ
R
c
¼c
1
Àc
2
t ð3Þ
where p
1
and c
1
are the production and consumption
rates at time = 0, respectively, and p
2
and c
2
are
the changes of production and consumption rates over

time.
By integrating Eqn (1), we obtain a second degree
polynomial function for the lipid accumulation:
MLðtÞ¼m
1
t
2
þm
2
tþm
3
ð4Þ
where m
1
=(c
2
) p
2
) ⁄ 2, m
2
=(p
1
) c
1
) and therefore
are related to the equilibrium between lipid production
and consumption, and m
3
is the starting ML value.
The best fit with Eqn (4) of data ML ⁄ M and

A ⁄ (Lys + Ala) versus time from independent experi-
ments on HeLa and MCF-7 cells gave a chi-square
value that was always < 1 for both the 1D and 2D
COSY experiments. Figure 5A,A¢,B,B¢ shows these
fittings for a representative experiment for both cell
lines. Figure 8 reports the parameters m
1
, m
2
and m
3
(mean ± SD) as obtained from fittings with Eqn (4)
for the data obtained from ten samples for HeLa and
ten samples for MCF-7 of cells harvested irrespective
of the growth phase.
(p.p.m.)
(p.p.m.)
(p.p.m.)
A′
B′
A
B
C
Fig. 6. Glycerol region of representative 1D
1
H NMR spectra from
total lipids extracted from HeLa (A, A¢) and MCF-7 cells (B, B¢).
Lower traces: control samples (A, B), upper traces: irradiated
(D = 20 Gy) samples (A¢,B¢). Spectra were acquired 2 days after
irradiation. Example deconvolution pattern (C) of 1D

1
H NMR
spectrum for sn-2 glycerol protons of TGs (couple of doublets
centered at 4.32 p.p.m., coupling 12 Hz and 4 Hz) and PLs (couple
of doublets centered at 4.42 p.p.m., coupling 12 and 3 Hz). Accord-
ing to this deconvolution pattern, in this spectrum, the ratio
TG ⁄ (PL + TG) was 0.34.
1
H NMR of mobile lipids in tumour cells A. M. Luciani et al.
1340 FEBS Journal 276 (2009) 1333–1346 ª 2009 The Authors Journal compilation ª 2009 FEBS
The minimum value t
m
of the parabolic fitting curve,
representing the time value for which production and
consumption rates are equal, and the corresponding
lipid value ML
m
are also reported.
1D and 2D data were in good agreement, although
the starting intensity values and values at the minimum
were different, reflecting different pools of ML inside
the cells. By changing the seeding density, the minimum
of the curve was shifted, but the shape of the curves
did not change. It is worth noting that m
1
was always
positive (i.e. c
2
was always greater than p
2

), whereas m
2
was always negative (i.e. c
1
was always greater than p
1
).
For m
1
> 0, the parabola was concave upward. For
m
2
< 0, the minimum was after time zero.
(p.p.m.)
(p.p.m.)
(p.p.m.)(p.p.m.)
A′ B′
AB
C
C′
Fig. 7. 1D
1
H NMR spectra of the choline-
related metabolites region from PCA
extracts of HeLa (A, A¢) and MCF-7 cells
(B, B¢). Lower traces: control samples
(A, B), upper traces (A¢,B¢): irradiated
samples (D = 20 Gy). Spectra were acquired
2 days after irradiation. Deconvolution
pattern of the same region (C) and of the

reference signal (C¢).
Table 2. Mean values of parameters d (p.p.m.), Dm (Hz) and IR
after deconvolution of 1D spectra from PCA extracts of HeLa cells
(Fig. 7C,C¢). Values were obtained from the spectra of three differ-
ent samples derived from the same culture. The standard deviation
was 0.005 p.p.m. for chemical shift values and 10% for linewidths
and IR.
D (p.p.m.) Dm (Hz) IR
M (reference) 0.939 25.0 1.00
Choline 3.219 1.88 0.45
PC 3.226 2.20 0.70
GPC 3.236 1.65 0.53
Table 3. Mean values (three independent experiments) of IRs of
the choline-related metabolites from PCA extracts of HeLa and
MCF-7 cells for control and irradiated samples. The standard devia-
tion was 10%.
GPC PC Cho
HeLacells
Controlsample 0.43 0.71 0.50
Irradiated sample 0.28 0.86 0.72
MCF-7 cells
Control sample 1.12 1.31 0.60
Irradiated sample 0.90 0.77 0.63
A. M. Luciani et al.
1
H NMR of mobile lipids in tumour cells
FEBS Journal 276 (2009) 1333–1346 ª 2009 The Authors Journal compilation ª 2009 FEBS 1341
Although the t-test showed that differences were not
significant between the two groups (HeLa and MCF-
7), m

