Intermolecular Forces in Spread Phospholipid ... - ACS Publications

Aug 6, 2004 - 119, Stainbank Rd., Kendal, Cumbria, LA9 5BG, England, and Department of Chemical. Engineering, Princeton University, Engineering ...
1 downloads 0 Views 61KB Size
Langmuir 2004, 20, 7493-7498

7493

Intermolecular Forces in Spread Phospholipid Monolayers at Oil/Water Interfaces James Mingins and Brian A. Pethica* 119, Stainbank Rd., Kendal, Cumbria, LA9 5BG, England, and Department of Chemical Engineering, Princeton University, Engineering Quadrangle, Princeton, New Jersey 08544 Received February 2, 2004. In Final Form: June 1, 2004 The lateral intermolecular forces between phospholipids are of particular relevance to the behavior of biomembranes, and have been approached via studies of monolayer isotherms at aqueous interfaces, mostly restricted to air/water (A/W) systems. For thermodynamic properties, the oil/water (O/W) interface has major advantages but is experimentally more difficult and less studied. A comprehensive reanalysis of the available thermodynamic data on spread monolayers of phosphatidyl cholines (PC) and phosphatidyl ethanolamines (PE) at O/W interfaces is conducted to identify the secure key features that will underpin further development of molecular models. Relevant recourse is made to isotherms of single-chain molecules and of mixed monolayers to identify the contributions of chain-chain interactions and interionic forces. The emphasis is on the properties of the phase transitions for a range of oil phases. Apparent published discrepancies in thermodynamic properties are resolved and substantial agreement emerges on the main features of these phospholipid monolayer systems. In compression to low areas, the forces between the zwitterions of like phospholipids are repulsive. The molecular model for phospholipid headgroup interactions developed by Stigter et al. accounts well for the virial coefficients in expanded phospholipid O/W monolayers. Inclusion of the changes in configuration and orientation of the zwitterion headgroups on compression, which are indicated by the surface potentials in the phase transition region, and inclusion of the energy of chain demixing from the oil phase will be required for molecular modeling of the phase transitions.

Introduction The high cohesive interactions between monolayer chains observed at the air/water (A/W) interface are much reduced at the oil/water (O/W) interface, leading to higher surface pressures (Π) at a given monolayer density and the elimination of one phase transition. The resultant continuous isotherms, from high Π at low areas/molecule (A) and out to near-ideal two-dimensional gas properties at large areas and over a range of temperatures allow a full thermodynamic analysis of the molecular interactions, which is difficult to achieve at the air/water (A/W) interface. Study of the effects of changing chain length and headgroup facilitates identification of the component forces in the monolayers. Despite the greater technical difficulties compared with the A/W interface, the effort is worthwhile because thermodynamic properties are made available for testing and developing two-dimensional molecular models relevant to a variety of colloid systems and to the forces acting between major components of biomembranes. Phillips and Rideal1 showed that two-dimensional phase transitions could occur in a spread monolayer at the O/W interface. With introduction of the Brooks2-4 frame, the existence of these transitions was confirmed by more accurate isotherms in monolayers comprised of an equimolar mixture of cationic and anionic single-chain molecules. The transitions were attributed to the attractive electrostatic interactions of the ionic headgroups. It was therefore surprising when Demel5 found a pronounced * Princeton University. (1) Phillips, J. N.; Rideal, E. K. Proc. R. Soc. London, Ser. A 1955, 232, 149. (2) Brooks, J. H.; Pethica, B. A. Trans. Faraday Soc. 1954, 60, 208. (3) Brooks, J. H.; Pethica, B. A. Proceedings of the 4th International Congress on Surface Activity: 1964; p 191. (4) Brooks, J. H.; Pethica, B. A. Trans. Faraday Soc. 1965, 61, 571. (5) Demel, R. A. Unpublished results.

