E. F., Schlegel, R. A., Schroit, A. J., Weiss, H. J., Williamson, P.,. 17. Lichtenberg, D., Robson ... Hurd, R. E., and John, B. K. (1991) J. Magn. Reson. 91, 648â. 8.
ANALYTICAL BIOCHEMISTRY ARTICLE NO.
245, 38–47 (1997)
AB969907
Organic Solvent Systems for 31P Nuclear Magnetic Resonance Analysis of Lecithin Phospholipids: Applications to Two-Dimensional Gradient-Enhanced 1H-Detected Heteronuclear Multiple Quantum Coherence Experiments M. Bosco,* N. Culeddu,† R. Toffanin,* and P. Pollesello‡ *POLY-bio´s Research Centre, Area Science Park, Padriciano 99, I-34012 Trieste, Italy; †CNR IATCAPA, via Vienna 2, I-07100 Sassari, Italy; and ‡Biostructure Department, Drug Discovery, R&D, Orion-Pharma, P.O. Box 65, FIN-02101 Espoo, Finland
Received August 5, 1996
31
P NMR of lipid extracts is a reproducible, rapid, and nondegradative method for qualitative and quantitative analyses of phospholipid mixtures. This analysis, however, is hampered by the instability of the solvent system commonly used for NMR spectroscopy (CHCl3/ CH3OH/H2O–EDTA). In this work we have investigated the effects of several monophasic solvent mixtures to overcome this disadvantage. Among these mixtures we have selected a solution of triethylamine, dimethylformamide, and guanidinium chloride (Et3N/DMF– GH/) as the most efficient system. In this solvent the chemical shift dispersion of the 31P signals is about four times the frequency range observed in the standard chloroform–methanol–water system. Moreover, the stability of this solvent, as a monophasic system, allows easy reproducibility of the analysis. The use of two-dimensional 1H– 31P gradient-enhanced heteronuclear multiple quantum coherence experiments can further exploit the higher resolution of the signals obtained with this solvent system for the structure elucidation of known and unidentified phospholipids. q 1997 Academic Press
The study of phospholipid (PL)1 composition in cell membranes has become of great importance in the past 1 Abbreviations used: THF, tetrahydrofuran; FA, formamide; DMA, dimethylacetamide; Et3N, triethylamine; GH/, guanidinium chloride; CH3OH, methanol; DMF, dimethylformamide; CHCl3 , chloroform; py, pyridine; EDTA, ethylenediaminetetraacetic acid; PL, phospholipids; PC, phosphatidylcholine; PE, phosphatidylethanolamine; PS, phosphatidylserine; PI, phosphatidylinositol; PG, phosphatidylglycerol; PA, phosphatidic acid; SM, sphingomyelin; lysophospholipid and plasmalogen related to a phospholipid are indicated using the prefixes L and pl, respectively; ge-HMQC, gradient enhanced 1H-detected heteronuclear multiple quantum coherence.
few years because some investigations have indicated that PLs have not only a structural function in building up the membrane bilayer, but also an active role in regulating of important biological processes such as selective cation interaction and adhesion phenomena (1– 3). The composition of PLs, the asymmetry of their distribution in the inner and outer leaflets of the membrane, and the PL compositions of different segments of Golgi and endoplasmic reticulum membranes are carefully regulated by the cell, as demonstrated by the existence of specific enzymes (4, 5). In addition, the composition and asymmetric distribution of PLs are probably affected by modifications in cell metabolism and/or in the external environment of the cell (5, 6) and represent a message that can induce the activation of other biochemical processes. It has also been demonstrated that a modification of PL composition also affects membrane fluidity, which has been related to pathological modifications of the cells (7, 8). For these reasons much effort was put into developing a nondegrading, reproducible method for qualitative and quantitative analyses of PL mixtures. Thin-layer chromatography and HPLC are the leading techniques for PL analysis, but quantitative determinations can be affected by errors (9). Moreover, suitable standards of PL from the same biological origin as the investigated sample are required for standardization curves. Recently, better precision was obtained using fluorescent probes that are incubated with the PLs; for this analysis, however, a fluorescent detector is required and complicated procedures must be followed (11). In contrast, 31P NMR spectroscopy is an easy and fast alternative for PL analysis; moreover, it allows recovery of nondegraded samples, which is important mainly for rare samples. In addition, quantitative determination of the components does not require stan-
38
AID
0003-2697/97 $25.00 Copyright q 1997 by Academic Press All rights of reproduction in any form reserved.