1
values were generally higher in MCF-7 than in
HeLa, thus reflecting the tendency for higher final
values of ML signal intensities in MCF-7 cells (Fig. 8).
Cells were then irradiated to arrest cell growth; data
from irradiated cells could be still fitted through Eqn
(4), as shown in Fig. 5. Figure 9 shows the parameters
(mean ± SD) obtained by fitting the data from five
independent experiments on HeLa cells and five inde-
pendent experiments on MCF-7 cells (both irradiated
and non-irradiated cells). The parameter m
1
decreased
and m
2
increased after irradiation in both cell lines
(Fig. 9), but only the m
2
increase in MCF-7 cells was
statistically significant in independent experiments
(P < 0.05; t-test). This may be due either to an
increase in c
1
(consumption rate at time = 0) or a
decrease in p
1
(production rate at time = 0). The
relevant m
2
increase in MCF-7 cells produced the

great increase of ML signals with respect to controls
(Fig. 5B,B¢). In some experiments conducted on
MCF-7 cells, the minimum of the parabolic curve at
time > 0 was no more evident and m
2
became posi-
tive. Finally, t
m
shifted to lower values in MCF-7 cells
and the ML
m
value was considerably higher in
irradiated MCF-7 cells compared to controls.
Discussion
The appearance of intense signals from bulk methylene
of fatty acid chains in high resolution
1
H NMR spec-
tra of cells has been studied subsequent to the first
observations being made in cancer cells, lymphocytes
and developing cells. On the other hand, a correlation
of the intensity of these signals with metabolic parame-
ters is not straightforward. Some studies noted that
the signal intensity of bulk methylene is influenced by
cell proliferation, as T lymphocyte activation [21] and
in tumour cells by the different proliferation state
[4,22]; other studies found that the onset of apoptosis
correlates with the increase of lipid signals, whereas
others did not [5–9]. Finally, some studies found that
these signals can be affected by extreme pH conditions,

which is more likely due to the effects of low pH on
cell proliferation [22].
In the present study, ionizing radiation was used to
affect cell growth and induce cell death in cells show-
ing different attitudes with respect to undergoing radi-
ation-induced apoptosis. The relevant sub-G1 peak
observed in the cell cycle profile of radiation-arrested
HeLa cells (Fig. 4A¢) points to significant apoptosis, in
agreement with the previously observed radiation-
induced apoptosis determined by phosphatidylserine
5.0
0.0
–5.0
–
10.0
5.0
0.0
–5.0
Fig. 8. Parameters m
1
, m
2
and m
3
obtained
from fitting with Eqn (4) ML data, as in
Fig. 5. The minimum values (t
m
) of the para-
bolic fitting curve and the corresponding

lipid values (ML
m
) are also reported. Bars
represent the mean ± SD of ten indepen-
dent experiments for each cell line.
5
0
–5
–
10
5
0
–5
5
0
–5
5
0
–5
–
10
Fig. 9. Parameters m
1
, m
2
and m
3
obtained
from fitting with Eqn (4) ML data, as in
Fig. 5, for irradiated (I) and non-irradiated (C)

HeLa and MCF-7 cells. The minimum values
(t
m
) of the parabolic fitting curve and the
corresponding lipid values (ML
m
) are also
reported. Bars represent the mean ± SD of
five independent experiments on each cell
line. *P < 0.05 (t-test).
1
H NMR of mobile lipids in tumour cells A. M. Luciani et al.
1342 FEBS Journal 276 (2009) 1333–1346 ª 2009 The Authors Journal compilation ª 2009 FEBS
externalization [14]. Similar results were observed
in different studies [23,24]. Irradiated MCF-7 cells
showed very little apoptosis, according to previous
observations [14], and an insignificant G1 peak.
Ionizing radiation, similar to the effect of cytotoxic
drugs, induces DNA damage (either directly or indi-
rectly), causing proliferating cells to undergo cell cycle
arrest, which is assumed to provide an opportunity for
DNA repair. Irradiated cells may arrest either in the
G1 phase, due to the cell checkpoint mechanism under
the control of p53 protein, or in G2. After a time delay
that depends on the extent of the damage, cells may
enter mitosis, resulting in normal cell duplication, cell
mutation or cell death. Inadequate DNA repair often
results in apoptosis.
Lack of functional p53 in HeLa cells fails to induce
cell cycle arrest in G1. Cells can only proceed towards