phase transition in a spread monolayer at the O/W interface with a neutral zwitterionic phospholipid. The subsequent experimental program on O/W phospholipid monolayers by ourselves and colleagues6-15 confirmed these transitions and demonstrated a marked dependence on temperature, chain length, and headgroup of the spread molecules. The data were confirmed by experiments in the United States and the United Kingdom, based on separate syntheses, analyses, and monolayer measurements. The way was opened for critical thermodynamic analysis of the contribution of the chains and headgroups to phospholipid interactions in two dimensions. Other workers have extended the range of phospholipids and oils examined, mostly using different monolayer methods.16-19 The systematic study of dipalmitoyl phosphatidyl ethanolamine (DPPE) monolayers spread (6) Taylor, J. A. G.; Mingins, J; Pethica, B. A.; Tan, B. Y. J.; Jackson, C. M. Biochim. Biophys. Acta 1973, 323, 157. (7) Yue, B. Y. J.; Jackson, C. M.; Taylor, J. A. G.; Mingins, J.; Pethica, B. A. J. Chem. Soc., Faraday Trans. 1 1976, 72, 2685. (8) Taylor, J. A. G.; Mingins, J.; Pethica, B. A. J. Chem. Soc., Faraday Trans. 1 1976, 72, 2694. (9) Mingins, J.; Taylor, J. A. G.; Pethica, B. A.; Jackson, C.M.; Yue, B. Y. T. J. Chem. Soc., Faraday Trans. 1 1982, 78, 323. (10) Pethica, B. A.; Mingins, J.; Taylor, J. A. G. J. Colloid Interface Sci. 1976, 55, 2. (11) Jackson, C. M.; Yue, B. Y. J. Adv. Chem. Ser. 1975, 144, 14. (12) Llerenas, E; Mingins, J. Biochim. Biophys. Acta 1976, 419, 381. (13) Taylor, J. A. G.; Mingins, J. J. Chem. Soc., Faraday Trans. 1 1975, 71, 1161. (14) Pickard, W. F.; Sehgal, K. C.; Jackson, C. M. Biochim. Biophys. Acta 1979, 552, 1. (15) Sehgal, K. C.; Pickard, W. F.; Jackson, C. M. Biochim. Biophys. Acta 1979, 552, 11. (16) Zhou, N. F.; Neuman, R. D. Colloids Surf. 1992, 63, 201. (17) Brezesinski, G.; Thoma, M.; Struth, B.; Mohwald, H. J. Phys. Chem. 1996, 100, 3126. (18) Hayashi, M.; Kobayashi, T.; Seimiya, T. Chem. Phys. Lipids 1981, 29, 289. (19) Thoma, M.; Mohwald, H. J. Colloid Interface Sci. 1994, 162, 340.

10.1021/la040016t CCC: $27.50 © 2004 American Chemical Society Published on Web 08/06/2004

7494

Langmuir, Vol. 20, No. 18, 2004

at the interface of water and three oils by Thoma and Mohwald19 is of particular interest since a wide temperature range was investigated. A related study of oil penetration into phospholipid monolayers at the A/W interface is also available.17 Since there are apparent differences between the thermodynamic conclusions and other aspects of some of these studies, and given the potential utility of reliable phospholipid monolayer data, it is worthwhile reviewing the available data to identify areas of agreement, the salient features to be addressed by theoretical models, and the leading requirements for new experiments. Experimental Section Although the measurement of surface pressure (Π) and surface potential (∆V) for spread monolayers has become more convenient with new instrumentation, acquisition of accurate reproducible isotherms still depends on the necessary standards of purity, monolayer stability, and quantitative spreading. For critical thermodynamic-quality data, demonstration of reversibility is also required. Methods for attaining such data have been studied and published,6-13 but not widely employed, presumably because they are time-consuming. With respect to O/W systems, which are particularly susceptible to contamination, it is necessary to check the water, salts and the as-received oils and monolayer substances; check to confirm agreement with established surface and interfacial tensions; check for monolayer substance stability in the spreading solvent; check for spreading losses and solvent retention; check by testing for monolayer desorption during compression, and check to ensure that the pressure measurements are free of artifacts such as nonzero contact angles with Wilhelmy plates, as discussed in detail in Taylor et al.13 These procedures were developed and applied during a program of studies of phospholipid monolayers (predominantly phosphatidyl cholines, PC) by the present authors and their colleagues.6-15 and we denote this body of data as JB to recognize the contributions of the late Dr. John Brooks to O/W studies. Further details specific to phospholipids are given in these papers. These precautions form part of the general standards for critical studies on liquid interfaces, which are widely accepted, if less widely practiced, and we note recent additional studies on experimental standards in surface chemistry, which give further evidence on the impact of impurities.20 We also observe that Zhou and Neuman16 used similar methods to those of the JB group with dipalmitoyl phosphatidyl choline (DPPC) at one temperature at the heptane/water interface and obtained an isotherm in good agreement with our results. The thermodynamic functions from the JB data set for the phase transitions considered in the present paper are mostly in Yue et al.7 and Mingins et al.,9 with extensions given below. The effect of changing the oil phase on phospholipid monolayer isotherms was first observed by Jackson and Yue11 and later by Hayashi et al.,18 the latter using a phosphate buffer solution as the aqueous phase. For the shorter n-alkane oils in particular, no effect of phospholipid chain length was observed at areas above the phase transitions, but the parameters of the phase transitions themselves were significantly modified. Interestingly, a substantial difference in the isotherms for spread monolayers of equimolar mixtures of octadecyl trimethyl ammonium bromide (OTAB) and sodium octadecyl sulfate (SOS) at the dilute NaCl solution interface with carbon tetrachloride or heptane was again only observed in the phase transition regions.3,4 The phase transitions in these mixed monolayers have properties similar to those of the phospholipids, which they resemble in having fully ionized positive and negative head groups at overall zero net charge. The isotherms for both the zwitterionic and mixed zero-charge monolayers show little or no dependence on the NaCl concentration. We now discuss the set of experiments of Thoma and Mohwald19 on the effect of temperature on spread DPPE O/W monolayers. These results for a 98% pure DPPE from 2 nm2/molecule to lower (20) Persson, C. M.; Claesson, P. M.; Lunkenheimer, K. J. Colloid Interface Sci. 2002, 251, 182.

