Quantitative HRMAS proton total correlation spectroscopy applied to cultured melanoma cells treated by chloroethyl nitrosourea: Demonstration of phospholipid metabolism alterations



Recent NMR spectroscopy developments, such as high-resolution magic angle spinning (HRMAS) probes and correlation-enhanced 2D sequences, now allow improved investigations of phospholipid (Plp) metabolism. Using these modalities we previously demonstrated that a mouse-bearing melanoma tumor responded to chloroethyl nitrosourea (CENU) treatment in vivo by altering its Plp metabolism. The aims of the present study were to investigate whether HRMAS proton total correlation spectroscopy (TOCSY) could be used as a quantitative technique to probe Plp metabolism, and to determine the Plp metabolism response of cultured B16 melanoma cells to CENU treatment in vitro. The exploited TOCSY signals of Plp derivatives arose from scalar coupling among the protons of neighbor methylene groups within base headgroups (choline and ethanolamine). For strongly expressed Plp derivatives, TOCSY signals were compared to saturation recovery signals and demonstrated a linear relationship. HRMAS proton TOCSY was thus used to provide concentrations of Plp derivatives during long-term follow-up of CENU-treated cell cultures. Strong Plp metabolism alteration was observed in treated cultured cells in vitro involving a down-regulation of phosphocholine, and a dramatic and irreversible increase of phosphoethanolamine. These findings are discussed in relation to previous in vivo data, and to Plp metabolism enzymatic involvement. Magn Reson Med 49:241–248, 2003. © 2003 Wiley-Liss, Inc.

The ability to follow phospholipid (Plp) metabolism is of paramount importance in many circumstances in which cell survival and cell proliferation are of concern—for example in neurological disorders and cancer (1, 2). NMR spectroscopy has long been known to be important tool for exploring Plp metabolism (3–6). Novel NMR developments, such as high-resolution magic angle spinning (HRMAS) probes, have provided improved spectra of cultured cells (7) and normal (8, 9) and cancer (10–12) tissues. Recently, using HRMAS together with proton total correlation spectroscopy (TOCSY), we showed that a mouse-bearing B16 melanoma tumor responded to chloroethyl nitrosourea (CENU) treatment in vivo by altering its Plp metabolism (12).

The first purpose of the present study was to demonstrate on cultured B16 melanoma cells in vitro that HRMAS proton TOCSY cross-peak signal variations of Plp derivatives reflected concentration variations. The second purpose was to determine the Plp metabolism response of cultured B16 melanoma cells to CENU treatment in vitro, as a model for Plp metabolism alterations previously observed in vivo.

The Plp derivatives that were simultaneously observed using proton TOCSY were choline (Cho), phosphocholine (PC), cytidyl-diphosphate-choline (CDP-Cho), ethanolamine (Eth), phosphoethanolamine (PE), CDP-ethanolamine (CDP-Eth) (all of which are derivatives of the phosphatidylcholine (PtdCho) and phosphatidylethanolamine (PtdEth) biosynthetic pathways), and glycerophosphocholine (GPC) and glycerophosphoethanolamine (GPE) (derivatives of PtdCho and PtdEth hydrolysis) (1–3).

As regards the most expressed Plp derivatives in cultured B16 melanoma cells, HRMAS proton TOCSY signals revealed linearity with 1D saturation recovery signals, the NMR spectroscopy reference for quantifying concentrations. Therefore, HRMAS proton TOCSY was used to quantify concentration changes of water-soluble Plp derivatives in CENU-treated cultured melanoma cells. The response of cultured B16 melanoma cells to CENU treatment involved a transient accumulation of GPC and GPE, a down-regulation of PC and increase of CDP-Eth during cell proliferation inhibition, and a dramatic and irreversible rise of PE during and after cell proliferation inhibition. These data are discussed in relation to mouse-bearing B16 melanoma tumor response to CENU in vivo, and to Plp metabolism enzymatic involvement.