the S and G2 phases while trying to repair the dam-
aged DNA, with a subsequent block in the G2 phase,
at the expense of G1 (Fig. 4A¢ ). On the other hand,
HeLa cells undergo apoptosis after irradiation, accord-
ing to that previously reported for other cells lacking
p53 [23].
MCF-7 cells were found to be more resistant to irra-
diation (Fig. 4B), displaying an absent or a very low
apoptotic percentage compared to controls [14,24],
despite the presence of p53 and the subsequent block
in G1 (Fig. 4B¢). Cells are also arrested in G2, whereas
the S phase is markedly reduced (Fig. 4B¢). Cells are
unable to enter mitosis, although they escape an apop-
totic fate. This finding is in agreement with other stud-
ies that also found very low apoptosis and arrest in
G2 and G1 in these cells after irradiation, whereas the
same cells could undergo apoptotic death upon treat-
ment with daunorubicin [24] or buthionine sulphoxi-
mine [14,24].
The two cell lines showed similar behaviour with
respect to the intensity of variation in bulk methylene
signals (ML) when kept in culture, displaying high
intensities, followed by a decrease and subsequent
recovery. This was not true after proliferation arrest
induced by irradiation because MCF-7 cells displayed
a faster recovery of ML signals with respect to non-
irradiated cells, whereas the opposite was the case for
HeLa cells (Fig. 5).
On comparison, the ML intensity in cells (Fig. 5)
correlated with TG intensity in extracted lipids both

before [13] and after irradiation (Fig. 6). Only one
glycerol cross peak T, attributable to TG, was visible
in cells (Fig. 1B, insert) in the presence of high-inten-
sity lipid cross peaks, A, B and F (Fig. 3B). On the
other hand, spectra from extracted lipids showed two
separated cross peaks from PL and TG (Fig. 2B,
insert). We therefore attributed ML signals to neutral
lipids, mostly TG, in agreement with the data available
in the literature [1,13,15].
Consequently, the observed ML variations were
analysed in terms of production ⁄ consumption of TG.
In mammalian cells, TG may be deleted to meet the
needs of cell metabolism, either as a source of energy
or as a reservoir of fatty acids, after controlled hydro-
lysis. These molecules may also accumulate as stored
fatty acids, either originating from an external source
or due to phosphatidylcholine breakdown or synthesis
blockage. Indeed, there is evidence that, after induc-
tion of apoptosis, modification in phosphatidylcholine
synthesis accompanies TG accumulation in cells [7].
From the data shown in Fig. 8 relative to proliferat-
ing cells, for m
1
> 0 and m
2
< 0, we may infer that
c
2
> p
2

and c
1
> p
1
. This means that the consump-
tion lipid rate starts faster than production, but slows
down over time more than production. This latter
effect produces the very intense ML ⁄ TG signals that
are observed in the final days in culture. On the other
hand, the high rate of proliferation in the exponential
growth curve is accompanied by a high consumption
rate of TG (i.e. the ML ⁄ TG signal in cells decreases
shortly after seeding) (Fig. 5). This can be explained
by an increased fatty acid demand for PL synthesis as
cells proliferate and is in agreement with the high PC
signals observed in PCA extracts for both cell lines
shortly after cell seeding (not shown). It is worth not-
ing that the reduction in TG, hydrolyzed either due to
energy demands or as a source of fatty acids, occurs
when cells are in the presence of fresh growth medium,
from which both energy and fatty acids can be taken.
When cell growth slows down, the PL demand also
decreases, allowing the storage of fatty acids in TG. It
has been demonstrated that TG may be accumulated
through the diversion of PL synthesis from diacylglyce-
rols when phosphatidylcholine synthesis is reduced or
in the presence of phosphatidylcholine degradation
[18,25]. In particular, inhibition of the final step of
phosphatidylcholine synthesis via the Kennedy path-
way may divert diacylglycerol molecules into TG.