Mingins and Pethica areas and over a range of temperatures at the three interfaces of water (pH 5.5, no added salts) with n-dodecane, n-hexadecane, and biscyclohexane (BCH), all used as received, are potentially a source of thermodynamic information, and a stimulus to experimental critique through comparison with related published data. We denote this data set as TM. The TM data were obtained by two spreading techniques with good agreement, suggesting that the chloroform used for spreading is not significantly retained. They state that there may be a zeroing error in Π of up to 2 mNm-1 in their results, which if constant would not affect their calculations of the enthalpies of the phase transitions. No comparisons were made with other published results, which are given below and which indicate that the TM data’s potential error in Π is probably small. Of further concern is that their isotherms were obtained by continuous compression, apparently at only one scan rate corresponding to about 17 min per experiment. This leaves open the questions of monolayer stability and loss. No traces are given for expansion, which would have given a further check. Taylor22 found slow instability at higher pressures at 20 °C for the lower chain length DMPE at the heptane/water interface. This finding would indicate likely monolayer dissolution at the highest temperatures and pressures of the TM series with DPPE and hexadecane oil. With regard to manometric methods, our experience with Teflon Wilhelmy plates (used in the TM experiments) was erratic, leading to the adoption of the completely oil-wetted carbon black-coated platinum plates used in our experiments. Despite these concerns, there are adequate grounds to reconsider the TM data for further comparisons. For example, continuous scans at the rate indicated will not be much in error for the pressure and area of onset of the transitions, and the data can be analyzed for the thermodynamic properties of the DPPE monolayer transitions, bearing in mind the qualifications made above. We find no other publication on precisely any one of their chosen conditions, but useful comparisons can be made with other PE studies at areas above the phase transitions since it is well established that in this region the isotherms of homologous phospholipids are not dependent on the lipid chain length and are very similar for several n-alkanes and some other oils.18,19 This allows a cross-check on the isotherms for monolayer spreading, retention, and manometry in the lower pressure ranges for the DPPE results from the TM (no salt) and JB (NaCl solutions) data sets since the isotherms for dimyristoyl phosphatidyl ethanolamine (DMPE)10 do not change with salt concentration. In fact, the two sets of PE isotherms give fair agreement at areas above the onset of the phase transitions. In contrast, the data of Hayashi et al.18 at 25 °C with DPPE at the interface of a 0.01 M phosphate buffer (pH 6, with added EDTA) against several n-alkanes show large differences from the TM and JB results in the high area regions of the isotherms as well as in the Π and A values for the transitions. It is known for DPPE monolayers at the A/W interface that ∆V at constant area and ionic strength (NaCl with either HCl or NaOH, no buffer salts) varies with pH, but is almost constant over a short pH range near neutral at higher salt concentrations.21 As the ionic strength is lowered, the region of constancy decreases, the crossover with the plateau at higher salt being close to pH 6. This behavior reflects the ionization of the amine group of the DPPE monolayer. It is reasonable to conclude that the TM data with salt-free water at pH 5.5 are for the zwitterionic form of DPPE, and hence in agreement with the earlier JB result for DMPE on salt solutions.10 The results of Hayashi et al.18 for DPPE at n-alkane interfaces with a phosphate buffer solution probably reflect interaction of the phosphate ions with the monolayer amine groups, the resulting electrostatic charging causing the observed expansion of the monolayer. The effect of EDTA on phospholipid monolayers has not yet been checked.