AB 9907
/
6m25$$$221
12-30-96 22:25:24
abal
SOLVENT SYSTEMS FOR
31
P NMR ANALYSIS OF PHOSPHOLIPIDS
39
dardization since the intensity of the 31P signals is not A 100-ml Hamilton syringe was used to prepare the affected by the nature of the esterified fatty acid on the solvent mixtures directly in a 5-mm NMR tube. Typiglycerol backbone as it is for HPLC. Also, 1H NMR cally, for a 1D 31P NMR experiment 15 mg of lecithin spectroscopy may be used for the study of PL mixtures, was dissolved in 500 ml of the organic solvent mixture. but in this case, high-field spectrometers are absolutely The 1D NMR spectra were acquired in unlocked mode required to resolve the overcrowded signals of proton on a Bruker AC 200 spectrometer equipped with a 5spectra (12–15). In contrast, in 31P decoupled spectra, mm multinuclear probe operating at 81.015 MHz for each PL component gives rise to one singlet that may the 31P nucleus or on a Varian VXR 300-S instrument be easily integrated or processed by means of numerical operating at 121.44 MHz for 31P nucleus. 31P decoupled analysis. For each PL class, the lyso-PL and the plas- spectra were typically acquired over 8K data points malogen can be identified separately. with an acquisition time of 2.05 s; the reading pulse Two alternative approaches are currently used for was 7 ms, equivalent to a flip angle of 757. The spectral 31 P NMR analysis: solubilization of PLs in water me- width was 2000 Hz, giving a digital resolution of 0.244 dium by means of detergents or solubilization in an Hz/pt after zero filling. Except for a few specific tests, organic solvent. In the first method, a large excess of all the experiments were performed at 296 K. Two detergent is used to reduce the signal linewidth; this thousand scans were generally used to obtain an optieffect is due to the dispersion of the PLs inside the mal signal-to-noise ratio. T1 values were obtained by a micelles caused by the detergents. Unfortunately, the standard inversion recovery experiment performed at resolution of the signals is dependent on the nature 7.05 T. Gradient-enhanced (ge) HMQC experiments and total concentration of the detergents and is not were performed on a Bruker 400 ARX spectrometer always optimal (16–18). Moreover, information from using a pulse sequence with z- gradients (21) as dethe 1H spectrum is lost due to the dominating signals of scribed by Pollesello et al. (22) the detergents. In the second method, the well-known The maximum amplitudes of the three sine-bellCHCl3/CH3OH/H2O–EDTA system is barely reproduc- shaped gradients (G1 , G2 , and G3), calculated from the ible due to the separation of two phases whose composi- gyromagnetic ratios of proton and phosphorus, were 9, tions are very sensitive to solvent treatment and sam- 9, and 7.26 G/cm. The duration of the gradient pulses ple manipulation (19, 20). was 2 ms, the delays for the gradient recovery were In this work many different organic solvents rang- 100 ms, and the relaxation delay was 1.5 s. The delay ing from less polar (benzene, triethylamine, ethyl ace- for the evolution of the long-range couplings was in the tate, and chloroform) to more polar (tetrahydrofuran, range 25–335 ms, and it was calculated from values of 1,4-dioxane, pyridine, dimethylacetamide, dimethyl- 3JPH and 4JPH ranging from 1.4 to 20 Hz. A delay as formamide, and water) were tested to overcome these long as 335 ms can be used due to the relatively long disadvantages. Much effort was made to use solvent T2 values of the phosphorus metabolites under examimixtures that do not produce phase separation and nation. The spectra were acquired over spectral winallow easy reproducibility of the analysis. Moreover, dows of 1000 Hz in F1 and of 4000 Hz in F2 . The number 2D gradient-assisted 1H – 31P heteronuclear multiple of transients was 8 per increment and 256 increments quantum coherence (HMQC) experiments were per- were collected. The total experimental time was 2 h. formed on a mixture of PLs (a) to facilitate the assign- All the 2D data were zero-filled in the F1 dimension ment of the signals in both the 1H and the 31P NMR and processed with sine-bell functions in both domains. spectra and (b) to determine that the chemical shift Data were expressed in magnitude mode after Fourier dispersion in this solvent system is sufficient for the transformation. identification and partial structure elucidation of known and unidentified phospholipids in mixtures. RESULTS AND DISCUSSION MATERIALS AND METHODS
Effect of Water
The NMR experiments were performed on a commercial sample of lecithin (Mr-Millers). All the organic solvents were analytical grade solvents obtained from Sigma–Aldrich Chemical Co. (tetrahydrofuran, 1,4-dioxane, formamide, dimethylacetamide, triethylamine, guanidinium chloride), Carlo Erba (methanol), or BHD Laboratory Supplies Poole (dimethylformamide, chloroform, pyridine). Pure phospholipid samples, i.e., phosphatidylcholine (PC), phosphatidylethanolamine (PE), phosphatidylserine (PS), phosphatidylinositol (PI), and sphingomyelin (SM), were obtained from Sigma Chemical Co.