Moreover, PL catabolism via phospholipases, produc-
ing free fatty acids and lysophosphatidylcholine, may
also result in increased TG. A net accumulation of
TG is then bound to both the anabolic and catabolic
pathways of phosphatidylcholine.
After arrested growth, induced by irradiation, the
cell cycle profiles point to different mechanisms
(Fig. 4). The changes of m
1
and m
2
parameters move
in the same way for both cell lines after irradiation
(Fig. 9). Differences are more relevant and statistically
significant in independent experiments on MCF-7 cells,
A. M. Luciani et al.
1
H NMR of mobile lipids in tumour cells
FEBS Journal 276 (2009) 1333–1346 ª 2009 The Authors Journal compilation ª 2009 FEBS 1343
where the relevant increase of m
2
produces the relevant
intensity changes of ML ⁄ TG signals. In principle, such
an increase can be due to both a decrease in c
1
and an
increase in p
1
after growth arrest. Because this parame-
ter has a tendency to become positive in MCF-7 cells,

we may infer that c
1
tends to become lower than p
1
after growth arrest, at least in this cell line. The effect
was less intense in HeLa cells, probably due to the ten-
dency of the S phase to remain high in this cell line, in
contrast to MCF-7 cells (Fig. 4A¢,B¢) where the
S phase is low after irradiation. Due to a net accumu-
lation of PL synthesis in the S phase, if a reduced
S phase is paralleled by a reduction of the final step of
PL, TG accumulates, as observed in MCF-7 cells after
irradiation. A low PL synthesis in MCF-7 cells is also
sustained, as demonstrated by the observation in PCA
extracts (Fig. 7) where growth-arrested MCF-7 cells
show low PC and increased GPC, indicating an imbal-
ance between PL synthesis and degradation and, con-
sequently, an increase of TG. Other studies suggest
that the simultaneous accumulation of ML and GPC
may be linked to PL catabolism [7].
Modulation of ML intensity can then be ascribed to
an equilibrium between the build up and consumption
of these molecules due to the cellular state. In control
cells, these changes can only be ascribed to a depletion
of TG when cells are actively duplicating and a subse-
quent restoration of TG storage when cells approach
confluence. The timing of this equilibrium is influenced
by cell seeding conditions. On the other hand, when
cells stop growth due to the effects of irradiation, dif-
ferent trends are offered by the two cell lines.

Two major observations can be made. First, varia-
tion in ML intensity is not consistent with the onset of
apoptosis because ML signals decrease in HeLa cells,
which are characterized by relevant apoptosis [14]
(Fig. 4), and increase in MCF-7 cells with low, nonrel-
evant apoptosis [14]. ML accumulation after the arrest
in cell growth, followed either by cell death due to
apoptosis or mitotic death, provides further evidence
that these NMR signals are not specifically bound to
the induction of apoptosis, in agreement with recent
findings [8].
Second, the intensity of the ML signal, correspond-
ing to TG levels, is increased in irradiated MCF-7
cells, which have very low S phase. In mammalian cells
a net accumulation of PL occurs in the S phase: a rela-
tionship between a PL synthesis block, concomitant
with a reduction of S phase, and high intensity MLs
can therefore be postulated. A strong decrease in the
S phase was also observed in DU145 human prostatic
carcinoma cells upon treatment with either phenylace-
tate or phenylbutyrate, with the effect being larger in
phenylbutyrate-treated cells, where a larger increase of
ML signals was found [7]. Furthermore, prevalent
phosphatidylcholine breakdown can be envisaged in
MCF-7 cells due to a net increase of the GPC ⁄ PC
ratio that was observed in PCA extracts (Fig. 7).
The data obtained in the present study suggest, for
the first time, that TGs in mammalian cells different
from adipocytes can either be consumed, by hydroly-
sis, or produced, with a balance between production

