A Calorimetric Study of Phospholipid Hydration. Simultaneous

Simultaneous Monitoring of Enthalpy and Free Energy ... We present a novel method for monitoring isothermal lipid hydration using a sorption microcalo...
0 downloads 0 Views 84KB Size
Download PDF
J. Phys. Chem. B 2000, 104, 8053-8060

8053

A Calorimetric Study of Phospholipid Hydration. Simultaneous Monitoring of Enthalpy and Free Energy Natalia Markova,† Emma Sparr,*,† Lars Wadso1 ,‡ and Håkan Wennerstro1 m† Physical Chemistry 1, Center for Chemistry and Chemical Engineering, P.O. Box 124, and DiVision of Building Materials, Box 118, UniVersity of Lund, S-221 00 Lund, Sweden ReceiVed: March 17, 2000; In Final Form: June 6, 2000

We present a novel method for monitoring isothermal lipid hydration using a sorption microcalorimeter. A measuring cell of the double-twin calorimeter consists of two vessels connected by a stainless steel tube. The upper vessel contains pure water, and the bottom vessel is loaded with the lipid sample. This calorimeter allows for simultaneous measurement of the partial molar enthalpy and the chemical potential (or the partial molar free energy) of the water. The versatility of the method is demonstrated by studies of the hydration of the phospholipids dipalmitoyl phosphatyl choline (DPPC), dimyristoyl phosphatidyl choline (DMPC), and dilauroyl phosphatidyle choline (DLPC) at 25 and 27 °C. The measurements provide a relation between water content and water chemical potential, which, in these lamellar systems, is often recast as a forcedistance relation and has been called the hydration force. Through the simultaneously monitored calorimetric values, the partial molar enthalpy of water is also obtained. The method consequently provides a rather unique combination of information on both partial molar enthalpy and partial molar free energy and thus also the partial molar entropy of the process. We find that the incorporation of the first three to four water molecules per lipid is exothermic. These water molecules presumably interact directly with oxygen atoms on the phosphate of the lipid headgroup. When the first waters have been added, the remaining ones are incorporated endothermically. This applies to the water molecules taken up both in the gel phase and in the liquid crystalline state. We also observe that the sorption process triggers a first-order phase change from a gel (Lβ′) to a liquid crystalline (LR) phase. For DLPC, this occurs at 25 °C at a relative humidity of 0.79 with an endothermic transition enthalpy of 42 ( 2 kJ/mol (DLPC) and, for DMPC, at 27 °C at 0.93 relative humidity with ∆H ) 56 ( 5 kJ/mol (DMPC). We use a previously established model to quantitatively interpret these phase transitions. Furthermore, the observed endothermic nature of the sorption process above three to four waters per lipid is fully consistent with the suggestion that the negative free energy of the sorption (swelling) is due to increased thermal excitations and thus a positive entropy. It is more problematic to reconcile the data with models proposing structuring effects in the water as the main cause of the swelling.

Introduction The sorption of water vapor by a condensed phase can strongly influence its physical properties, such as phase state, crystal structure, reactivity, and rate of dissolution. There are numerous practical applications in which the sorption process or its inverse is of major concern. Correct dosing of pharmaceuticals requires knowledge of the actual composition. Sorption can lead to sintering of crystallites and cake formation or to major restructuring.1-3 The quality of food is often strongly influenced by the state of hydration, and drying is a major unit operation in the food industry.4 Water is present nearly everywhere, and sorption/drying is important in many other contexts, such as dehydration of organisms or epidermal water loss.5 Water is the most common solvent. Its unique solvation properties are the result of the high polarity, small molecular volume, and structural versatility in the liquid state. A molecular understanding of the hydration of a particular compound must be based on an understanding of the direct interaction of water with this compound. Furthermore, in a condensed phase, the * Author to whom correspondence should be addressed at Department of Physical Chemistry 1, PO Box 124, SE-22100 Lund, Sweden. Fax: +46 46 222 44 13. E-mail: [email protected].

