J. Phys. Chem. 1996, 100, 15211-15217
15211
Multinuclear NMR Investigation of Phosphatidylcholine Organogels† Donatella Capitani,‡ Anna Laura Segre,*,‡ Frank Dreher,§ Peter Walde,§ and Pier Luigi Luisi§ Istituto di Strutturistica Chimica, G. Giacomello, CNR, Area della Ricerca di Roma, C.P. 10, I-00016 Monterotondo Stazione, Rome, Italy, and Institut fu¨ r Polymere, ETH-Zu¨ rich, CH-8092 Zu¨ rich, Switzerland ReceiVed: March 15, 1996; In Final Form: June 17, 1996X
A multinuclear NMR investigation of organogels formed by soybean lecithin and by a series of synthetic phosphatidylcholines in cyclohexane in the presence of a small amount of water is presented. The NMR measurements are based on 1H, 13C, and 31P dynamic parameters and the line width. To study the gelation process, measurements are carried out with samples at different amounts of added water. Both for proton and phosphorus resonances, the onset of the gel formation is clearly evidenced by a broadening of the line width. In the first set of measurements soybean lecithin is studied. It is shown that as water is being added, the line widths of the different protons of lecithin become broader, each to a different extent. Particularly significant is the stiffening of the geminal protons at the sn-1 position of the glycerol backbone. 31P NMR T2 measurements allow the distinction between gel-forming and nongel-forming solvents. The NMR line width broadening is also present in regions in which rheology data show no high viscosity, e.g., at high water content and/or at low lecithin concentration. This is thought to indicate that a considerable molecular stiffening of the glycerol moiety and of the phosphate is present even in the absence of a high viscosity macroscopic gel structure. To study the influence of the molecular structure on the dynamics of gel formation, studies have been extended to synthetic gel-forming phosphatidylcholines, such as 1,2-dioleoyl-sn-glycero-3phosphocholine (DOPC) between 281 and 300 K, 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine (POPC) at 300 K, and 1,2-dipalmitoyl-sn-glycero-3-phosphocholine (DPPC) between 313 and 333 K; 1,2-dilinoleoylsn-glycero-3-phosphocholine (DLPC) at 281 K. On all gels, differences in 31P NMR T2 values are quite small, while the line widths, both on protons as well as on phosphorus, appear to be much more sensitive to differences in the molecular architecture. Accordingly, this study allows one to draw a quite general picture of lecithin gels in which the molecular structure is linked to the dynamic parameters during gel formation, which are in turn linked to the macroscopic physical properties such as viscosity and phase transition temperature. By comparison of all these data, it appears DOPC is the closest model to natural lecithin. Even in this case, however, caution is required, since local motions in the glycerol moiety are more hindered in DOPC than in soybean lecithin.
Introduction It was reported some time ago by Scartazzini and Luisi1 that the addition of small amounts of water to solutions of soybean lecithin2 in apolar solvents induces the formation of transparent, highly viscous gels with a viscosity as high as 104 P. Schurtenberger et al.,3-5 based mostly on light- and neutron-scattering experiments, have shown that the high viscosity is due to an entangled dynamic network of long reverse micellar, wormlike aggregates (flexible cylinders), which in fact behave in several aspects like equilibrium polymers. From the applicative point of view, these gels are being investigated as biocompatible matrix for transdermal drug delivery.6-9 Whereas scattering and viscosity studies are instrumental to clarifying the overall extensive structural properties of the gels, clarification at the molecular level is rather a domain of NMR. A previous 31P NMR study on lecithin gels enabled one to draw a correlation between the mobility of the phosphate group and the molar ratio [water]/[lecithin] ) wo. Furthermore, during the gelation process, as a function of wo, a good correlation was found between viscosity and the broadening of the phosphorus line width and the broadening of the 13C signals of the glycerol moiety.10,11 This was interpreted as a progressive * To whom to address correspondence. † This work is dedicated to Professor A. Ballio on the occasion of his 75th birthday. ‡ Istituto di Strutturistica Chimica, CNR, Rome. § Institut fu ¨ r Polymere, ETH-Zu¨rich. X Abstract published in AdVance ACS Abstracts, August 1, 1996.