and consumption that depends on the proliferative
state of the cell.
Experimental procedures
HeLa cells, kindly provided by G. Aquilina (Istituto Superi-
ore di Sanita
`
, Rome, Italy) and purchased from the ICRF
Cell Production (Clare Hall, UK) and MCF-7 cells, kindly
donated by S. Meschini (Istituto Superiore di Sanita
`
,
Rome, Italy) and purchased from ATCC (Manassas, VA,
USA) were grown as described previously [14]. Cells of
both cell lines were routinely seeded in 175 cm
2
flasks at a
density of 4 · 10
5
cells per flask in 50 mL of medium.
Cell culture flasks were irradiated with 20 Gy
60
Co
gamma rays. Cells were detached at different times after
irradiation (day zero = 2 h after irradiation), counted and
samples were prepared for NMR measurements, as
described below.
Cell cycle measurements
Cells were detached, washed and suspended in NaCl ⁄ P
i
at

a concentration between 1–2 · 10
6
mL
)1
. Next, 0.5 mL of
propidium iodide staining solution (10 mg of propidium
iodide, 0.1 mL of Triton-X, 3.7 mg of EDTA in 100 mL of
NaCl ⁄ P
i
) was added to 0.5 mL of cell suspension. The sam-
ples were incubated in the dark in the presence of heat-
inactivated RNase (0.01 mL of 10 mgÆmL
)1
stock solution)
for 30 min at room temperature or overnight at 4 °C. The
analysis was carried out using a Becton-Dickinson 3 FAC-
Scan flow cytometer equipped with a doubler discrimina-
tion module for DNA analysis and tested with Becton
Dickinson Immunocitometry Systems (BDIS) DNA Quality
Control Particles (Becton-Dickinson Biosciences, Franklin
Lakes, NJ, USA). Ten thousand events were collected for
each sample. The cell cycle profiles were analyzed and the
cell distributions in the different phases were determined
using modfit lt curve fitting software (Verity Software
House, Topsham, ME, USA).
Sample preparation for
1
H NMR measurements
Cell samples were prepared as described previously [13].
PCA extracts were prepared by mixing the cell pellet

with 30% (weight : volume) cold PCA. After 30 min, the
1
H NMR of mobile lipids in tumour cells A. M. Luciani et al.
1344 FEBS Journal 276 (2009) 1333–1346 ª 2009 The Authors Journal compilation ª 2009 FEBS
solution was neutralized with KOH and centrifuged. The
liquid phase was then lyophilized.
Before NMR measurements, samples were suspended in
D
2
O and the pH was adjusted to 7.4. Lipid extracts were
prepared by suspending the pellet obtained from PCA
extracts preparation in distilled water. The solution was
then lyophilized and the lyophilized powder dissolved in a
2 : 1 CDCl
3
⁄ CD
3
OD solution for NMR measurements, as
described previously [13]. All reagents for cells were pur-
chased from Sigma (St Louis, MO, USA) and deuterated
NMR reagents were purchased from Cambridge Isotope
Laboratories, Inc. (Andover, MA, USA).
1
H NMR measurements
To perform NMR measurements on cells, a pellet of
approximately 5 · 10
6
cells was suspended in NaCl ⁄ P
i
with

10%
2
H
2
O and inserted into a 1 mm tube.
1
H NMR spectra were run at 400 MHz on a digital
Avance spectrometer (Bruker, AG, Darmstadt, Germany).
Extracted lipids spectra were performed at 700 MHz on the
digital Avance spectrometer. The experimental parameters
were as described previously [14].
2D COSY spectra were recorded under water suppression
by presaturation, with a total of 16 scans for cells and typi-
cally 1000 scans for extracts. The 2D raw matrix consisted
of 512 complex points along the first dimension and 128
points along the second dimension. A sine function was
applied in both dimensions of the time domain before
Fourier transformation.
Integration of 1D and 2D signals was performed using
1D winnmr and 2D winnmr software (Bruker, AG,
Darmstadt, Germany) as described previously [13]. 2D
spectra were reconstructed in the magnitude mode. The
peaks were identified using data available from the litera-
ture [1–9] and from our own PCA, lipid and single com-
pound experiments. In accordance with a previous study
[26], the size of each rectangle marking the area used for
the volume integration of a peak was determined such that
it just covered the peak (above the noise) but not the
neighbouring peaks. After evaluation in a few spectra, the
optimized sizes of the rectangles (areas of integration) were