net interaction is always an effective result of the interactions between several molecular components. Sorption calorimetry provides a method for direct measurement of the partial molar enthalpy change when a water molecule is transferred from the liquid to a sorbent condensed phase. The enthalpy change can then be interpreted as the difference between the water-sorbent interaction on one hand and the water-water interaction on the other hand. Another way to study the sorption process is to measure the relation between the amount of water taken up by a sorbent and the water chemical potential. This can, in general, be done using a gravimetric technique such as the desiccator method6 and the sorption microbalance;7 also, the so-called osmotic stress technique8,9 is based on this concept. In these cases, one gets a measure of the free energy changes as the water is incorporated into the sample. Recently, two of us have developed an isothermal sorption microcalorimeter based on a principle with two twin cells.10 The twin design has the effect of reducing the noise from external sources and decreasing the time lag between the thermal power produced in the measuring cell and the registered calorimetric signal. The measuring and reference cells consist of two vessels: the lower vessel of the measuring cell contains the sample under study, while the upper vessel contains a source

10.1021/jp001020q CCC: $19.00 © 2000 American Chemical Society Published on Web 07/29/2000

8054 J. Phys. Chem. B, Vol. 104, No. 33, 2000

Markova et al.

of pure water. The two vessels are connected, and the sorption process is followed calorimetrically in both. The heat effect due to the evaporation in the vapor source vessel provides the rate of vaporization. The heat effect in the sample vessel gives the enthalpy of sorption, and at steady state, the sorption rate is the same as the rate of vaporization. By integrating the sorption rate, one also obtains the amount of sorbed water. Thus, one obtains the enthalpy of sorption per mole of water. There is another piece of information in the rate of vaporization. Under the isothermal conditions, convection is negligible, and the transport between the two vessels is due to diffusion. Mass transport is then caused by a given gradient in vapor pressure. In the vapor source vessel, the vapor pressure corresponds to 100% relative humidity, and the transport rate is determined by the relative humidity in the sample vessel. Thus, once calibrated, the measured rate of vaporization also determines the relative humidity over the sample. In this way, the double calorimeter system is set up so that one simultaneously measures both the enthalpy and the free energy per mole of water involved in the sorption process. Biological processes occur in an aqueous environment, and the molecular processes of life depend on intermolecular and interaggregate interactions in such an environment.11 One approach to study the more fundamental aspects of these interactions is to use simple model systems. The swelling of polar lipids in a lamellar state has been one such key phenomenon that has given rise to many experimental and theoretical efforts.8,12,13 These studies are primarily concentrated on a description on the free energy level, whereas the enthalpic aspects of the swelling process have received much less attention. In this study, we use the sorption calorimeter for the simultaneous study of the partial molar enthalpy of water and the water chemical potential during the swelling of lamellar systems of the phospholipids dipalmitoyl phosphatidyl choline (DPPC), dimyristoyl phosphatidyl choline (DMPC), and dilauroyl phosphatidyle choline (DLPC). Such a simultaneous description of the enthalpy and the free energy has not previously been achieved for these systems.

The water flow, qm (g/s), from the vapor source vessel to the sample vessel is calculated as

Materials and Methods A novel double-twin isothermal microcalorimeter was used to study the water vapor sorption of the phospholipids. A detailed description of the instrument is presented elsewhere.10 The calorimetric cell consists of two vessels connected by a steel tube. At the start of the measurements the bottom vessel contains 40-100 mg of a dry sample, and the top vessel about 100 µL of water spread over the hydrophobic porous membrane [Durapore (0.22 µm), Millipore, Bedford, MA]. During the measurements, water vaporizes in the top vessel and diffuses through the tube down to the bottom vessel where it is taken up by the sample. Each measurement was repeated 3 times with almost identical result. The experimental setup could be looked upon as a continuous titration of an initially dry lipid with water vapor. The rate of water diffusion in the vapor is controlled by the geometry of the vessel and the boundary conditions. The maximum diffusional flow is about 0.4 µg of H2O per s. We have confirmed that the sorption process takes place under quasi-equilibrium conditions by conducting separate experiments with samples of different sizes. With the method, one can follow an isothermal line in the phase diagram of a substance-water system, scanning the water content from very low water chemical potentials up to 0.99 relative humidity. The thermal powers of vaporization, Pv (W), and sorption, Ps (W), are measured separately in the double microcalorimeter.