N′-[2-chloroethyl]-N[2-(methylsulfonyl)ethyl]-N′-nitrosourea (cystemustine) is a CENU antineoplastic agent (13) that has been proposed for the treatment of human malignant melanoma and glioma. Cystemustine (Orphachem, Clermont-Ferrand, France) was supplied as a 5-mM solution in 0.9% NaCl. D2O (SDS, Peypin, France) was used to lock the spectrometer.

Cell Cultures

Transplantable B16 melanoma cells originating from C57BL6/6J Ico mice (ICIG, Villejuif, France) were adapted to grow in culture. Melanocytes were maintained as monolayers in 75-cm2 culture flasks in Eagle's MEM-glutaMAX medium (Life Technologies) supplemented with 10% fetal calf serum (Boehringer), and 4 μg/ml Gentamicin (Boehringer).

B16 melanocytes were harvested by trypsinization and plated 20 hr before a 2-hr exposure to 200 μM cystemustine. At defined times (6, 15, 25, 34, and 45 days), cells were harvested by trypsinization and then counted. The cell suspension was washed in saline and centrifuged. The cell pellet (typically 2 × 106 cells, from one or two culture flasks during growth inhibition) was suspended in three drops of saline D2O solution. It was then transferred to the NMR rotor to be examined without delay, or frozen at −80°C for later examination. During each follow-up examination, four independent measurements were performed.

NMR Spectroscopy

NMR Spectroscopy was performed on a small-bore Bruker DRX 500 magnet (Bruker, Karlsruhe, Germany) equipped with an HRMAS probe. Samples at room temperature were set into ZrO2 rotor tubes (4 mm diameter, 50 μl), without an upper spacer. Rotors were spun at 4 kHz.

Assignment of Plp Derivatives

The proton resonances of the amine headgroups, Cho and Eth, are characteristic of each Plp derivative and, due to scalar coupling among hydrogen nuclei of neighbor methylene groups, they give rise to cross-correlation signals (chemical shifts are given in ppm; the hyphenated numbers are the chemical shifts of protons to which this proton is scalar coupled): Cho (3.20, 3.55-4.07), PC (3.23, 3.62-4.18), CDP-Cho (3.23, 3.68-4.39), GPC (3.23, 3.68-4.34), Eth (3.15-3.80), PE (3.22-3.99), CDP-Eth (3.30-4.20), GPE (3.30-4.12) (14). Model mixtures of CDP-Cho and CDP-Eth were used to unequivocally attribute their cross-peaks (not shown). A reaction scheme of Plp metabolism is given in Fig. 1 showing the relationships between these derivatives.

Figure 1.

Reaction scheme of phospholipid metabolism. The following water-soluble derivatives were investigated: Cho, choline; Eth, ethanolamine; PC, phosphocholine; PE, phosphoethanolamine; CDP-Cho, cytidine diphosphate choline; CDP-Eth, cytidine diphosphate ethanolamine; GPC, glycerophosphocholine; and GPE, glycerophosphoethanolamine. Phospholipids : PtdCho, phosphatidylcholine; PtdEth, phosphatidylethanolamine. Enzymes of phospholipid metabolism: PLA, phospholipase A (PLA1: EC; PLA2: EC; LPL, lysophospholipase (EC; PLC, phospholipase C (EC; PLD, phospholipase D (EC; CKI, Cho kinase (EC; CCT, CTP:PC cytidylyl transferase (EC; CPT, PC transferase (EC; EKI, Eth kinase (EC; ECT, CTP:PE cytidylyl transferase (EC; EPT, PE transferase (EC; PEMT, PtdEth N-methyl-transferase (significant only in the liver, PEMT-I: EC; PEMT-II: EC Left ascending vertical pathway: CDP-Cho pathway; right ascending vertical pathway: CDP-Eth pathway; transversal pathway: PEMT pathway.

1D Proton NMR Spectroscopy

1D spectra were acquired and processed from a workstation. After tuning the coil and shimming the magnet from a previously recorded shim table, the water signal was obtained and its frequency was retained for subsequent water signal suppression. An 8 μs radiofrequency pulse was applied, and proton free-induction decays (FIDs) were collected with water signal suppression at low power, 10 ppm spectral width, 8 K complex data points, 10-s repetition time (TR), and 32 accumulations. After Fourier transformation, a baseline correction was applied using a spline algorithm in the spectral domain of interest.