Thermodynamic Functions for the Phase Transitions A striking difference between the conclusions drawn from the TM and JB data sets on the temperature (21) Standish, M. M.; Pethica, B. A. Trans. Faraday Soc. 1968, 64, 1113. (22) Taylor, J. A. G. unpublished results.

Intermolecular Forces in Phospholipid Monolayers

dependence of the phase transitions is that the JB data for the PCs at the heptane/water interface showed no change in the heat capacity on compression between two areas embracing the transitions,9 whereas the TM data for DPPE were interpreted as showing a large change in a heat capacity (presumably at constant surface pressure) which varied markedly between the three oils studied. It will be shown that these differences arise from an incorrect use of the Clausius-Clapeyron equation for interpreting the TM data for the transitions. The Clausius-Clapeyron relation is strictly valid only for first-order transitions, for which Π should be constant across the transition region, and its frequent use for transitions in monolayers not showing Π constant is at best an approximation in some few cases. In the early stages of the JB studies, the Clausius-Clapeyron equation was also used, but this method was later abandoned and used only for comparative purposes since it was established that these transitions are definitely not first order.7,9,23 The evidence for this conclusion may be summarized as follows. The phospholipid syntheses and analyses were carried out in several laboratories in the United Kingdom and United States and the analytical data published. The samples were exchanged and monolayer measurements made in two collaborating laboratories with rigorous purification of heptane and water, and any discrepancies were addressed until full convergence between laboratories and samples was obtained, thereby minimizing error due to impurities.7,9 The possibility that the shape of the isotherms and the inconstancy of the phospholipid surface pressures across the transition regions are due to impurities is ruled out by the fact that the deviations from firstorder behavior are greater for the lower chain lengths and also that the deviations increase with temperature for all chain lengths.23 Instead of using the Clausius-Clapeyron equation, a rigorous analysis of the transitions was made by first obtaining the Helmholtz free energy of compression across the transitions (∆F) by integration of the isotherms as ∆F ) -∫ΠdA between two areas A1 and A2 chosen close to and spanning the region of all the phase transitions to be compared. For these phospholipids, Π at areas above the onset of any transition does not vary with the chain length, and not greatly with temperature, so that the precise choice of the reference area in this region is not critical. The chosen dense standard area can be closely defined by inspection on the steep part of the isotherms with little consequent uncertainty in ∆F. The entropy change across the transitions then follows from -∆S ) d∆F/dT, and the energy from ∆F ) ∆U - T∆S. The corresponding enthalpy (∆H) could be calculated from ∆H ) ∆U + ∆(ΠA) between the chosen reference areas (0.47 and 2.0 nm2/molecule for the PCs). Since the heat capacity change at constant Π cannot be obtained from the surface enthalpies at constant areas, they are not considered further. The change of Gibbs free energy can be obtained from ∫AdΠ between two pressures spanning all the phase transitions to be compared, but this offers no advantage over using the densities, and the choice of the dependent pressure variable is both less secure and less convenient for reference standards. For each of the PC monolayers at the heptane/water interface in the JB set, ∆F varies linearly with temperature. Thus, ∆S and ∆U are constant for each PC over the experimental range. As the chain length is increased, ∆S varies by Rln3/CH2 group in each chain (where R is the (23) Bell, G.; Mingins, J.; Levine, S. J. Chem. Soc., Faraday Trans. 2 1978, 74, 223.