In a previous work from one of the authors (19) it was demonstrated that the linewidth (n1/2) and chemical shift dispersion (Dd) of the 31P signals are very sensitive to the water content in the chloroform–methanol solvent system. In fact, n1/2 decreases with increased water content. Unfortunately, the miscibility of water in chloroform is limited and the role of methanol is to increase this solubility. Regardless, a physical limit cannot be overcome. As a first approach, we replaced CHCl3 with other less volatile organic solvents showing higher miscibility with water. In this set of experiments, 15 mg of a lecithin commercial sample con-
AID
AB 9907
/
6m25$$$221
12-30-96 22:25:24
abal
40
BOSCO ET AL.
taining a mixture of PC, PE, PI, and phosphatidic acid (PA) as main components, was dissolved in tetrahydrofuran (THF), 1,4-dioxane, pyridine (py), and triethylamine (Et3N), respectively. Increasing amounts of CH3OH/H2O–EDTA solution were added to the organic solvent and the 31P spectra were stored after each addition. The minimum n1/2 was obtained at different proportions of the components for each organic solvent. In all these proportions the system was monophasic. The first experimental result from these spectra was used to confirm the dependence of n1/2 and Dd on the signals from the water content as observed with chloroform. Moreover, with these solvents increased resolution of the signals was obtained in comparison to the standard Glonek–Meneses system (9) as shown in Fig. 1. Effect of pH This first set of experiments revealed that the 31P d in organic systems is also strongly dependent on the pH. This is not surprising since all water-soluble phosphorylated metabolites show typical sigmoidal curves of 31P d as a function of pH (23). Although a precise determination of pH is not possible in organic solvents, a ‘‘basic’’ environment is present in the py and, even more, in the Et3N system. This increases the separation of the d among the phosphate groups. The effect is dramatic for phosphomonoesters like PA, whose signals are shifted from 1.2 to 1.3 up to 4.5 to 5.5 ppm from the PC signal. Of course, the same effect can be observed in the ‘‘neutral’’ THF or 1,4-dioxane after addition of NaOH/H2O solutions. The fact that py and Et3N in the presence of water produce organic cations like pyH/ and Et3NH/, respectively, which can replace Na/ or K/ as counterions of the charged phosphate groups, must also be considered. It is known that the nature of the cations present in the system can affect the d of the 31P signals (9, 24). It is not surprising that each solvent system can induce a different, specific d for each PL. Moreover, the presence of EDTA was confirmed to reduce the signal n1/2 in any solvent because of its powerful ability to form complexes with divalent inorganic cations. In fact, when traces of paramagnetic cations are present in the PL mixture, the relaxation rates of 31 P nuclei are increased and the relative signals are broad. Although the basic character of the organic solvent is very efficient in spreading the d of the signals, thus increasing the resolution, the presence of OH0 anions produces hydrolysis of the fatty acid esters, which are labile under basic conditions. In contrast, the phosphate esters are labile only under acidic conditions. In Et3N/H2O solutions, the hydrolysis can be followed carefully: the intensity of the PL signals decreases and corresponding lyso-PL signals increase progressively (Fig. 2). The hydrolysis rate is decreased when samples are stored at 4–57C and when 0.5 M HCl solution is used instead of neutral water. Also in the
AID
AB 9907
/
6m25$$$221
12-30-96 22:25:24