fixed and applied to all the spectra. Mean ± SD values
calculated from 1D deconvolution and 2D integration of
at least five spectra of samples prepared from the same cell
culture were used to assess the error on the single mea-
surement. Fitting of data to obtain model parameters was
performed using origin software (OriginLab Corp.,
Northampton, MA, USA).
Statistical analysis
Student’s t-test was applied to the two-sample groups to
compare variations in intensity of the control and irradi-
ated samples. Student’s t-test works under the assumption
of a Gaussian distribution of data, but also works remark-
ably well for distributions that are not accurately Gaussian
[27]. Variations in intensity of the examined signals are not
linear with time and ⁄ or cell proliferation [13]. It is therefore
necessary to examine the signal intensity for samples of
cells seeded and harvested under similar conditions. Fur-
thermore, we always irradiated cells under similar growth
conditions. Under this experimental set-up, the value distri-
bution appears to be Gaussian, although the limited num-
ber of experiments (approximately ten for each time
interval) cannot guarantee the presence of a perfect Gauss-
ian shape. The second hypothesis necessary to perform
Student’s t-test requires that the variance of the two samples
is equal. By estimating the variance from the standard devia-
tions and by using the F distribution [27], we could not find
any statistical difference in the variance when examining
control and irradiated samples. This indicates that the use of
Student’s t-test is compatible with the examined data.
Acknowledgements

This work was partially supported by INFN, experi-
ment EPICA.
References
1 May GL, Wright LC, Holmes KT, Williams PG, Smith
ICP, Wrights PE, Fox RM & Mountford CE (1986)
Assignment of methylene proton resonances in NMR
spectra of embryonic and transformed cells to plasma
membrane triglyceride. J Biol Chem 261, 3048–3053.
2 Hakumaki JM & Kauppinen RA (2000)
1
H NMR visi-
ble lipids in the life and death of cells. Trends Biochem
Sci, 25, 357–362.
3 Delikatny EJ, Lander CM, Jeitner TM, Hancock R &
Mountford CE (1996) Modulation of MR-visible mobile
lipid levels by cell culture conditions and correlations
with chemotactic response. Int J Cancer 65, 238–245.
4 Rosi A, Luciani A, Matarrese P, Arancia G, Viti V &
Guidoni L (1999)
1
H-MRS lipid signal modulation and
morphological and ultrastructural changes related to
tumor cell proliferation. Magn Reson Med 42, 248–257.
5 Brauer M (2003) In vivo monitoring of apoptosis. Prog
Neuropsychopharmacol Biol Psychiatry 27, 323–331.
6 Hakumaki JM & Liimatainen T (2005) Molecular imag-
ing of apoptosis in cancer. Eur J Radiol 56, 143–153.
7 Milkevitch M, Shim H, Pilatus U, Pickup S, Wehrle JP,
Samid D, Poptani H, Glickson JD & Delikatny EJ
(2005) Increases in NMR-visible lipid and glycero-

phosphocholine during phenylbutyrate-induced apopto-
sis in human prostate cancer cells. Biochim Biophys
Acta 1734, 1–12.
8 Santini MT, Romano R, Rainaldi G, Ferrante A,
Motta A & Indovina PL (2006) Increases in
1
H-NMR
A. M. Luciani et al.
1
H NMR of mobile lipids in tumour cells
FEBS Journal 276 (2009) 1333–1346 ª 2009 The Authors Journal compilation ª 2009 FEBS 1345
mobile lipids are not always associated with overt apop-
tosis: evidence from MG-63 human osteosarcoma three-
dimensional spheroids exposed to a low dose (2 Gy) of
ionizing radiation. Radiat Res 165, 131–141.
9 Mannechez A, Reungpatthanaphong P, de Certaines
JD, Leray G & Le Moyec L (2005) Proton NMR
visible mobile lipid signals in sensitive and multi-
drug-resistant K562 cells are modulated by rafts.
Cancer Cell Int 5,2.
10 Perez Y, Lahrech H, Caban
˜
as ME, Barnadas R, Sabes
M, Remy C & Aru` s C (2002) Measurement by nuclear
magnetic resonance diffusion of the dimensions of the
mobile lipid compartment in C6 cells. Cancer Res 62,
5672–5677.
11 Delikatny EJ, Cooper WA, Brammah S, Sathasivam N
& Rideout DC (2002) Nuclear magnetic resonance-
visible lipids induced by cationic lipophilic chemo-