S0022-3654(96)00811-8 CCC: $12.00
stiffening of these two parts of the molecule during the process of gelation. However, no obvious stiffening of the acyl chains could be observed. There are several other questions concerning lecithin gels that can be tackled by NMR measurements. A general question is whether NMR allows one to distinguish between solvents in which the gelation process takes place (gel-forming solVents) and solvents in which no gels are obtained. Another general and important question is about the dynamic behavior of the individual atoms of the lecithin molecule during and after the gelation process in order to understand which parts of the molecule are more mobile and which are stiffened. Finally, there is the question as to whether and to what extent NMR and viscosity data (until now mainly used to characterize the gels macroscopically) are consistent with each other. In the first part of the paper, mostly the soybean lecithin/ cyclohexane/water system will be analyzed, since this is indeed the system that has been most extensively studied thus far. We will use a multinuclear NMR approach that involves the use of 31P, 1H, and 13C dynamic parameters and the line width. Since soybean lecithin, as a natural product, is a mixture of different phosphatidylcholines, the case of synthetic phosphatidylcholines with a well-specified alkyl chain structure will be also considered. The results will be compared with the naturally occurring phosphatidylcholine mixture from soybeans. Materials and Methods Reagents. Purified soybean lecithin (“Epikuron 200”, batch 1-6-1003, mean relative molecular mass of 760) was obtained © 1996 American Chemical Society
15212 J. Phys. Chem., Vol. 100, No. 37, 1996 with a purity of 95% from Lucas Meyer GmbH, Germany.12 The lysolecithin content was 0.6%, which could be detected by thin layer chromatography on silica gel plates using CHCl3/ CH3OH/40% (v/v) CH3NH2 at a volume ratio of 50:30:8 as eluant, stained with iodine or 0.65% (w/v) molybdenum blue in 4 M H2SO4. Lecithin stained at an Rf of 0.55 and lysolecithin at Rf ) 0.29. The main remaining impurities were water, other phospholipids, and R-tocopherol (∼0.2%). According to Lucas Meyer GmbH, the fatty acid composition of the hydrolyzed Epikuron 200 product is (values given relative to the total amount of fatty acids) 16-20% palmitic acid (C 16:0) and stearic acid (C 18:0), 8-12% oleic acid (C 18:1), 62-66% linoleic acid (C 18:2), and 6-8% linolenic acid (C 18:3).12 The dominating fatty acid residues esterified with the glycerol oxygen at the sn-1 position are palmitic acid, oleic acid, and linoleic acid, while the oxygen at the sn-2 position is over 70% esterified with linoleic acid.13 The water content in the soybean lecithin sample was determined by FTIR, a methodology described before.14 About 0.5-1 molecule of water per lecithin was already present in the sample used. All wo values given in this paper are relative to that for water added to the system, neglecting the small amount of water already present. 1,2-Dioleoyl-sn-glycero-3-phosphocholine (DOPC), 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine (POPC), 1,2-dilinoleoyl-sn-glycero-3-phosphocholine (DLPC), and 1,2-dipalmitoylsn-glycero-3-phosphocholine (DPPC)sall with a >99% purityswere purchased from Avanti Polar Lipids, Inc., Pelham, Alabama. Deuterated cyclohexane and deuterated benzene were purchased from Carlo Erba S.p.A., Milano, Italy. Preparation of the Samples. All samples were prepared according to a previously described procedure.1 NMR Measurements. 13C NMR T1 relaxation times were measured at 50.33 MHz on a Bruker AC 200 spectrometer. 31P NMR spectra were performed at 32 MHz on a Bruker WP 80, at 81 MHz on a Bruker AC 200, or at 162 MHz on a Bruker AC 400. All instruments were equipped with a Bruker variable temperature controller unit, and the temperature was kept at 300 K unless otherwise specified. 31P NMR spectra at 81 and 162 MHz were obtained in standard 5 mm tubes; owing to the poor signal/noise ratio at 32 MHz, 10 mm NMR tubes were used in that case. In all cases the sample was kept well within the receiver coil. All T1 relaxation measurements were performed using the standard inversion recovery sequence.15 T2 measurements were performed using the standard CPMG sequence,15 and relaxation delay was always kept longer than 5T1. The intensity of the echoes was used as an input for the exponential fit.16 31P NMR spectra were run both broad-band (BB) and nonbroad-band proton decoupled; the decoupling procedure did not affect the measurement of the relaxation times at all. In all cases the use of D2O instead of H2O or the use of deuterated organic solvents instead of nondeuterated solvents did not affect the value of measured T1 and T2 relaxation times. 1H NMR spectra were run at 600.13 MHz on a Bruker AMX 600 spectrometer. Full assignment of the observed resonances was performed with a 2D TOCSY experiment.17 The assignment of the peaks in the 13C NMR spectrum was made with a X-H correlated 2D experiment performed in reverse detection.18 Viscosity Measurements. For the determination of the zero shear viscosity (ηs) a Physica Rheolab MC 100 instrument (from Physica Messtechnik GmbH, Stuttgart, Germany) with a concentric cylinder geometry was used. The measurements were performed in analogy to those described before4 at a temperature of 298.0 ( 0.1 K.
Capitani et al.