The Plp derivative lines were then fitted a mixed Gaussian-Lorentzian shape using the XWINNMR software. Cho was integrated from the 3.20 ppm resonance (9 protons), PC from the 3.62 ppm resonance (2 protons), PE from the 3.99 ppm resonance (2 protons), and CDP-Eth from the 4.20 ppm resonance (2 protons) (Fig. 2).

Figure 2.

a and b: Typical 1D spectra of untreated and CENU-treated cultured B16 melanoma cells. Fully relaxed saturation recovery spectra of (a) untreated cultured B16 melanoma cells, and (b) cultured B16 melanoma cells 15 days after treatment by CENU. Proton resonances of interest arose from Cho, O[BOND]CH2[BOND]CH2[BOND]N(CH3)math image, and Eth, O[BOND]CH2[BOND]CH2[BOND]NHmath image, headgroups. Attribution of signals relevant to methods are shown.

2D Proton NMR Spectroscopy

Investigation of Cho- and Eth-containing derivatives was performed using 2D TOCSY. The sequence was used with water signal suppression, high-resolution sampling (spectral bandwidth of 6 ppm, 256 data points along f1, and 2 K data points along f2), 1-s TR, and 16 accumulations. The phased mode was achieved using the Time Proportional Phase Increments (TPPI) procedure (15). Raw data were processed with a quadratic sine function and then were Fourier transformed.

The TOCSY sequence was optimized for the mixing time using a cascade of measurements on several cell samples. The optimum signal-to-noise ratio (SNR) for most of the Plp derivatives was obtained for a mixing time of 75 ms (data not shown) that was retained for acquisitions. Isotropic mixing during TOCSY sequence was achieved using a DIPSI-2 pulse train (16).

The stability of cells inside the rotor was evaluated by a cascade of TOCSY measurements for several hours. We verified that the sample starting at room temperature, Plp derivative signals varied by 20% after a 3-hr stay inside the rotor. Our overall acquisition took 80 min, which was short enough to adequately reflect Plp derivative changes that take place in cultured cells.


For quantification, TOCSY spectra were phased along both frequency axes and baseline-corrected using a low-order spline function. Cross-peak volumes (CPVs) were integrated using the XWINNMR software from a recorded template of regions applied to cross-peaks of interest (Cho, PC, GPC, CDP-Cho, Eth, PE, GPE, and CPD-Eth). CPVs were measured in the upper half plane of the spectrum, where they were exempt from t1-noise and from partial saturation effects due to water signal suppression. Internal standardization was performed using the signal of glycine (Gly) at 3.56 ppm. The Gly resonance was easily separated and integrated, and provided same standardization in 1D and 2D spectra.

For the most highly expressed Plp derivatives, the ratio of their CPVs to the diagonal-peak volume (DPV) of Gly obtained from TOCSY spectra, equation image, was compared to the ratio of their signal integral (I) to that of Gly, equation image, obtained from 1D fully relaxed saturation recovery spectra. The latter spectra were averages of 1D spectra framing each TOCSY spectrum. Only the equation image ratio was corrected for the number of protons and had the meaning of a concentration ratio. Then, based on the calibration of TOCSY signals against 1D signals, 1D signal integral equivalents (I*[Plp derivative]) were calculated from TOCSY signals (CPV[Plp derivative]) according to

equation image(1)

where slope was the average slope of the established calibration curves, and equation image was a standardization constant between the 1D and 2D sequences. Then absolute concentrations were obtained using the internal water standard method (17)

equation image(2)

where C*[Plp derivative] was the Plp derivative concentration estimate in mM, I[Water] was the integral of water signal, N* was the number of protons associated to I* (e.g., 2 if I[Gly] held for 2 protons in Eq. [1]), and c[Water] was the NMR-visible water cell concentration in mM as given in Ref. 17.