Langmuir, Vol. 20, No. 18, 2004 7495

Figure 1. ∆F for the DPPE transitions at three oil/water interfaces. Experimental data of Thoma and Mohwald19.

gas constant), indicating that the entropy change is largely determined by the configurational restriction of the chains in the condensed phospholipid phase. The invariance of ∆E with temperature implies no heat capacity change (at constant reference areas) across the transition region. The same method via the Helmholtz free energy was applied to the TM data for DPPE, graphically integrating the isotherms directly from the published isotherm traces19 between 2.0 and 0.40 nm2/molecule (the dense phase area for PEs from both the TM and JB sets). The ∆F values were reproducible in repeat integrations, at 1% or better. The results are given in Figure 1 with the lines shown obtained by regression analysis. It will be seen that ∆F varies linearly with T for biscyclohexane and dodecane. For hexadecane, the ∆F dependence on temperature is only fairly linear, but with a trend perhaps reflecting stronger chain interaction when the chains of the lipid and oil are the same, as suggested by the authors. The linear regression lines shown in Figure 1 are taken for further comparisons, implying ∆S and ∆U are constant for the DPPE transitions, as was established for the DPPC and other PC monolayers studied in the JB set for the heptane interface. The TM DPPE results are shown in Table 1 for comparison with results from the JB DPPC and other monolayers as discussed below. Readers should note the integration limits for these functions, which depend on the molecular groups (PE, PC etc.) forming the phase diagrams. In the JB set, the only PE data are for DMPE at the heptane/water interface at two temperatures.10 At the upper temperature (20 °C), the isotherm given does not quite complete the transition due to the onset of slow desorption, the experimental points ending at 0.46 nm2/ molecule. The data thus give ∆F at two temperatures for compression from 2.0 to 0.47 nm2/molecule, the range for the PC transitions. For the nearest comparison with the TM set, ∆F values were also obtained for the same compression range for DPPE at the dodecane/water interface. These ∆F values are almost linear with temperature. Taking the linear regression again, these values of ∆F, ∆S, and ∆U from 2.0 to 0.47 nm2/molecule limits from the TM set for DPPE at the dodecane/water interface are given in Table 1 together with the JB results for DMPE at the heptane/water interface From the TM DPPE data, ∆S and ∆U for the two low-area compression limits appear virtually unchanged, which is improbable, and again suggests experimental error. If ∆S for the PE transitions is independent of temperature and varies linearly with

7496

Langmuir, Vol. 20, No. 18, 2004

Mingins and Pethica

Table 1. Thermodynamic Functions for Compression Across the Phase Transitions of Spread O/W Monolayers of Zwitterionic Phospholipids and of 1:1 Mixed Monolayers of n-octadecyl Anionic and Cationic Surfactants oil phase

monolayer

1. biscyclohexane 2. hexadecane 3. dodecane 4. dodecane

DPPE DPPE DPPE DPPE

5. heptane 6. heptane 7. heptane 8. heptane 9. carbon tetrachloride

DMPEa,b DPPCa DSPC OTAB/SOS pair OTAB/SOS pair

compression

∆F

TM data19 2.0 to 0.40 2.0 to 0.40 2.0 to 0.40 2.0 to 0.47

14.9 9.1 11.0 9.9

JB data3,9,10 2.0 to 0.47 2.0 to 0.47 2.0 to 0.47 2.0 to 0.55 2.0 to 0.55 nm2/molecule

13.3 17.1 15.0 11.7 13.3 kJ/mole

-∆S

-∆U

255 233 231 228

61.1 60.4 57.8 58.6

147 266 303 312 288 J/°K/mole

30.5 62.2 75.4 81.0 73.0 kJ/mole

∆F at 25 °C: ∆S and ∆U are invariant with temperature over the experimental ranges. The compression range and thermodynamic functions for the 1:1-mixed OTAB/SOS monolayers are for molecular pairs. a Short extrapolation to 25 °C. b ∆S and ∆U from two experimental temperatures only.