Et3N/CH3OH/H2O–EDTA system, the hydrolysis rate is reduced as a consequence of the lower concentration of water. Under these experimental conditions the lysoPL signals increased by 9% after 24 h. Since the basic environment is due to py or Et3N and the hydrolysis is performed by OH0 anions, the two effects can be separated. A basic-induced separation of the signals in the absence of hydrolysis can be obtained only in waterfree systems. Other liquid amines such as diethylamine (Et2N), benzylamine, and tert.-butylamine were tentatively used on the lecithin sample, but gave worse results than Et3N: PE and PI signals overlapped and the solubility of PLs was lower. Solvent Nature and Composition As previously mentioned, PLs in anhydrous organic solvents always give very broad 31P signals (lw Å 30– 50 Hz). Many trials to obtain sharp signals with organic molecules instead of water have failed. Nevertheless, a similar trend in the spectral variations was obtained using molecules with high dielectric constants, high dipolar moments, and the ability to produce a hydrogenbonding network (25). Among these, some amides, dimethylacetamide (DMA), dimethylformamide (DMF), and formamide (FA), gave better results. It is interesting to note that FA, which has a dielectric constant even higher than that of water (e Å 111), is not able to mimic the water molecules better than other molecules. This confirms the unique properties of water. Since the n1/2 of the signals is related mainly to the mobility of the PL molecules, indicating their packing inside macroscopical aggregations, we used denaturing agents able to destroy weak van der Waals interactions and hydrogen-bonding networks. Although urea and urea derivatives, e.g., tetramethylurea (TMU), were not effective in reducing signals n1/2 , guanidinium chloride (GH/) was very efficient. The average linewidths of 31P signals of lecithin PLs solubilized in several organic solvents and binary or ternary systems are listed in Table 1. Other solvents not included in this table were also tested (n-esane, benzene, ethyl acetate, acetonitrile, TMU, FA, acetone, dimethyl sulfoxide). They were finally excluded because of the low solubility of PLs or the low miscibility with highly polar solvents or because they produced selective precipitation of some PLs from the lecithin mixture. In Table 1, the solvents are listed according to their solubility index Sp0 (25) from Et3N to H2O. The table displays the pure solvents on the diagonal and the progressively more ‘‘heterogeneous’’ mixtures far from the diagonal. Some trends can be evaluated from the results reported in this order: (i) The solubility order of the PLs follows the Sp0 order of this set of solvents. (ii) The PLs signals are always very broad in pure solvents independent of their solubility.
abal
SOLVENT SYSTEMS FOR
31
P NMR ANALYSIS OF PHOSPHOLIPIDS
41
FIG. 1. 31P NMR spectra of lecithin in four water-containing solvent mixtures; note that only the CHCl3/CH3OH/H2O–EDTA system undergoes separation of two nonmiscible phases. The EDTA concentration in water solution was 0.2 M, pH 6.0; the volume ratios among the three liquid components were 10:4:1 for CHCl3 :CH3OH:H2O, 1.82:0.73:1 for THF:CH3OH:H2O, 6.30:4.05:1 for py:CH3OH:H2O, and 8:1:1 for Et3N:CH3OH:H2O, respectively.
(iii) There is always a remarkable decrease in n1/2 after addition of increasing amounts of highly polar components, especially water, and this is evident in the cells far from the diagonal. The lowest linewidth values are depicted in bold. (iv) Obviously, the relative miscibility of the solvents decreases with increased polarity difference (i.e., far from the diagonal). For this reason, in some cases, a
AID
AB 9907
/
6m25$$$221
12-30-96 22:25:24
third component as CH3OH was used to keep the system monophasic. The effect of each solvent on the chemical shift of P signals was seldom forecasted a priori because we observed very large variations of d for each PL class with respect to external 85% (w/w) orthophosphoric acid. Of course, this depends on the specific interaction 31
abal
42
BOSCO ET AL.