h represents the enthalpy change as a In the experiments, ∆sorpH small amount of water is taken from the liquid water at the atmospheric pressure and added to the sample. If equilibrium is maintained, this enthalpy change represents the difference between the partial molar enthalpy of the water in the sample and in the liquid water.

qm ) Pv/∆vHMw

(1)

where ∆vH (J/mol) is the enthalpy of vaporization of water. Fick’s law is sufficient to describe the diffusion from the top to the bottom vessel

qm ) Dk(psat - pw)

(2)

where Dk is the diffusional permeability for the water vapor through the gas phase between the top and bottom vessels, D is the diffusion coefficient of water vapor in air, k denotes the coefficient describing geometry of the sorption cell, psat is the saturation vapor pressure of water realized in the vapor source vessel, and pw is the vapor pressure of water in the sample vessel. An experiment can be design so that the water vapor pressure above the sample is kept negligibly small (with a drying agent such as molecular sieves). Then, the vapor flow would be

qm ) qm,max ) Dkpsat

(3)

Hence, combining eqs 1-3, the water chemical potential in the sample cell expressed in terms of the relative humidity, RH)pw/ psat, in any other experiment can be calculated as

RH ) 1 - qm/qm,max

(4)

The water gain, cw (g/g), by a sample in the bottom vessel is

cw )

∫(qm dt)/mo

(5)

where mo is a mass of a sample at the start of a measurement. The differential enthalpy of sorption, ∆sorpH h [J/mol(H2O)], relative to the liquid state of water at steady state is

h ) (Ps - Pv)/qmMw ∆sorpH

∆sorpH h )H hw - H h ow ) ∆H hw

(6)

(7)

A sorption microbalance DVS-1000 (Surface Measurement Systems Ltd, London, U.K.) was used to verify sorption isotherms of DMPC obtained by sorption microcalorimetry at 25 and 27 °C. In these experiments, 13-16 mg of the lipids were subjected to a stream of nitrogen gas with a varying water vapor content. The relative humidity of the gas stream was increased stepwise (with a maximum equilibration time of 12 h for a 3% increase in the relative humidity near saturation). Simultaneously changes in the mass of the sample were monitored. A MicroCal-2 (MicroCal, Northampton, MA) high-sensitivity differential scanning calorimeter (DSC) was used to verify the purity of the DMPC sample. A 30-µM dispersion was prepared by adding water to the dry powder of the phospholipid, heating for 4 min at 40 °C, and then vortexing for 4 min of vortexing. A scan rate of 30 °C/h was used. The phase transition temperatures as measured by high-sensitivity DSC in a dilute

Calorimetric Study of Phospholipid Hydration

Figure 1. Microcalorimetric sorption data for (a) DLPC at 25 °C, (b) DMPC at 25 °C, (c) DMPC at 27 °C, and (d) DPPC at 25 °C: water content [mol(water)/mol(lipid)] versus realtive humidity (s). In panels b and c, sorption data on similarly prepared samples obtained using DVS microbalance (O) are included. For comparison, panels a and d include osmotic stress data (×) and exponential curve fit (- - -) from Lis et al.25 Panel d also includes results of climatic chamber technique for DPPC at 22 °C ()) from Jendrasiak et al.17