Figure 1. 1H NMR spectrum, at 600.13 MHz, of 30 mM soybean lecithin in cyclohexane at wo ) 1. Chemical structure of soybean phosphatidylcholine and proton labeling is also reported. Number of C atoms in the two acyl chains of soybean phosphatidylcholine is predominantly12 16 or 18. All double bonds in the unsaturated fatty acid residues are unconjugated and have a cis configuration.13 Carbon atoms of the glycerol backbone are numbered according to the rules for the stereospecific numbering of phospholipids.2
Results and Discussion A. Case of Soybean Lecithin. 1H NMR Measurements. It is worth reporting a complete assignment of the signals in the 1H NMR spectrum of the lecithin in cyclohexane/water (Figure 1). For that, it is necessary to have an assessment of the molecular structure of the soybean lecithin used in our experiments. This is not trivial, since naturally occurring lecithin is a mixture of phosphatidylcholines in which the alkyl chains of the acyl moieties vary in length and in number/type of double bonds. A simplified representation of the chemical structure of soybean lecithin is shown in Figure 1. Soybean lecithin is rich in nonconjugated double bonds that are separated by only one methylene group (sCHdCHsCH2sCHdCHs), so actually the best model compound for soybean lecithin would be represented by 1,2-dilinoleoyl-sn-glycero-3-phosphocholine (see later in the paper). Let us consider first the 1H NMR spectrum at 600.13 MHz of the soybean lecithin/cyclohexane/water system at wo ) 1, as shown in Figure 1, with the full assignment of the signals. It agrees with data obtained on similar compounds in another solvent.19 By the increase of the water content, some of the resonances become broader while others become sharper (Figure 1S in Supporting Information). The most significant broadening occurs for the resonance at ∼4 ppm arising from the AA′ protons at the sn-1 position of the glycerol backbone. This signal becomes broader and broader until, at wo ) 6, it fully disappears. Upon addition of water, in the full wo range, protons CC′ at the sn-3 position broaden, remaining well observable. Thus, the two pivot CH2 groups of the glycerol skeleton show a completely different mobility. It is worth noting that from wo ) 1 to wo ) 4, the resonances due to protons P and Q (-CO-CH2- of the fatty acid) broaden more than twice. As the water content increases, all resonances due to the fatty chains progressively broaden, although this effect is considerably less marked than in the case of the glycerol signals. The further away from the glycerol backbone the protons are, the less dramatic the observed broadening is.
Phosphatidylcholine Organogels
Figure 2. Soybean lecithin (30 mM)/cyclohexane/water system: wo dependence of the line width of the resonances for the methylene protons (A, C, C′, M, N) and the methyl protons R of the glycerol and head group part of the lecithin.
In the case of the resonance at ∼3.4 ppm, due to the methyls (R) of the choline head group (-N+(CH3)3), the line width initially sharpens and then from wo ) 4 to wo ) 16, it modestly broadens from ∼6 to ∼12 Hz. A similar trend is also found for the methylene groups of the polar head labeled M (P-OCH2-) and N (-CH2-N+). A summary of the behavior of the line width of a few selected resonances is reported in Figure 2. When the temperature is lowered, all resonances become broader, maintaining, however, the relative broadenings previously reported in Figure 2. No phase separation is observable down to temperatures as low as 278 K. Proton spectra were also performed on samples at low lecithin concentration (3 mM) where the viscosity is rather low and a true gel is absent. At relatively large wo values (wo ) 16-20), the resonance due to AA′ broadens up to 55 Hz. Thus, at low lecithin and high water content a considerable stiffening of the lecithin molecule must exist, although no dramatic increase in the macroscopic viscosity is observed. We will come back later to this comparison between NMR molecular stiffening and viscosity measurements. All 1H NMR data regarding the dependence of the line width of the proton resonances on wo can be rationalized on the basis of a previously proposed model for the water-induced lecithin gelation.3-5 The addition of water to small lecithin reverse micelles (in cyclohexane present at wo ≈ 2) leads to a progressive stiffening of the lecithin molecules, in particular around the protons at the sn-1 carbon atom and, to a lesser extent, around the phosphate group. This stiffening occurs when long cylinders are formed. The viscosity dramatically increases owing to the entangling present in the nondilute regime of rodlike cylinders.20 To verify the local stiffening of the molecules as a function of wo, and also to check the gel stability, the proton line widths were studied as a function of the temperature. These data are shown in Figure 3 (circles), where only a few selected resonances are reported. 31P NMR Measurements. 1H NMR measurements allow an overview of the overall behavior of the lecithin molecule during the gelation process. To clarify specific structural issues, 31P NMR measurements have been performed. As already mentioned, the formation of lecithin gels is observed in a large variety of organic (mostly apolar) solvents.1,20 However, in
J. Phys. Chem., Vol. 100, No. 37, 1996 15213
Figure 3. 1H NMR at 600.13 MHz showing line widths of few selected resonances (signals C, M, N, R) as a function of wo and temperature; O refers to soybean lecithin data, while 3 refers to DOPC data.
Figure 4. wo dependence of the 31P NMR T2 spin-spin relaxation time at 81 MHz for the gel-forming system soybean lecithin/cyclohexane/water (circles) and for the nongel-forming system soybean lecithin/ benzene/water (squares). Lecithin concentration in both cases was 40 mM.
some apolar solventssmost notably benzene and toluenesno gels are formed. Thus, organic solvents can be regarded as gelforming and nongel-forming solvents. A question arises of whether a difference between the two classes of solvents can be detected by NMR measurements. To address this question, 31P NMR T2 measurements on the soybean lecithin/cyclohexane/water (gel-forming) systems and soybean lecithin/benzene/water (nongel-forming) systems at different wo values were performed. Results are shown in Figure 4. Note that already at wo ) 0 (no added water), the T2 relaxation time in cyclohexane is much shorter than in benzene, most likely reflecting differences in the state of aggregation of lecithin in these two (dry) solvents. Upon addition of the first water molecule (wo ) 1), T2 increases in both systems. At higher water content, T2 decreases in cyclohexane, whereas in benzene the reverse is true. At wo ) 4, the difference reaches 1 order of magnitude: 350 ms in benzene vs 30 ms in cyclohexane. As discussed above, the decrease in T2 values in cyclohexane can be explained by the formation of cylindrical, wormlike lecithin aggregates, while in benzene spherical reverse micelles are present21-23 in which the mobility of the phosphate group is not restricted.