therapeutic agents are accompanied by increased lipid
droplet formation and damaged mitochondria.
Cancer Res 62, 1394–1400.
12 Quintero MR, Caban
˜
as ME & Aru´ s C (2007) A possi-
ble cellular explanation for the NMR visible mobile
lipid (ML) changes in cultured C6 glioma cells with
growth. Biochim Biophys Acta 1771, 31–44.
13 Rosi A, Grande S, Luciani AM, Barone P, Mlynarik V,
Viti V & Guidoni L (2004)
1
H MRS studies of signals
from mobile lipids and from lipid metabolites: compari-
son of the behaviour in cultured tumour cells and in
spheroids. NMR Biomed 17, 76–91.
14 Rosi A, Grande S, Luciani AM, Palma A, Giovannini
C, Guidoni L, Sapora O & Viti V (2007) Role of gluta-
thione in apoptosis induced by radiation as determined
by
1
H MR spectra of cultured tumor cells. Radiat Res
167, 268–282.
15 Sathasivam N, Brammah S, Wright LC & Delikatny EJ
(2003) Inhibition of tetraphenylphosphonium-induced
NMR-visible lipid accumulation in human breast cells
by chlorpromazine. Biochim Biophys Acta 1633, 149–
160.
16 Willker W & Leibfritz D (1998) Assignment of mono-
and polyunsaturated fatty acids in lipids of tissues and

body fluids. Magn Reson Chem 36, S79–S84.
17 Choi GTY, Casu M & Gibbons WA (1993) NMR lipid
profiles of cells, tissues and body fluids. Biochem J 290,
717–721.
18 Bagnato C & Igal RA (2003) Overexpression of diacyl-
glycerol acyltransferase-1 reduces phospholipid synthe-
sis, proliferation, and invasiveness in simian virus
40-transformed human lung fibroblasts. J Biol Chem
278, 52203–52211.
19 Caviglia JM, de Go
´
mez Dumm INT, Coleman RA &
Igal RA (2004) Phosphatidylcholine deficiency upregu-
lates enzymes of triacylglycerol metabolism in CHO
cells. J Lipid Res 45, 1500–1509.
20 Igal RA, Caviglia JM, de Gomez Dumm INT & Cole-
man RA (2001) Diacylglycerol generated in CHO cell
plasma membrane by phospholipase C is used for
triacylglycerol synthesis. J Lipid Res 42, 88–95.
21 Veale MF, Roberts NJ, King GF & King NJC (1997)
The generation of
1
H-NMR detectable mobile lipids in
stimulated lymphocytes: relationship to cellular activa-
tion, the cell cycle, and phosphatidylcholine-specific
phospholipase C. Biochem Biophys Res Commun 239,
868–874.
22 Barba I, Mann P, Caban
˜
as ME, Aru´ s C & Gasparovic

C (2001) Mobile lipid production after confluence and
pH stress in perfused C6 cells. NMR Biomed 14, 33–40.
23 Wilson GD (2004) Radiation and the cell cycle, revis-
ited. Cancer Metastasis Rev 23 , 209–225.
24 Essmann F, Engels IH, Totzke G, Schulze-Osthoff K &
Ja
¨
nicke R (2004) Apoptosis resistance of MCF-7 breast
carcinoma cells to ionizing radiation is independent of
p53 and cell cycle control but caused by the lack of
caspase-3 and a caffeine-inhibitable event. Cancer Res
64, 7065–7072.
25 Jackowski S & Fagone P (2005) CTP: phosphocholine
cytidylyltransferase: paving the way from gene to mem-
brane. J Biol Chem 280 , 853–856.
26 Binesh N, Yue K, Fairbanks L & Thomas MA (2002)
Reproducibility of localized 2D correlated MR spectro-
scopy. Magn Reson Med 48, 942–948.
27 Barlow RJ (1989) Statistics: A Guide to the Use of
Statistical Methods in the Physical Sciences. John Wiley
and Sons, Chichester, pp. 159–160.
1
H NMR of mobile lipids in tumour cells A. M. Luciani et al.
1346 FEBS Journal 276 (2009) 1333–1346 ª 2009 The Authors Journal compilation ª 2009 FEBS