chain length, as found with the PCs, the value for DMPE should be 4Rln3 (36.5 J mole-1 K-1) less negative than for DPPE for the same area limits. In fact, the DMPE value with heptane in Table 1 is lower by 80 J mole-1 K-1 compared to the TM results for DPPE with dodecane. Since the TM results for DPPE shown in Table 1 are similar for the two n-alkanes, the discrepancy is notable. Further experiments on a range of PE chain lengths at several temperatures are required to settle whether the TM and JB sets are in full accord for the PEs, but we conclude that the two sets of data are sufficently consistent for the water/ n-alkane interfaces to make further analysis of the available data useful. Also shown in Table 1 are the functions for DPPC9 at the heptane/water interface for comparison with the TM DPPE results, which show very similar values for ∆S and ∆U at both the dodecane and hexadecane/water interfaces. Both ∆S and ∆U are negative and of smaller magnitude for DPPE than for DPPC. For all cases where comparison with phosphatidyl cholines is available, the PE isotherms always show a little lower Π for a given A above the transitions and a much lower Π at and above the onset of transitions (see Taylor et al.6 for a direct comparison of DPPE and DPPC at the heptane/water interface). In the absence of better experimental evidence on the chainlength dependence of the thermodynamic functions for PEs, speculation is best limited to concluding qualitatively that PE headgroups are less repulsive than PC headgroups at comparable areas. This is in accord with the fact that at areas above the transitions where the isotherms are probably independent of chain length for the PEs (and certainly for the PCs), the second virials for PEs are less positive than for PCs.24 This difference in headgroup repulsion appears to be even larger in the condensed phases. The solvent effects seen in the TM study and by Hayashi18 extend the earlier information from JB collaborators. The isotherms of Pickard et al.14,15 for distearoyl phosphatidyl choline (DSPC) at the 2,2,4 trimethylpentane/0.01M NaCl interface again show the pattern of agreement with DSPC at the heptane interface at areas above the transition, with different values of Π and A at the onset of the transition.. Analysis of the temperature dependence of the transition with trimethylpentane as the oil phase is precluded by inconsistencies in some of the isotherms. Jackson and Yue11 also examined solvent effects for DSPC O/W monolayers. They found similar pressures at areas above the phase transitions for a range of lower (24) Mingins, J.; Stigter, D.; Dill, K. A. Biophys. J. 1992, 61, 1603.

n-alkanes and isooctane, and small but significant differences in the Π and A values at the transitions at two temperatures. Cyclohexane gave substantially higher pressures than the n-alkanes for DSPC monolayers at areas above the transitions, reminiscent of the TM finding for biscyclohexane, and may indicate a systematic effect with cyclic hydrocarbons. Cyclohexane and n-undecane oil phases showed the largest differences from the evennumbered alkanes at the transitions. Approximate estimates from the data in the paper for isooctane and the n-alkane homologues up to n-nonane at two temperatures gave entropies of transition similar to the value for DSPC from the extensive data for PCs at the heptane/water interface discussed above.9 The thermodynamic parameters for phospholipid monolayer phase transitions at O/W interfaces are usefully compared with corresponding data over a range of temperatures for equimolar mixed films of sodium octadecyl sulfate (SOS) and octadecyl trimethylammonium bromide (OTAB) spread at the interfaces of heptane and carbon tetrachloride with dilute NaCl solutions.3 These monolayers have no net charge and the anion/cation pairs can be regarded as dissociated zwitterion analogues. They show phase transitions similar to those for the corresponding chain-length phospholipids, and once again Π at areas above the transitions is similar for both oils, with the phase transition onset parameters definitely different. The thermodynamic functions for the transitions have been recalculated via ∆F as described above (with reference areas for integrating the isotherms of 0.51 and 2.0 nm2/molecule) and are shown in Table 1. The striking result is that the -∆S/mole of OTAB/SOS pairs is 312 and 288 J mole-1K-1 for the heptane and carbon tetrachloride interfaces, respectively. These entropies are close to the experimental (heptane/water) result of 303 J mole-1 K-1 for DSPC, a phospholipid of the same n-alkyl chain length, a convincing confirmation of the importance of the chain conformations for the entropies of transition. Future molecular models will take note of the differences in the molecular areas in the several dense phases, namely 0.40, 0.47, and 0.51 nm2/molecule for the PEs, PCs, and longchain cation/anion pairs. We note that at high areas, the mixed SOS/OTAB monolayers show a positive second virial coefficient of 0.32 nm2/chain, corresponding well to the close packed area of the two species. This result is in sharp contrast with the large second virial coefficients for the phospholipid monolayers. Furthermore, the phase transition in these mixed films can be qualitatively accounted for by the electrostatic attraction of the cation/anion array in the surface, again in contrast with the phospholipids,