15214 J. Phys. Chem., Vol. 100, No. 37, 1996
Capitani et al.
TABLE 1: wo Dependence of the 31P NMR T2 Spin-Spin Relaxation Time (Values in ms) at 81 MHz for the System Soybean Lecithin/Cyclohexane/Water at 300 K and at Different Lecithin Concentrations and of 30 mM DPPC/ Cyclohexane/Water at 333 K soybean lecithin conc (mM) 10 20 30 40 80 160
0
1
2
wo 3
4
6
8
75.7 135.1 108.3 92.7 46.2 15.9 16.9 83.0 117.0 98.0 76.2 39.6 8.7 9.6 67.7 94.3 106.5 51.5 36.1 10.9 8.5 73.2 91.8 92.3 54.4 25.6 a 7.4 63.6 78.0 70.0 52.7 23.6 10.8 6.9 62.4 69.8 70.8 44.6 8.6 6.0 5.9 wo
DPPC conc (mM)
0
1
2
3
30
317.4
372.6
353.6
328.1
a
4
6
8
108.7 34.4 21.3
Not determined.
Thus, 31P NMR data allow a clear discrimination between the two classes of solvents. In nongel-forming solvents a relatively large molecular mobility of the phosphate group is present that increases by increasing the amount of water. Conversely, in gel-forming solvents, cylinders are formed and the phosphate mobility is restricted; further addition of water enhances the stiffness. A tentative explanation of this difference between the two solvents may invoke the higher polarizability of benzene with respect to cyclohexane. Owing to this, the charged lecithin molecule may interact more stronglysi.e., may be much more solvatedswith benzene than with cyclohexane. Thus, intermolecular interactions among lecithin molecules, very important in building intermolecular “pregel aggregates”, are not favored in the aromatic solvent. The remarkable solvating tendency of benzene has been documented some time ago by Eicke, although in a somewhat different context.24 Owing to its high sensitivity to the local stiffening, 31P NMR T2 was studied as a function of the lecithin concentration as well as that of wo. The gelation process as a function of the lecithin concentration has been previously studied in terms of shear viscosity.3 Thus, a related, interesting question is whether a correlation can be found between viscosity and T2 data. 31P NMR T values as a function of w and of the lecithin 2 o concentration were measured (Table 1). For all concentrations, a maximum in 31P NMR T2 is observed between wo ) 1 and wo ) 2. This maximum was previously observed in the lecithin/ isooctane/water system and attributed to a conformation change of the polar head.10 Thus, yet again the maximal mobility of the phosphate group is present near wo ) 1-2. In all cases, at wo > 2, a sharp decrease of the mobility is observed, i.e., a kind of abrupt stiffening (or a kind of “melting”, if the data trend is analyzed going from wo > 2 down to wo ) 1-2). In the dilute concentration regime and at high wo values, NMR data point to the presence of molecular stiffening on the phosphate (see Figure 5), where the 31P NMR line width (circles) and the 31P NMR T2 (triangles) are reported as a function of the added water for a 3 mM lecithin concentration. The line width broadens, from 6 up to 50-60 Hz, and the sharp decrease occurs in T2 values at wo > 8. The line width increase is far from trivial and, in agreement with T2 data, suggests that even at low lecithin concentration, a molecular stiffening occurs. Once again, this can be interpreted assuming that cylindrical inverted micelles containing stiffened lecithin molecules are formed even at 3 mM (the cmc in cyclohexane is below 1 mM25). However, owing to the low lecithin content, the concentration of these flexible cylinders and, therefore, the frequency of entanglement are low. This is
Figure 5. wo dependence of the 31P NMR line width (b) and of the 31P NMR T spin-spin relaxation time (1) at 81 MHz for the soybean 2 lecithin (3mM)/cyclohexane/water system.
Figure 6. wo dependence of the 31P NMR line width (b) and of the 31P NMR T spin-spin relaxation time (1) at 81 MHz for (a) the 2 soybean lecithin/cyclohexane/water system, (b) the DOPC system, (c) the POPC system, where in the case of T2, values for 160 mM POPC are also given (3), and (d) the DLPC system at 280 K. In each figure the phosphatidylcholine concentration is 30 mM.
actually the definition of the dilute regime for rodlike molecules.26 This would also imply that rheology and NMR data are not in contradiction with each other. Simply, NMR is sensitive to molecular mobility while viscosity is a direct measure of the overall three-dimensional entanglement. In view of these data, a 31P NMR T2 study was also performed at high wo values using a higher lecithin concentration. Results are shown in Figure 6a. Note that at 30 mM lecithin concentration a broad phosphorus line width of ca. 100 Hz is observed, even at wo ) 20 where the viscosity is low. When the gel is formed, spin-spin relaxation time values drop down to 5-7 ms, pointing to a loss of molecular mobility from wo ) 6 to wo ) 20. The comparison of viscosity and NMR data shows that in general the two sets of data correlate well in the region of low wo whereas in the high wo region serious discrepancies are present: here, the viscosity sharply decreases while the 31P line
Phosphatidylcholine Organogels
J. Phys. Chem., Vol. 100, No. 37, 1996 15215
Figure 8. Zero shear viscosity as a function of wo for 30 mM soybean lecithin (a), 30 mM DOPC (b), and 30 mM POPC (c) cyclohexane/ water systems.