(30-µM) aqueous dispersion were 15.8 and 24.1 °C for the preand main transition of DMPC, respectively. DLPC, 1,2-dilauroyl-sn-glycero-3-phosphatidylcholine, Mw ) 621 g/mol, >99% pure (Avanti Polar Lipids, Birmingham, AL) was obtained dissolved in chloroform. After evaporation of the chloroform in air, the samples were annealed by adding small amount of water and subsequently drying for 2 days in a vacuum at room temperature. DPPC, 1,2-dipalmitoyl-sn-glycero-3phosphatadylcholine (>98% pure), Mw ) 734 g/mol, and DMPC, 1,2-dimyristoyl-sn-glycero-3-phosphatadylcholine (>98% pure), Mw ) 678 g/mol, in powder form were obtained from Larodan Fine Chemicals (Malmo¨, Sweden) and were used directly after drying in a vacuum at room temperature or being annealed by prehydration and subsequent drying. According to Nilsson et al.14 and McIntosh et al.,9 such a careful drying procedure is sufficient to remove all water from the lipid sample. The fact that the lipids prior to drying were stored at -4 °C further ensures the dry state of the lipids.15 However, other authors claim that one or a maximum of two water molecules per lipid are so strongly associated to the lipid headgroups that they are very difficult to remove in practice.16 Below, we report compositions on the assumption that there is no water in the maximally dried sample. Results and Discussion Free Energy of Sorption. The calorimetric sorption measurement provides a relation between water content and relative humidity and, Figure 1a-d shows the results for (a) DLPC at 25 °C, (b) DMPC at 25 °C, (c) DMPC at 27 °C, and (d) DPPC at 25 °C. In the cases of DMPC at 25 and 27 °C, the results were checked by independent measurements on similarly prepared samples using a commercial sorption microbalance. Data from these measurements are shown as open circles in Figure 1b and c. For comparison, results for DPPC sorption at 22 °C obtained with the climatic chamber technique by

J. Phys. Chem. B, Vol. 104, No. 33, 2000 8055 Jendrasiak et al.17 are shown in Figure 1d. There is, within experimental error, quantitative agreement between these results. The sorption curves show four different regimes. At low relative humidities, there is a continuous uptake of water molecules. This is followed by a region with an almost constant water content over a large range of relative humidity, which changes over to a stepwise sorption at almost constant relative humidity. In the fourth regime, there is again a continuous increase in the water content as the relative humidity approaches saturation. The latter regime is only seen for DLPC at 25 °C and DMPC at 27 °C (Figure 1a and c), with onsets at relative humidities of 0.79 and 0.93, respectively. The thermodynamic properties of binary phospholipid-water systems have been studied extensively.18 There exists a large body of data on the relation between water content and water chemical potential, usually expressed in terms of force-distance curves.8 In addition, experimental and theoretical phase diagrams have been established in a few cases (DMPC, DPPC).19-23 On the basis of these investigations, it is straightforward to obtain a qualitative interpretation of the observed sorption isotherms. At 25 °C in excess water, one is above the main gel-to-liquid crystal phase transition for both DLPC and DMPC. From phase equilibrium studies of DMPC and DPPC, it is established that the transition can be reversed by decreasing the water content. The stepwise sorption in Figure 1a and c thus reflects a transition from gel (Lβ′) to liquid crystalline (LR) phase induced by an increase in the water chemical potential, which is analogous to the temperature-induced transition in excess water. It also follows that the increase in water content at relative humidities above the transition reflects the so-called hydration force24 in the lamellar liquid crystal. This force is usually reported in terms of a relation between osmotic pressure and the thickness of the water layer. Whereas the osmotic pressure is just one way of expressing the water chemical potential, the water content is a more solid thermodynamic variable than the water layer thickness. The interlamellar separation can be translated into water/lipid ratio by a conversion factor of ap/(2VH2O). Here, ap represents the average area per lipid headgroup, which varies in response to the water content.25 The sorption curves calculated from the osmotic stress experiments25 for the liquid crystalline (DLPC, 25 °C) and the gel (DPPC, 25 °C) phases are shown in Figure 1a and d. Despite the uncertainties in the conversion of the data and in the absolute values of the water content, the agreement between the sets of data fully demonstrates the basic equivalence between the methods. At low water contents, the sorption isotherm shows only a moderate uptake of water over a large span in relative humidity. In terms of the osmotic stress technique, the data reflect the hydration force in the gel phase. Such data are more scarce, but also in this regime, there is a reasonable agreement between our sorption data and previous force measurements. The hydration force is generally considered as exponentially decaying with the interlamellar distance for both the liquid crystalline and gel phases.25 Our sorption data for the gel phase show a clear deviation from the exponential behavior at low water contents (Figure 1d), as was also observed by McIntosh et al.9 In the liquid crystalline phase, the sorption data fit well to the exponentially decaying repulsive force (Figure 1a). At a given water content, the repulsive interbilayer interaction has been found to be considerably weaker in the gel phase than in the liquid crystal, and this is confirmed by the present data. We can go one step further and attempt to obtain a quantitative interpretation of the data. Guldbrand et al. presented a model describing the phase behavior of the DPPC-water