Intermolecular Forces in Phospholipid Monolayers

for which the zwitterion headgroups appear to be play a repulsive role, as discussed below. Discussion The comparison of the entropies of the phase transitions for the PE and PC lipids of the same chain length at n-alkane/water interfaces, and the established linear chain-length dependence of ∆S for PCs confirm the importance of chain conformation in the formation of the dense phases. The suggestion19 that chain mixing in dilute monolayers at alkane/water interfaces can take on the character of a weak association when the oil and phospholipid chain lengths match, thereby contributing to the free energy of the transition, is weakly supported by the thermodynamic results discussed above, but has support from TM monolayer structural studies.20 Definite thermodynamic support of the significance of chain-matching in O/W phospholipid monolayers is seen at the heptane/ water interface in a binary mixed monolayer of two homologous PCs differing by six methylene groups per chain.6 Two distinct phase separations are observed, each homologue forming its own dense phase at surface pressures near those for the single component monolayers. Further confirmation of the significance of chain packing comes from a comparison of monolayers of the 1,3 and 1,2 isomers of distearoyl PCs at the heptane/water interface at 20 °C.12 The two isotherms are quite distinct over the entire area range, with the phase change onset at similar areas but very different Π values. The interaction of pairs of PC headgroups can be estimated from comparison of the energies and entropies of compression across the phase transitions.9 The entropies vary linearly with chain length and extrapolate to zero at a chain length of one carbon in each chain, for which there is no chain configurational entropy. The energies also vary linearly with chain length and extrapolate to zero at a chain length of 9-10 carbons in each chain, implying a repulsive energy term for the PC headgroup of 40-45 kJ/mole on compression from 2.0 to 0.47 nm2/molecule. A direct estimate of the repulsive energy contribution for PE zwitterions in the dense phase is unavailable in the absence of isotherms for a range of chain lengths. If we assume the same chain length dependence as for the PCs, the ∆U would be about 20 kJ/mole more repulsive for the PE zwitterions than the PCs in the dense phase. This does not necessarily contradict the conclusion that PE is less repulsive than PC for areas at and above 0.47 nm2/ molecule, given the substantially closer packing of the PE zwitterions in the dense phase as compared to PCs and the probably different headgroup reorientations on compression. There is striking evidence that the PE and PC headgroups do not readily mix in monolayers at O/W interfaces. An early experiment with an equimolar mixed monolayer of DPPC and DPPE at the isooctane/water interface at 1 °C showed an increase in Π over the (almost equal) pressures for the individual lipids alone at high areas, and two distinct phase separations at pressures approximating those for the phase transitions of the separate lipids.6 Demixing on compression is thus dependent on the lipid chain length for two phospholipids with the same headgroup, and on the headgroup for two phospholipids of the same chain length.. Further experiments using critical methods with mixed phospholipid O/W monolayers will prove valuable in quantifying these chain and headgroup interactions, which are clearly relevant to the mixed-lipid arrays in biomembranes. Molecular theories for the phase transitions should strictly include the effects of water structure changes at