TABLE 2: DOPC (30 mM)/Cyclohexane/Water System and POPC (30 mM)/Cyclohexane/Water System: wo Dependence of the 1H NMR Line Width (Values in Hz) of the Resonances for the Methylene Protons (A, C, C′, M, N) and of the Methyl Protons R of the Glycerol and Head Group Part of the Lecithinc Figure 7. wo dependence of the 31P NMR T2 spin-spin relaxation time for the soybean lecithin (160 mM)/cyclohexane/water system at different magnetic fields (32 MHz (b), 81 MHz (1), and 162 MHz (9)). Insert shows the expanded scale relative to the measurements performed at 162 MHz.
width and T2 still suggest the presence of stiffened structures. As mentioned before, since NMR measures the molecular mobility, whereas viscosity reflects the overall, gross behavior of the material, a perfect correlation in all concentration domains cannot be expected.27 Let us consider the physical meaning of 31P NMR relaxation times in more detail. As is well-known, two major terms contribute to the phosphorus relaxation process, namely, the dipole-dipole and the chemical shift anisotropy mechanism.28,29 This second mechanism becomes dominant at higher magnetic fields, whereas the dipole-dipole term is more efficient at low field. To clarify the extent of these mechanisms in lecithin gels, T1 and T2 measurements were performed at various magnetic field intensities. Results are shown in Figure 7 where 31P NMR T2 spin-spin relaxation times, measured at three different frequencies (i.e., 32, 81, and 162 MHz), are reported as a function of wo. The maximum in T2 values is always observed in the same wo region, between 1 and 2. At wo ) 8, where high shear viscosity is observed,5 the T2 trend observed at 162 MHz is similar to that previously reported at 81 MHz but with much shorter values; T2 values, measured at 32 MHz, are longer, going from 100 to 19 ms. In all cases, when gel formation occurs, an asymptotic behavior is reached. Note that the difference between values at the asymptote and at the maximum increases by decreasing the frequency. Thus, the relaxation mechanism responsible for the observed T2 shortening upon gel formation is mostly a dipole-dipole contribution, which reflects well the loss of mobility in the phosphate group. 31P NMR T relaxation times as a function of w were also 1 o measured, but no significant dependence on wo and/or the magnetic field was found. Since all T1 values lie within the 0.8-1.3 s range (data not shown), they are not a suitable source of information. B. Cases of DOPC, POPC, DPPC, and DLPC. To understand the influence of the molecular structure on gel formation, we have investigated organogels formed from welldefined, single molecular lecithin species. In this regard, one should recognize that such gels are not necessarily stable in the same temperature range as soybean lecithin gels. Let us consider first the two systems DOPC/cyclohexane/ water and POPC/cyclohexane/water as examined by 31P NMR. Both lecithins form stable gels at 300 K, but the preparation and storage of these systems for high values of wo must be carried out at temperatures not lower than 293 K, otherwise the systems become cloudy. NMR provides a sensitive method
wo 0
1
2
3
4
6
8
10
12
14
16
b 58 55 50 27 12
b 54 52 49 28 12
b 60 56 44 32 16
b 49 48 43 30 16
b 50 46 43 32 13
b 48 69 39 32 14
b 68 73 53 40 19
b 62 68 45 38 17
b a a a a a
b 54 50 54 38 24
A C C′ M N R
a a a a a a
23 20 25 20 15 7
31 27 29 25 17 8
53 26 30 24 19 8
DOPC 59 b 29 33 35 40 25 32 17 17 8 10
A C C′ M N R
16 27 17 17 34 13
31 30 35 33 25 9
38 28 31 36 20 9
59 30 37 28 19 7
POPC 111 b 31 44 43 64 30 37 20 25 9 10
a Not determined. b Not measurable (signal too broad). c Labeling of the different protons is as shown for soybean lecithin in Figure 1.