FIG. 2. 31P NMR spectra of lecithin in Et3N:H2O (9:1); the analysis was repeated on the same sample 3 and 17 days after the preparation to follow the effect of hydrolysis.
of the solvent with the polar head of the PL, including hydrogen bonding, dipolar, van der Waals, and electrostatic interactions. Nevertheless, one clear trend can be observed in the d effect on the basic nature of the solvent as already mentioned. On the other hand, the
AID
AB 9907
/
6m25$$$221
12-30-96 22:25:24
solvent effect on the signal linewidth can be discussed in greater detail. Since the relaxation parameters (T1 , T2) are strongly dependent on the mobility of the molecules, the n1/2 of the signals can be used to investigate the aggregation pattern of the PLs. It is well known
abal
SOLVENT SYSTEMS FOR
31
43
P NMR ANALYSIS OF PHOSPHOLIPIDS TABLE 1
Measured Linewidths (in Hz) of
31
P Lecithin Signals Obtained in Pure Solvents and Mixed Solvent Systems 6.0/4.0
É30 Et3N
4.6/4.6/0.8
3.5/5.3/1.2
5.8/3.9/0.3
ss É30 l 5.0
CHCl3
7.5/2.5 6.7/2.9/0.4 7.9/2.1
1.4/8.6 1.4/8.6 1.0/8.0/1.0
7.7/2.3 7.6/2.3/0.1
0.3/5.0/4.7 0.7/6.6/2.7
6.7/3.3
3.6/6.4 2.1/5.1/2.8
3.3/4.5/2.2
2.6/7.4
4.9/5.1
1.8/8.2 0.9/5.5/3.6
ú50 s
THF
7.6/2.4 5.2/2.5/0.3 7.7/2.3
ú50 É50
m ss
l 1.5
7.7/2.3 4.1/5.7/0.2
20
ú50
1,4-diox l
py
DMA
DMF
H2O
Sp0 Å eÅ
40 l 6.5
m m
3.7 l 1.0 m 1.7
ú50 7
l 7.5 m 3.3
2.0 j 1.2 m 1.3
m 5.5 ml1.5
1.3 ❖ 1.3 j.1.5
j 5.2 j.2.0
Et3N 8.0 2.42
CHCl3 9.3 4.81
m 4.4
ll ; 6.5
m 4.4 ll
.l0.8 j.1.1 THF 9.5 7.58
ml1.4 1,4-diox 10.0 2.21
ml0.8 j.1.3 py 10.6 12.91
ll DMA 11.1 37.78
DMF 12.1 36.71
H2O 23.4 78.3
Note. In the ternary systems the third component is indicated by the following symbols: l, H2O; j, CH3OH; m, guanidinium chloride; ., EDTA; l, NaOH/H2O; HCl/H2O; ;, Et3N. The solubility of lecithin sample in each pure solvent is indicated in the diagonal cells with the following symbols: ss, very high; s, high; m, medium; l, low; ll, almost insoluble. The relative proportions of the solvents in binary and ternary systems are given above the diagonal; the first number refers to the solvent reported in the vertical axis, the second number refers to the solvent in the horizontal axis, and the third number refers to the solvent reported in symbol. The concentration of guanidinium chloride was almost the same in all systems (77 mg/ml).