T1 Relaxation Time Measurements

T1 estimates of Plp derivative cross-peaks were obtained using IR-TOCSY (an inversion-recovery sequence with TOCSY as the read block). The sequence was implemented with water signal suppression during the inversion period. To limit the consequences of sample degradation, acquisitions were performed alternating long and short inversion times until the data-point series was completed. T1's were thus estimated from a three-parameter model using a nonlinear fit (Levenberg-Marquardt algorithm) (18):

equation image(3)

where P0 was the maximum signal, P1 was the fraction of inversion, P2 was the T1 to determine, τ was the waiting time of the sequence, and x was the time of inversion that was allowed to vary 12 times, between 10 ms and 5 s.

Statistical Analysis

Data are presented as mean ± SEM. Kinetics were analyzed using ANOVA followed by a post-hoc test with comparisons to the pretreatment data point.


Quantification of Plp Derivatives

Calibration of TOCSY Signals Against Saturation- Recovery Signals

Typical 1D fully relaxed saturation recovery spectra are displayed in Fig. 2. Highly expressed Plp derivatives or Plp derivatives with easily separable resonances in 1D spectra (Cho, PC, PE, and CDP-Eth) were used to verify the linearity of TOCSY measurements against 1D fully relaxed saturation recovery measurements. Since PC was low in treated cultures, a second untreated data set (three untreated cell cultures) was added to the plots. Linear fits with strong correlation coefficients were obtained for the investigated Plp derivatives (Fig. 3a–d). We inferred that linearity held for other Plp derivatives that, due to poorer expression or overlapping signals, could not be directly measured from 1D saturation recovery spectra.

Figure 3.

a–d: Calibration of 2D against 1D spectra. Plot of 2D TOCSY cross-peak volume normalized to Gly diagonal-peak volume, against 1D saturation recovery signal integral normalized to Gly signal integral, for the following Plp derivatives: (a) Cho, (b) PC, (c) PE, and (d) CDP-Eth. Open squares: untreated; crosses: additional untreated; full squares: CENU-treated cultured cells.

T1 Measurements

T1 estimates of Plp derivative cross-peaks in intact untreated and CENU-treated cells are given in Table 1. T1's of Cho, PC, and Gly were not significantly different between both groups. In addition, from one Plp derivative to another, T1 differences were moderate, and thus did not underlie the strong differences that could be observed between Plp derivative CPVs.

Table 1. T1 Estimates (ms) of Plp Derivative Cross-Peaks in Cultured B16 Melanoma Cells
  • Data obtained from IR-TOCSY measurements (6 measurements in each group, mean ± SEM) at 500 MHz and room temperature, and using Eq. [3].

  • a

    Diagonal-peak volume measurement.

  • b

    Not significant, CENU-treated vs. untreated cultured cells, unpaired t-test.

Untreated cells774 ± 48716 ± 421404 ± 50
Treated cells882 ± 43863 ± 53728 ± 83715 ± 89b740 ± 82b753 ± 511411 ± 106b

Cell Count Curve

Cells were counted before collection for NMR spectroscopy. Cell counts dropped at day 6 and remained low until day 25, indicating growth inhibition. From day 35, cultured B16 melanoma cells resumed growth (Fig. 4).

Figure 4.

B16 melanoma cell count curve. Cell count in monolayer per cell culture flask (75 cm2). CENU treatment (200 μM cystemustine) was given for 2 hr on day 1. Bar: SEM. P < 0.01 for data points between days 6 and 25, vs. day 0 (untreated cultured cells).

Plp Metabolism Alterations

Plp derivative concentrations were calculated from TOCSY spectra, using Eqs. [1] and [2]. PC decreased strongly during the growth inhibition phase and was quite low during regrowth, while Cho and CDP-Cho levels showed weak variation (Figs. 5a–d and 6a). Eth levels were low and showed weak variation throughout the experiment. In contrast, CDP-Eth pathway compounds were up-regulated. PE increased dramatically at day 6, reaching a maximum at day 25 (Figs. 5a–d and 6b). During late follow-up and after cell regrowth, PE remained strongly elevated. CDP-Eth was low at day 6, elevated at days 15 and 25, and reached baseline at day 34 (Figs. 5a–d and 6b). GPC and GPE increased between days 6 and 15 (day 25 for GPE), during the growth inhibition period, and then returned to baseline (Figs. 5a–d and 6c).