for detecting phase separation. In fact, when demixing occurs, two almost overlapped resonances instead of a single one are observed in the 31P NMR spectrum.11 For example, in DOPC/cyclohexane/water at wo ) 20, two different transparent phases coexist, one with a line width of ∼119 Hz, well within the gel region, and another with a much narrower line width, ∼14 Hz, i.e., corresponding to the line width observed for samples having wo ) 3-4. None of these two lines arises from monomeric, nonassociated DOPC. In this case, in fact, a line width of ∼7 Hz would be observed. In parts b and c of Figure 6, for both DOPC and POPC, the 31P NMR line width and T as a function of w are reported. In 2 o both cases the phospholipid concentration is 30 mM. The line width broadening and the corresponding T2 shortening reflect the gelation process (formation of long cylindrical micelles); the maximum of stiffness is reached at wo ≈ 12-14. Within the range wo ) 4-12, the phosphorus line width broadening correlates well with the increase in zero shear viscosity as shown in Figure 8. The behavior of the DOPC or POPC/cyclohexane/water system is comparable to the soybean lecithin/cyclohexane/water system. The good agreement among these three systems is verified by the similarity of the corresponding line width in the respective proton spectra (Table 2). In analogy to the case of soybean lecithin/cyclohexane/water (see Figure 2), a remarkable stiffening of the two protons connected to the sn-1 carbon atom of the glycerol backbone (A, A′) is observed. The influence of surfactant concentration has been also considered. 1H NMR line widths were measured for two different POPC concentrations, 30 and 160 mM within the wo range 1-4 (at larger wo values the lines are too broad to be correctly analyzed) (see Figure 2S in Supporting Information). As observed in the case of soybean lecithin, the higher the
15216 J. Phys. Chem., Vol. 100, No. 37, 1996 surfactant concentration, the broader the line width. Again, at wo ) 3 and 160 mM, the resonance due to A, A′ disappears, while at 30 mM and at the same wo value, it is still observable. NMR data are thus in qualitative agreement with rheological measurements,20 i.e., the higher the concentration, the easier the formation of transparent and viscous gels. To gain a more detailed insight into local motions within the phospholipid molecules, 13C spin-lattice relaxation times (T1) have been measured at 50.33 MHz for most of the POPC resonances at 160 mM and wo ) 0-6. These data are shown in Table 1S in Supporting Information. All 13C NMR resonances remain clearly observable even at wo > 4, where the gels are very viscous. Only the glycerol resonances broaden to the extent that T1 measurements become impracticable. Upon further addition of water, 13C NMR T1 relaxation times barely change: in particular T1 values due to double bonds remain almost constant in the range 0.5-0.6 s. The only clear effect can be observed at 15.89 ppm on the resonance due to the terminal methyl: upon addition of water T1 clearly shortens, going from 4.1 to 2.7 s. Note that T1 spin-lattice relaxation is mostly due to reorientation of the dipole of the attached proton so that it is very sensitive to local motions. For the organogels, different motions contribute to the total relaxation:28,29 an overall motion that encompasses the macroscopic properties of the system, and which is clearly reflected in the viscosity, and localized motions that are still observable after gel formation. T1 relaxation responds more to these latter motions and as such is not very useful for studying the gelation process per se. Let us now consider the case of DPPC. With respect to the temperature of the organogel formation, POPC and DOPC behave quite similarly to each another and to soybean lecithin, whereas DPPC behaves differently. In particular, the DPPC/ cyclohexane/water system does not form a gel at room temperature. It does form a gel, however, at a temperature above ∼313 K. In this regard, one should remember that aqueous bilayer surfactant aggregates are characterized by distinct transition temperatures of which perhaps the most relevant is the one which is responsible for the main transition (Tm) from a crystalline (or solid) analogue state to a fluid analogue state.30 In general, the transition temperature Tm increases by decreasing the number of double bonds in the fatty chains. The physical properties of liposomes, for example, such as permeability and overall stability, are very different13 depending on whether one operates above or below Tm. The importance of Tm in surfactant gel aggregates is less known. It seems that gels can be formed only in the fluid analogue state, namely above Tm, mostly because below such temperature the surfactant cannot be solubilized. In fact, in the case of soybean lecithin, POPC, and DOPC, Tm is around 258, 270, and 255 K respectively,31 while for DPPC Tm is ∼314 K.31 The critical temperature above which the three components DPPC, cyclohexane, and water form a transparent system depends on wo. 31P NMR T values at 333 K in the full w range have been 2 o measured (Table 1). Owing to the high operational temperature, T2 values are per se longer than the corresponding ones measured at 300 K in the other systems. The qualitative behavior is similar to the data observed for soybean lecithin, DOPC, and POPC gels: an increase of T2 upon addition of the first water molecules and a sharp decrease afterward. At wo ) 8-10, where the gelation occurs, T2 values are very short, less than 40 ms. In the full wo range, T2 values span over more than 1 order of magnitude.