(16) that small micelles in H2O–detergents (and inverse micelles in organic solvent) allow a high degree of molecular motion; as a consequence T1 values were found in the range 1–3 s and the n1/2 values on the order of 1–2 Hz. In contrast, vesicles or other forms of double-layer packing do not allow sufficient tumbling and the observed signals are very broad. Although we do not have detailed information about the real packing in the organic solvents investigated, in analogy with other studies, we can suppose that high-molecularweight aggregations of PLs or clusters of micelles are present in all the pure solvents. Since the entropy (DS) drives toward the dissolution of PLs, it is reasonable to assume that these large aggregations are stable if the interactions with the solvent are not enthalpically (DH) efficient. We could not find any solvent molecule with chemical functions so versatile as to interact fa-
AID
AB 9907
/
6m25$$$221
12-30-96 22:25:24
vorably with both the hydrophobic acyl chains and the charged head groups of the PLs. For this reason, only a mixture of highly different solvents can produce lowmolecular-weight aggregations such as micelles. In other words, the critical concentration at which vesicles are formed from micelles is strongly dependent on the solvent composition and nature; it increases when highly hydrophobic solvents are mixed with highly polar solvents. The critical micellization concentration probably also shows similar behavior, but we could not find a way to increase it enough to reach values reasonable for a 31P NMR analysis. Finally, we identified some of the main properties in a search for the ‘‘ideal’’ solvent system: (i) The system should have at least two components miscible in the optimal proportions; eventually a third
abal
44
BOSCO ET AL. TABLE 2 31
P Chemical Shifts and T1 Values of the Most Common PLs
Solvents
PC
LPC
PS
PE
SM
PI
PG
LPI
PA
LPA
A(d/ppm) B(d/ppm) T1B/s
0.00 0.00 1.14 { 0.02
0.56 0.45 1.3 { 0.1
0.63a 0.54
0.84 0.56 1.35 { 0.07
0.75 0.84
0.56 1.02 0.85 { 0.04
1.41 1.25 1.3 { 0.2
1.03 1.51
1.14a 5.34 1.22 { 0.09
1.75 6.01
Note. A: CHCl3/CH3OH/H2O–EDTA; B:Et3N/DMF–GH/. The d of PLs was referred to PC to show the chemical shift differences and signal dispersion (Dd) in the two solvent systems. In system A the d of PC was 00.84 ppm as referred to external 85% (w/w) orthophosphoric acid; in system B the d of PC was /0.38 ppm with respect to the same standard. The third row of data shows the T1 values measured on the lecithin sample in solvent B. a In three different papers, Glonek et al. (9, 10, 24) assigned PS and PA signals in a variable range of frequencies (0.63–0.79 ppm for PS and 1.14–1.63 ppm for PA), although the samples were prepared under the same experimental conditions (K/ form).
component can be used to increase the miscibility and to inhibit the PLs aggregations. (ii) One of the components should be hydrophobic for proper interaction with acyl chains of the fatty acid esters. (iii) The second component should be polar, with a high dielectric constant. (iv) One of the components should have a strong basic character but, in this case, water must be avoided to prevent hydrolysis. (v) The solubility of the PLs should be as high as possible. (vi) For further NMR investigations, common perdeuterated solvents are preferred. After several steps we succeeded in fulfilling the previous conditions: Et3N was chosen for its basic nature, for its low polarity, and because it ensures high solubility of PLs. DMF was found to be the most compatible polar partner for Et3N, and it is easily available as a perdeuterated solvent. Guanidinium chloride (GH/) was very efficient in preventing macroscopical aggregations due to its known denaturing properties (26). This novel solvent system can be easily obtained by mixing 1 ml of DMF, 0.3 ml of Et3N, and 100 mg of GH/. The monophasic mixture obtained can be stored for a long time at room temperature without undergoing any detectable modification. The stability of the proportion of components, in contrast with the CHCl3/CH3OH/H2O– EDTA, is ensured by the relatively high boiling point of both solvents. When the system is stored at 47C, it undergoes separation of two nonmiscible phases, but the process is reversible upon warming at room temperature. Since the system is anhydrous, it does not produce any hydrolysis, as demonstrated by repeat analysis of the same sample after several weeks. The relative integrals of the lecithin signals for all PL classes were found to be the same as those in the standard CHCl3/ CH3OH/H2O–EDTA system. To recover the sample, the solution can be evaporated under reduced pressure at room temperature and then dissolved in chloroform; the GH/Cl0 is not soluble and can be separated by filtration.