8056 J. Phys. Chem. B, Vol. 104, No. 33, 2000

Figure 2. Calculated phase diagram according to model by Guldbrand et al.22 for (a) DLPC and (b) DMPC. Experimental sorption data are used to describe interaction forces.

system by combining DSC data for the phase transition in excess water with data from osmotic stress measurements.22 The same model can be applied to the DMPC and DLPC systems. The transition enthalpies in excess water are known (vide infra), and for the variation of the water chemical potential upon hydration within one phase, we use the sorption data of Figure 1. For the liquid crystalline phase, the sorption data in Figure 1a were extrapolated into the metastable regions, assuming the exponentially decaying hydration repulsion according to Lis et al.25 Details of the calculations are given in Appendix 1. Figure 2 shows the calculated temperature-composition phase diagrams for the DLPC and DMPC systems. The model allows us to equally well represent the phase equilibrium in terms of the variables temperature and relative humidity as shown in Figure 3a and b. We find that, for DLPC, the calculated relative humidity at the gel-to-liquid crystal transition at 25 °C is 0.80 (observed 0.79), while the calculated water uptake is from 4.1 to 6.6 mol/mol (observed 2.9-6.7). For DMPC at 27 °C, the transition is calculated to occur at RH ) 0.94 (observed 0.93), with a water uptake from 6.7 to 10.0 mol/mol (observed 3.610.8). The calculated phase diagram also shows very good agreement with the experimental T-RH phase diagram for DMPC reported by Smith et al.26 The model thus reproduces the relative humidities extremely well, while the water uptake is slightly underestimated. The latter property is sensitive to the exact expressions for the water chemical potential in both phases, which are difficult to achieve experimentally at these low water contents. Further, the experimental boundaries for the water uptake may be overestimated because of small inhomogenities in the sample upon the phase transition. We can thus conclude this section by noting that the calorimetric results for the free energy of water sorption are consistent with previous studies carried out with quite different techniques, such as osmotic stress and climatic chamber (cf. Figure 1a and d). The calorimetric method appears particularly useful at low water contents. Furthermore, we have demonstrated that one can use a previously established model to nearly

Markova et al.

Figure 3. Calculated phase diagram for (a) DLPC and (b) DMPC as a function of temperature and relative humidity. Experimental sorption data are used to describe interaction forces.

TABLE 1: Integrals of the Exothermic Peaks of the Partial Molar Enthalpy of Water at Low Water Contents lipid

integral of peak kJ/mol(lipid)

DLPC, 25 °C DMPC, 25 °C DMPC, 27 °C DPPC, 25 °C

-16 ( 2 -20 ( 2 -19 ( 1 -10 ( 2

quantitatively interpret the data for the transition between gel and liquid crystalline phases. Enthalpy of Sorption. The novel feature of the present experimental method is that the partial molar enthalpy of water in the sorption process is obtained simultaneously with the free energy. The observed heat effects can be directly interpreted as the partial molar enthalpy of water measured relative to a bulk aqueous phase. The results for the four cases studied are shown in Figure 4a-d. Also here, we can recognize four different regimes. At low water contents, the partial molar enthalpies are large and negative in all four samples. The curves are reproducible, but as indicated also by the kink on the isotherm of Figure 1a for DLPC, the sorption is not occurring at complete equilibrium, and there are some kinetic delays. This results in artificially sharp peaks in the curves (Figure 4a-d), whereas their integrated values, see Table 1, should correctly reflect the changes in partial molar enthalpy in this concentrated regime. At somewhat higher water contents, there is a change to a regime with a positive partial molar enthalpy. In Figure 4b and d, the partial molar enthalpy thereafter decays toward zero when the limit of full hydration of the gel phase is approached. In the curves of Figure 4a and c, on the other hand, there is a change h . This regime of to a constant strongly positive value of ∆sorpH constant partial molar enthalpy corresponds to the transition from gel to liquid crystalline phase discussed above. At even higher water contents follows a region where the partial molar enthalpies are small but positive and decaying toward zero as the limit of full hydration of the liquid crystalline phase is approached.