Langmuir, Vol. 20, No. 18, 2004 7497

the headgroups as they adopt dense-phase conformations. This combined reorientation of water and headgroups causes changes in the mean normal component of dipole moment of the monolayers, as is seen from the surface potentials, which track the phase transitions closely.10,14,15 Further data on the surface potentials as a guide to these reorientations at the phase transitions are being prepared for publication. Zwitterion orientation for PC and PE molecules at the O/W interface has been calculated from the electrostatics of zwitterion pairs and groups, including images, and applied to give a good account of the two-dimensional virial coefficients in dilute monolayers by Stigter and coworkers24-27 The zwitterions are slightly out of the plane parallel to the interface, leading to a net repulsion accounting for the virial coefficients, with the positive choline group residing increasingly in the low-dielectric oil with the increase in temperature, accounting for the larger repulsion seen with the PCs. This conclusion for dilute phospholipid monolayers is supported by the finding that 1,2-dipalmitin adsorbed at the heptane/water interface gives a two-dimensional second virial coefficient that is positive, but significantly lower than the close-packed molecular area.28 This effectively rules out a repulsive contribution from the two ester carbonyl dipoles to the large second virial coefficients observed for the phospholipids. Dill and Stigter26,27 consider that the in-plane component of the zwitterion dipole (with its images in the oil layer) in dilute monolayers makes a minor contribution to the lateral pair potentials, based on treating the zwitterion’s two ionic charges as inside a disk and separated by 0.45 nm, which is close to the covalent bond distance between the two ionized groups. In dense monolayers with close-packed zwitterions, these orientations are likely to be altered by intermolecular polarization and steric factors. An early calculation of the electrostatic energies (without images) for a number of stereochemically feasible twodimensional planar lattices of dumbbell-shaped PC zwitterions in close-packed phospholipid monolayers showed that the energy of these lattices could be lower or higher than the electrostatic self-energy of the isolated zwitterion dipole, depending on the lattice pattern and the internal polarization of the zwitterion charges.29 The internal polarization of the charge distribution in the isolated zwitterion and its change on association and formation of a close-packed array will be important factors in defining the repulsion energies observed. These mutual polarizations will also bear on the role of the dispersion forces between zwitterions, as indicated by the growing evidence of the importance of dispersion forces between ions in electrolyte solutions and in colloid phenomena such as the Hofmeister series.30 Conclusions For spread monolayers of 1,2 dialkanoyl phosphatidyl choline and ethanolamine phospholipids at O/W interfaces, the apparent substantial differences in interpretation of the phase transitions observed in published isotherms over a range of temperatures are largely resolved when reanalyzed by rigorous thermodynamics. Interpretation (25) Stigter, D.; Mingins, J.; Dill, K. A. Biophys. J. 1992, 61, 1616. (26) Dill, K. A.; Stigter, D. Biochemistry 1988, 27, 3446. (27) Stigter, D.; Dill, K. A. Langmuir 1988, 4, 200. (28) Pallas, N. R.; Pethica, B. A. Unpublished results. (29) Pethica, B. A. Surface Activity and the Microbial Cell; Society of Chemical Industry Monograph 19; Society of Chemical Industry: 1964; p 85. (30) Ninham, B. W.; Yaminsky, V. Langmuir 1997, 13, 2097.

7498

Langmuir, Vol. 20, No. 18, 2004

Figure 2. ∆F for the phase transitions of equimolar mixed monolayers of SOS and OTAB at the interfaces of CCl4 and n-heptane with dilute aqueous NaCl solutions.

of these phase transitions is assisted by data on the phase transitions observed in O/W monolayers of equimolar mixtures of long-chain anionic and cationic compounds. The combined body of thermodynamic data provides the basis for molecular modeling of these phase transitions and advances understanding of the contribution of phospholipids to biomembrane structures. No phase transitions have yet been observed in O/W monolayers with a net charge, and simple salt effects on the phospholipid zwitterion isotherms are not discernible up to at least 0.1M NaCl. Changing the oil gives minor

Mingins and Pethica

isotherm shifts at high areas with oils studied to date, except with cyclic alkanes. In dilute PC and PE monolayers, net repulsion between the lipid molecules is shown in the large positive second and third virial coefficients, which are shown to be independent of chain length for the PCs. The virial coefficients indicate that both PC and PE headgroups are repulsive, almost equally so at low T, with the repulsion between PC headgroups increasing markedly with T as a result of changing orientation of the zwitterion dipole.25 At the phase transition, further changes in headgroup orientations are indicated by surface potentials and the substantial differences in ∆F between PEs and PCs (Table 1). These differences in headgroup repulsion are confirmed by the observed demixing of mixed PE and PC monolayers of the same chain length at the phase transitions. The ∆S for these phase transitions is largely accounted for by the chain packing, as is confirmed by the demixing above and below the transition regions of some mixed monolayers of homologous phospholipids, and by the large isotherm differences between the 1,2 and 1,3 glyceroisomers of the same chain lengths. New monolayer experiments with PEs are desirable to resolve remaining experimental questions, to establish the chain-length dependence of ∆S and ∆U for the phase transitions, and to confirm the PE headgroup contribution to the interaction potentials in the dense phase.. Studies on mixed O/W monolayers with additional components of biological significance for thermodynamic characterization will expand the understanding of the complexities of biological membranes. LA040016T