AID
AB 9907
/
6m25$$$221
12-30-96 22:25:24
Many other possible solutions were tested but, in our experience, this combination was the most stable, reproducible, and efficient. The chemical shift of PC was found to be /0.37 { 0.01 ppm with respect to external 85% orthophosphoric acid. Moreover, in a preliminary study using crude extracts from heterogeneous tissues, the standard deviation of the chemical shift for all the PL classes was found to be 0.009 ppm under conditions of widely differing concentrations, relative compositions, and biological origin of PL mixtures. The chemical shift order of the different PL classes is obviously different from that in the CHCl3/CH3OH/H2O– EDTA solvent that we assigned from commercial standard phospholipids and from ge-HMQC spectra (see Table 2 and Fig. 3). The redistribution of the resonances allows easier identification and quantitative determination of PL components from the 31P NMR spectra. For instance, SM is 0.30 ppm downfield from PE (it is 0.09 ppm in CHCl3/CH3OH/H2O–EDTA) and PI is 0.48 ppm downfield from PE (it is 0.28 ppm in CHCl3/ CH3OH/H2O–EDTA and overlaps with lyso-PC) so that they can be integrated separately. Also, plasmalogens and lyso-PL can be distinguished from the corresponding 1,2-diacylglycerophospholipids. The average linewidth of all the signals was 1.2 Hz and their T1 values (at 7.05 T) were in the range 0.85– 1.2 s as reported in Table 2; these values are compatible with mobile molecules inside small micelles similar to those observed in water after addition of a considerable amount of detergents (16–18). Effect of Temperature Since some of the solvent systems investigated have high boiling points, in contrast with CHCl3/ CH3OH/H2O – EDTA, the sample can also be analyzed at higher temperatures. The temperature very often increases the relative miscibility of solvents and PLs; moreover, it should increase the mobility of the PL molecules, thus reducing the linewidths in the spectra. Unfortunately, as a result, the n1/2 was not significantly reduced in the investigated range of temperatures (296 – 353 K). Moreover, the kinetic rates
abal
SOLVENT SYSTEMS FOR
31
P NMR ANALYSIS OF PHOSPHOLIPIDS
45
FIG. 3. 31P NMR spectra of lecithin in the standard CHCl3:CH3OH:H2O–EDTA (10:4:1) solvent system and in Et3N:DMF–GH/ (0.3:1); the guanidinium chloride concentration in the final mixture was 77 g/liter.
of hydrolysis or other degrading phenomena (i.e., migration of phosphate groups inside the glycerol) were increased. Unexpectedly, in some cases, such as Et3N/H2O, the signals were sharper when the temperature was decreased to 273 K, but the separation of two phases occurred.
AID
AB 9907
/
6m25$$$221
12-30-96 22:25:24
2D Spectroscopy 1
H– 31P ge-HMQC experiments were performed on a PL mixture in the Et3N/DMF–GH/ solvent system containing DMF-d7 . Using this method it was possible to assign rapidly and unequivocally most of the protons
abal
46
BOSCO ET AL.
FIG. 4. Expansions of a 1H– 31P ge-HMQC NMR spectrum acquired for a PL mixture in the Et3N/DMF–GH/ solvent system containing DMF-d7 . The spectrum was recorded in 2 h with an evolution delay of 250 ms and without decoupling during acquisition. Peaks attributed to the most abundant components of the mixture, i.e., PC, PME (phosphatidyl-N-monomethylethanolamine), PDE (phosphatidyl-N,Ndimethylethanolamine), LPC, PS, PE, SM, PI, PG, LPI, PA, and LPA, are marked on the projection in F1 .
coupled to 31P, both from the polar heads and from the glycerol or sphingosine moiety. It is interesting to note that a cross peak (Fig. 4) relative to the nitrogen-bound methylene protons of ethanolamine (at 2.8 ppm in F1) was also detected, although the 4JPH coupling constant is very small. No cross peaks relative to the methine proton at position 2 on the glycerol backbone of the phospholipids (5.0–5.3 ppm in F2) were detected. Another advantage of using gradient-enhanced experiments is the good suppression of the T1 noise generated by undeuterated solvents. The good resolution of the signals in both dimensions makes this technique useful for the rapid and correct identification of PL components in complex mixtures. CONCLUSION
Some clear advantages in the study of PL mixtures can be obtained by using the Et3N/DMF–GH/Cl0 sol-
AID
AB 9907
/
6m25$$$221
12-30-96 22:25:24
vent system instead of the well-known CHCl3/CH3OH/ H2O–EDTA system. Among these, we emphasize the stability of the solution (as a monophasic system) so that a new preparation is not needed for each analysis. The system produces a reduced linewidth of 31P NMR signals and an increased dispersion of their chemical shifts. The solution is stable and allows the recovery of the PL samples. Although other still untested combinations of organic solvents might be more efficient, the system presented here can be a further step in improving the NMR potential in analysis of PL mixtures.