Figure 5.

a–d: Typical TOCSY spectra of untreated and CENU-treated cultured B16 melanoma cells. TOCSY spectra obtained (a) in untreated cultured B16 melanoma cells, and at (b) day 6, (c) day 25, and (d) day 45 after CENU treatment. Attributions of most relevant signals are displayed as dotted boxes.

Figure 6.

a–c: Plp derivative kinetics of CENU-treated cultured B16 melanoma cells. Long-term kinetics of concentrations of several Plp derivatives: (a) Cho (full triangles), PC (full squares), and CDP-Cho (open circles); (b) Eth (full triangles), PE (full squares), and CDP-Eth (open circles); and (c) GPC (full lozenges) and GPE (open lozenges). CENU treatment was given for 2 hr on day 1. Each data point was the average of four independent measurements. Bar: SEM. PC: P < 0.05 between days 15 and 34; PE: P < 0.01 between days 6 and 45; CDP-Eth: P < 0.01 between days 15 and 25; GPC: P < 0.05 between days 6 and 15; GPE: P < 0.01 between days 6 and 25, vs. day 0 (untreated cultured cells).


This work demonstrates that HRMAS proton TOCSY can be used as a quantitative tool to follow Plp derivative content, and that cultured B16 melanoma cells respond to CENU treatment by down-regulating PC, and dramatically and irreversibly increasing PE. In addition, transient activation of the CDP-Eth pathway was demonstrated during cell proliferation inhibition.

HRMAS Proton TOCSY as a Quantitative Tool: Validation and Limitations

In previous studies 2D techniques have been used successfully for quantification purposes. In the field of cancer, and among homonuclear 2D spectroscopy sequences, proton COSY had enabled the quantification of relative concentrations of some Plp derivatives and other species from perchloric acid extracts of colon carcinoma (19). COSY has also been used to obtain fucose:Cho ratios in several cancer cell lines (20). However, most authors acknowledge the superiority of TOCSY as applied to heterogeneous biological material, because of its increased sensitivity, ability to elicit long-range correlations, and pure absorption mode of cross-peaks (21, 22). Quantitative TOCSY was used to obtain a GPC:Cho ratio that was found to be increased in leiomyosarcomas in comparison with leiomyomas (21). It has also been used to perform lipid group analysis, showing increased fatty acyl unsaturation in leiomyosarcomas according to their mitotic activity (23), and increased fatty acyl saturation of cholesteryl ester and triglyceride in multidrug-resistant carcinoma cell lines (24).

In the present study there were moderate differences in the slopes of calibration curves from one Plp derivative to another, which indicates that caution should be used when making between-derivative quantitative comparisons. These differences may be related to T1 and/or polarization transfer effects. However, T1 estimates of cross-peaks were obtained in untreated and CENU-treated cultured cells for some Plp derivatives without significant differences between groups. Physico-chemical changes occur in treated cell cultures, such as the accumulation of melanin (a solid pigment that carries free radicals and traps paramagnetic metal ions) (25), but these changes do not have a significant impact on T1 values. Therefore, the proton TOCSY CPV variations observed in this study were mostly attributable to concentration variations. Absolute concentrations calculated this way were in very good agreement with literature data (3), so we can infer that this quantitative procedure is valid in other cultured cell types and tissue samples.

Plp Metabolism Alterations Under CENU Treatment

We found that pulse exposure of B16 melanoma cell cultures to CENU resulted in dramatic changes in Plp derivative expression. Cystemustine is known to produce DNA damage (13) and induce growth arrest but not apoptosis in B16 melanoma (26, 27). A few days after cystemustine treatment, cultured B16 melanoma cells exhibit blockage in the G2 phase. After 2 weeks they exit the G2 phase, cross the M phase, and enter the G1 phase again, while progressively losing synchronization (26). Our NMR observations may be explained in part by the progression of the cultured cells through the cell cycle after they were blocked in the G2 phase.