Capitani et al. 1H NMR spectra were also recorded as a function of the temperature and wo. The line widths due to resonances AA′ are the most affected by the gelation process. Upon gelation, a significant broadening of this resonance is observed and, as expected, the lower the temperature, the more pronounced the effect (see Figure 3S in Supporting Information). Finally, the system DLPC/cyclohexane/water has been investigated. Tm for DLPC31 is 220 K. The formation of viscous and transparent gels is observed at 280 K, once again with the highest viscosity at wo ) 10-12. These gels eventually melt at a slightly higher temperature. In Figure 6d, as a function of wo, the phosphorus line width and the spin-spin relaxation times are reported. Owing to the low temperature, an increased viscosity of the solution itself (DLPC and cyclohexane) is observed; at wo ) 0 (no added water), the measured 31P NMR line width is in fact 18 Hz while at 300 K, it is 7 Hz. However, upon addition of water, the phosphorus line width increases from 18 up to 150-160 Hz, clearly reflecting the corresponding marked increase in the viscosity values. As a function of wo, the 31P NMR T2 measurements follow the usual trend: when the gel is formed, T2 values drop down to 5-7 ms. Owing to the lower temperature, the measured T2 times at low water content are shorter than the values previously reported in other systems. As a function of the water content, the releVant features in the proton spectra are shown in Figure 4S in Supporting Information. The corresponding proton spectra, as a function of wo, show an enormous broadening of the resonance A and a clear broadening of resonances P and Q, as in all other systems. To conclude, it is useful to directly compare all NMR data of synthetic lecithins with those of natural soybean lecithin. This we will do with POPC and DOPC, which are phenomenologically more similar to lecithin in the sense that gels are obtained at room temperature. The dependence of the 31P NMR line width as a function of wo for the three different systems (DOPC, POPC, soybean lecithin) shows that in the wo range 10-18, the line widths measured in synthetic phospholipids are always larger than the corresponding ones of the soybean lecithin system. See Figure 5S in Supporting Information. In the range wo ) 10-16, the line width in DOPC and POPC is ∼160 Hz compared to ∼100 Hz observed in soybean lecithin. Since local motions are mostly responsible for the observed line broadening, it can be concluded that independent of the viscosity values, organogels from synthetic chemically pure phospholipids show a higher local stiffening of the phosphate group than those from the naturally occurring soybean lecithin. The line width of protons C, M, A, and N in the three different gels can be compared. In the case of gels obtained at high wo values, line widths for POPC organogels are always broader than the corresponding line widths of the organogels from DOPC and natural soybean lecithin, directly reflecting the higher gel viscosity of the POPC system (see Figure 6S). However, DOPC line widths are always broader than the corresponding ones of soybean lecithin while the viscosity is about the same. This suggests that the stiffening of the polar head and the glycerol must be higher in DOPC than in soybean lecithin. All data, see Figure 3, show that gels from DOPC and soybean lecithin are stable in the same range of temperatures and that this range is much broader than the corresponding ones for the other synthetic lecithins used. In fact, DPPC is stable only at T > 313 K, while at lower temperatures precipitation occurs. POPC is stable at T > 286 K, below which the system becomes cloudy. DLPC is stable only in the small range 273-
Phosphatidylcholine Organogels 283 K; in fact, at T < 273 K, precipitation occurs, while at T > 283 K the system melts. In general, the comparison of all these data shows that DOPC is a closer model of soybean lecithin than the other synthetic phosphatidylcholines. Even in this case, however, caution is required in pushing the comparison too far, since local motions in the glycerol moiety are more hindered in DOPC than in soybean lecithin. Concluding Remarks Despite its relative molecular heterogeneity, soybean lecithin has a clearly interpretable NMR spectrum that allows one to identify the local group mobility attending the formation of organogels. As a result of this analysis, one obtains a detailed picture of the dynamic behavior of the entire lecithin molecule. Several proton resonances show a significant broadening of the line width during the gelation process, which is per se not surprising. What is surprising is the large difference in dynamic behavior of nuclei, which are geometrically very close to each other, as is the case of the methylenes of the glycerol moiety. This indicates that the gelation process is nucleated by the rigidification of only selected atomic groups of the lecithin molecule. These are centered around the glycerol moiety, whereas the alkyl chains show a higher mobility even after gelation has taken place. A similar glycerol backbone stiffening has also been observed for lecithin and other phospholipids in aqueous micelles and liposome dispersions.32,33 Therefore, it seems to be a general property of the phospholipid packing in aggregates, independent of whether they are formed in water or in organic solvents. Also, the large mobility of water under conditions in which the gel displays a very large viscosity is quite interesting.11 This suggests that water, although sequestered in the cylindrical wormlike micelles, can flow very freely inside them. The capability of distinguishing between gelforming and nongel-forming solvents is another additional insight provided by NMR data. The information obtained for the soybean lecithin/cyclohexane/water system is in good agreement with the detailed analysis of the synthetic lecithin systems POPC/cyclohexane/water and DOPC/cyclohexane/water. All these systems behave very similarly to soybean lecithin. However, the molecular stiffening during the gelation process is stronger in the case of DOPC and POPC than in the case of soybean lecithin. This observation most probably reflects the packing properties within the initial cylindrical reverse micelles, which is optimal in the case of the chemically pure synthetic lipids. In the case of DPPC, the importance of the temperature is evident, since below 313 K DPPC is not soluble in cyclohexane. However, at higher temperature, the NMR properties and the gelation process are again comparable to those of POPC, DOPC, or soybean lecithin. Generally speaking, the data obtained here are in agreement with the accepted model for lecithin gels based on the entanglement of long rodlike micellar aggregates. However, the comparison of NMR and rheology data indicates also the existence of conditions under which there is still a remarkable rigidity of portions of the lecithin molecule prior to reaching conditions of high viscosity. Acknowledgment. We thank PD Dr. P. Schurtenberger for helpful discussions. This work was partly supported by the Swiss National Science Foundation, Project Number 4024-027 203. A NATO contribution (Grant 921409) is acknowledged.