ACKNOWLEDGMENTS We are grateful to Dr. Luca Stucchi for useful suggestions on the chemical effects of the organic solvents. We also thank Professor Mario Branca and Dr. Debora Schneider for helpful discussions.
abal
SOLVENT SYSTEMS FOR
31
P NMR ANALYSIS OF PHOSPHOLIPIDS
REFERENCES 1. Utsugi, T., Schroit, A. J., Connor, J., Bucana, C. D., and Fidler, I. J. (1991) Cancer Res. 51, 3062–3066. 2. Blumenfeld, N., Zachowski, A., Galacteros, F., Beuzard, Y., and Devaux, P. F. (1991) Blood 77, 849–854. 3. Bevers, E. M., Wiedmer, T., Comfurius, P., Zhao, J., Smeets, E. F., Schlegel, R. A., Schroit, A. J., Weiss, H. J., Williamson, P., Zwaal, R. F. A., and Sims, P. J. (1995) Blood 86, 1983–1991. 4. Williamson, P., Antia, R., and Schlegel, R. A. (1987) FEBS Lett. 219, 316–329. 5. Schroit, A. J., and Zwaal, R. F. A. (1991) Biochim. Biophys. Acta 1071, 313–329. 6. Op den Camp, J. A. F. (1979) Ann. Res. Biochem. 48, 47–71. 7. Mu¨ller, P., Zachowski, A., Beuzard, Y., and Devaux, P. F. (1993) Biochim. Biophys. Acta 1151, 7–12. 8. Sentjurc, M., Schara, M., and Auersperg, M. (1988) Arch. Geschwulstforsch 58, 35–42. 9. Meneses, P., and Glonek, T. (1988) J. Lipid Res. 29, 679–689. 10. Meneses, P., Navarro, J. N., and Glonek, T. (1993) Int. J. Biochem. 25, 903–910. 11. Bernhard, W., Linck, M., Creutzburg, H., Postle, A. D., Arning, A., Martin-Carrera, I., and Fr.Sewing, K. (1994) Anal. Biochem. 220, 172–180. 12. Sparling, M. L., Zidovetzki, R., Muller, L., and Chan, S. I. (1989) Anal. Biochem. 178, 67–76.
AID
AB 9907
/
6m25$$$221
12-30-96 22:25:24
47
13. Casu, M., Lai, A., Pilia, A., Casti, G., Zedda, S., and Gibbons, W. A. (1992) Arch. Gerontol. Geriatr. Suppl. 3, 111–122. 14. Choi, G. T. Y., Casu, M., and Gibbons, W. A. (1993) Biochem. J. 290, 717–721. 15. Casu, M., Anderson, J. G., Choi, G., and Gibbons, W. A (1991) Magn. Reson. Chem. 29, 594–602. 16. London, E., and Feigenson, G. (1979) J. Lipid Res. 20, 408–412. 17. Lichtenberg, D., Robson, R. J., and Dennis, E. A. (1983) Biochim. Biophys. Acta 737, 285–304. 18 Bradamante, S., Barchiesi, E., Barenghi, L., and Zoppi, F. (1990) Anal. Biochem. 185, 199–303. 19. Branca, M., Culeddu, N., Fruianu, M., and Serra, M. V. (1995) Anal. Biochem. 232, 1–6. 20. Hedzes, H. T., Teerlink, T., Van der Knaap, M. S., and Valk, J. (1991) Magn. Reson. Med. 21, 107–116. 21. Hurd, R. E., and John, B. K. (1991) J. Magn. Reson. 91, 648– 653. 22. Pollesello, P., Eriksson, O., Vittur, F., Paoletti, S., Geimonen, E., and Toffanin, R. (1995) NMR Biomed. 8, 190–196. 23. Robitaille, P. M. L., Robitaille, P. A., Brown, G. G., Jr., and Brown, G. G. (1991) J. Magn. Reson. 92, 73–84. 24. Merchant, T. E., and Glonek, T. (1992) Lipids 27, 551–559. 25. Barton, A. F. M. (1975) Chem. Rev. 75, 731–753. 26. Faltz, L. L., Reddi, A. H., Hascall, G., Martin, D., Pita, J. C., and Hascall, V. C. (1979) J. Biol. Chem. 254, 1375–1380.
abal