In untreated B16 melanoma cells, the most highly expressed Plp derivative was PC. The PC level may be related to activation of phospholipase C, activation of Cho-kinase, and/or blockage of CTP:PC-cytidylyl transferase (CCT, the key-enzyme of PtdCho biosynthesis) (Fig. 1) (1–3).

During growth inhibition of CENU-treated cells, GPC and GPE were transiently up-regulated. This pattern may be attributed to phospholipase A-mediated hydrolysis of PtdCho and PtdEth (28) because of a sudden accumulation of Plp following the arrest of proliferation. It could also be related to decreased GPC and GPE hydrolysis.

In CENU-treated cells, the CDP-Cho pathway was strongly altered with a down-regulation of PC. This could be due to phospholipase C inactivation, blockage of Cho-kinase, and/or consumption of PC through the CDP-Cho pathway. However, phospholipase C is required for the observed mitogenic effect (29), and resumption of growth was not accompanied by PC-level recovery. In addition, sustained blockage of Cho-kinase is unlikely because Cho levels did not rise over the long term. Finally, it has been shown that activation of phospholipase A, as hypothesized from the observed increase in GPC and GPE, is aimed at counterbalancing the activation of the CDP-Cho pathway (28). Therefore, the observed changes in the CDP-Cho pathway metabolites may reflect increased activation of this pathway.

The CDP-Eth pathway was markedly altered under treatment. PE was dramatically and irreversibly up-regulated from day 6 during growth inhibition and regrowth, attesting to a transformed Plp phenotype of cultured B16 melanoma cells having been exposed to CENU. Also CDP-Eth was transiently but strongly up-regulated on days 15–25. CDP-Eth pathway activation is aimed at producing PtdEth, an important Plp, possibly to compensate for insufficient PtdCho production, or to participate in the cleavage furrow of dividing cells or in cell signaling processes (1–3). CTP:PC-cytidylyl transferase (CCT) activity has been shown to be maximal during early G1 phase, and then to decrease during late G1 and S phases (29). From our data this pattern appears to be valid for CTP:PE-cytidylyl transferase (ECT) as well. Long-term PE increase could result from sustained activation of phospholipase C, Eth-kinase or sphingosine-1-phosphate lyase (1–3). However, PtdEth-specific phospholipase C is not demonstrated in mammalian cells (3). In addition, activation of sphingosine-1-phosphate lyase should result in an accumulation of Eth, which was not found. Therefore the observed increase of PE should reflect activation of Eth-kinase.

Comparison With Treated B16 Melanoma Tumor Response In Vivo

There were some similarities between the cultured cell response to CENU treatment in vitro and that of the mouse-bearing B16 melanoma tumor in vivo. In the treated cultured cells there was transient variation in GPC and GPE, as well as a sustained rise of PE, as in the treated tumor (12). Possible mechanisms have been discussed above and are probably shared by both melanoma models. These similar responses validate cultured cells as a model by which we can better understand Plp derivative alterations induced by CENU in vivo.

However, there were differences between cultured cells and tissue as regards derivatives of the CDP-Cho pathway. PC was down-regulated in treated cultured cells, whereas it was rather elevated in treated tumors. Furthermore, Cho remained low in treated cultured cells although it transiently increased in treated tumor. One possibility is that Cho was insufficiently supplied to cultured cells by the medium used, although tumors in vivo would have increased Cho extraction from the blood. Another hypothesis is that phospholipase D or phosphatidylserine synthase activation (1–3), which is an intracellular process that releases Cho, was not involved in the cultured cells.

In conclusion, this study demonstrates that the response of cultured B16 melanoma cells to CENU treatment involves the down-regulation of PC, transient increase of CDP-Eth, and irreversible rise of PE. This response validates cultured melanoma cells as a model for melanoma tumor response to CENU in vivo. In addition, the response of the cultured cells could be related to the activation of Plp metabolism enzymes, which, at the cell biology level, should be triggered by the progression of cultures through cell cycle, proliferation, and survival mechanisms.