J. Phys. Chem., Vol. 100, No. 37, 1996 15217 Supporting Information Available: Figures 1S-6S showing 1H NMR spectra and wo dependences and Table 1S listing 13C NMR relaxation times (7 pages). Ordering information is given on any current masthead page. References and Notes (1) Scartazzini, R.; Luisi, P. L. J. Phys. Chem. 1988, 92, 829-833. (2) As accepted by the IUPAC-IUB commission of Biochemical Nomenclature, lecithin and phosphatidylcholine are used synonymously as general terms to designate any 1,2-diacyl-sn-glycero-3-phosphocholine molecule. Silvius, J. R. In Phospholipids Handbook; Cevc, G., Ed.; Marcel Dekker, Inc.: New York, 1993; pp 1-22. (3) Schurtenberger, P.; Scartazzini, R.; Luisi, P. L. Rheol. Acta 1989, 28, 3782-381. (4) Schurtenberger, P.; Scartazzini, R.; Magid, L. J.; Leser, M. E.; Luisi, P. L. J. Phys. Chem. 1990, 94, 3695-3701. (5) Schurtenberger, P.; Cavaco, C. Langmuir 1994, 10, 100-108. (6) Willimann, H.; Luisi, P. L. Biochem. Biophys. Res. Commun. 1991, 177, 897-900. (7) Willimann, H.; Walde, P.; Luisi, P. L.; Gazzaniga, A.; Stroppolo, F. J. Pharm. Sci. 1992, 81, 871-874. (8) Nastruzzi, C.; Gambari, R. J. Controlled Release 1994, 29, 5362. (9) Dreher, F.; Walde, P.; Luisi, P. L.; Elsner, P. Skin Pharmacol. 1996, 9, 124-129. (10) Capitani, D.; Segre, A. L.; Sparapani, R.; Giustini, M.; Scartazzini, R.; Luisi, P. L. Langmuir 1991, 7, 250-253. (11) Capitani, D.; Rossi, E.; Segre, A. L.; Giustini, M.; Luisi, P. L. Langmuir 1993, 9, 685-689. (12) Product information provided by Lucas Meyer GmbH, Hamburg, Germany. (13) New, R. R. C. In Liposomes: a Practical Approach; New, R. R. C., Ed.; Oxford University Press: Oxford, 1992; pp 1-32. (14) Walde, P.; Luisi, P. L. Biochemistry 1989, 28, 3353-3360. (15) Farrar, T. C.; Becker, E. D. Pulse and Fourier Transform NMR. Introduction to Theory and Methods; Academic Press: New York, 1971. (16) Sykora, S. Program “FIT”, part of the software of the spectrometer “Spinmaster”. (17) Bax, A.; Davis, D. G. J. Magn. Reson. 1985, 65, 355-360. (18) Bax, A.; Griffey, R. H.; Hawkins, B. L. J. Magn. Reson. 1983, 55, 301-315. (19) Wehrli, F. W.; Wirthlin, T. Interpretation of Carbon-13 NMR Spectra; Heyden & Son Ltd: London, 1976; p 269. (20) Scartazzini, R. Organogels from Lecithin Water-in-Oil Microemulsions: Phase Behaviour, Viscoelastic Properties and Applications. Dissertation, ETH-Zu¨rich, Nr. 9186, 1990. (21) Elworthy, P. H. J. Chem. Soc. 1959, 1951-1956. (22) Elworthy, P. H.; McIntosh, D. S. J. Phys. Chem. 1964, 68, 34483452. (23) Barclay, L. R. C.; MacNeil, J. M.; VanKessel, J.; Forrest, B. J.; Porter, N. A.; Lehman, L. S.; Smith, K. J.; Ellington, J. C. J. Am. Chem. Soc. 1984, 106, 6740-6747. (24) Eicke, H.-F. Top. Curr. Chem. 1980, 87, 85-145. (25) Schurtenberger, P. Personal communication. (26) Doi, M.; Edwards, S. F. The Theory of Polymer Dynamics, Oxford Science Publication; Clarendon Press: Oxford, 1986. (27) The question arises whether this discrepancy might also be due to the different handling of the samples. During the viscosity measurements, the sample must be submitted to a shear force, and this might in principle be enough to partially destroy a metastable network. However, this seems to be unlikely based on the fact that shear viscosity measurements have been analyzed in the linear viscoelastic regime of the system.4 (28) Abragam, A. The Principles of Nuclear Magnetism; Oxford University Press: Oxford, 1961. (29) So¨derman, O.; Walderhaug, H.; Henriksson, U.; Stilbs, P. J. Phys. Chem. 1985, 89, 3693-3701. (30) Very often, a rather confusing terminology is used in the literature to describe the different states of the lipids. The fluid analogue state is often called the “liquid crystalline” state, while the solid analogue state is called the “gel phase”.13 In order to avoid confusion with the term “gel” used for describing high-viscosity soft materials like the lecithin organogels on which we report here, we use the clearer term “solid analogue” state to denote the physical arrangement of the lipid molecules below the main phase transition temperature. (31) Cevc, G. In Phospholipids Handbook; Cevc, G., Ed.; Marcel Dekker, Inc.: New York, 1993; pp 939-956. (32) Strenk, L. M.; Westerman, P. W.; Doane, J. W. Biophys. J. 1985, 48, 765-773. (33) Hauser, H.; Pascher, I.; Sundell, S. Biochemistry 1988, 27, 9166-9174.
JP960811I