Calorimetric Study of Phospholipid Hydration

J. Phys. Chem. B, Vol. 104, No. 33, 2000 8057

Figure 5. A thermodynamic scheme for DLPC demonstrating consistency between the enthalpy measured for the transition from gel to liquid crystalline phase in the sorption process and the phase transition enthalpy measured by DSC at full hydration.

Figure 4. Partial molar enthalpy of water in the sorption process as a function of water content for (a) DLPC, 25 °C; (b) DMPC, 25 °C; (c) DMPC, 27 °C; and (d) DPPC, 25 °C. The inset in panel a shows a magnification of the partial molar enthalpy of water, ∆H h w, and the water chemical potential, ∆µw, in the region of swelling of the LR phase.

The lipid samples were prepared by drying in a vacuum without special precautions taken to ensure the formation of equilibrium crystals. It was confirmed by small and wide-angle X-ray diffraction that the lamellar structure is retained even at low relative humidities, although the sample appears to lack long-range order in the lateral direction.19 This was also confirmed for our system. The first exothermic peak observed at low water contents in all three samples can be interpreted as due to the primary hydration of the phosphate part of the headgroup, as found by Pohle et al.27 This primary hydration also induces the formation of the proper gel phase.19 When the gel phase of DMPC and DPPC is further hydrated (Figure 4b and d), the hydration is accompanied by a positive value of the partial molar enthalpy in the range of 4-3 kJ/mol, which is gradually decreases toward zero. In this range, the partial molar free energy RT ln(RH) changes from -0.4 to -0.13 kJ/mol, showing that the free energy and enthalpy have opposite signs and that the uptake of water is favored by entropic effects. Ideally, at a first-order phase change, as occurs for the samples of Figure 4a and c, the partial molar enthalpy should change discontinuously, in contrast to the chemical potential. Under actual experimental conditions, there are always smearing effects that give a continuous character to the transition. At the onset of the transition, there are large changes in the partial molar enthalpy. However, a comparison with Figure 4b and d, where the transition is absent, shows that a major part of this change from negative to positive values also occurs in the absence of the transition. It is thus difficult to precisely locate the onset of the discontinuity. For the DLPC sample, the estimate gives a water content of 2.8 ( 0.2 at the start of the transition, while for the DMPC sample at 27 °C, it occurs at a water content of 3.4 ( 0.3 mol/mol. At the end of the transition, the partial molar enthalpy falls from one positive value to a smaller one. There is a somewhat larger uncertainty in the estimated water content: 6.9 ( 0.5 and 10.7 ( 0.4 mol/mol for DLPC and DMPC, respectively. This uncertainty may contribute to the discrepancy between the experimental and calculated values for the water content in the two-phase Lβ′-LR region. Because of the discontinuities for the ideal case, the estimates based on

the enthalpy data should be slightly more accurate than those based on the vapor pressure data, but the limits cited above are fully consistent with the data of Figure 1a and c. By integrating the enthalpy over the phase transition, we arrive at a total transition enthalpy of 42 ( 2 kJ/mol for DLPC at 25 °C and 56 ( 5 kJ/mol for DMPC at 27 °C. The change in partial molar enthalpy of water is strongly endothermic, consistent with the interpretation that the main molecular event is the “melting” of the alkyl chains. The transition from the gel to the liquid crystalline phase has been extensively studied by differential scanning calorimetry for the case of fully hydrated systems. For DLPC, the transition enthalpy is 18 kJ/mol at -1 °C,15 while for DMPC, the transition occurs in two steps with a pretransition at 14 ( 4 °C having ∆prH ) 4 ( 1 kJ/mol and a main transition at 23.9 ( 1 °C having ∆trH ) 27 ( 2kJ/mol.28 The transition enthalpies at full hydration are thus significantly smaller than our observed values for the isothermal sorption process. There are two sources of this difference. The heat capacities are larger in the liquid crystalline phases than in the corresponding gel phases, as is normal for any pair of solids and liquids. Thus, the enthalpy difference should be larger at the higher temperatures of the sorption process relative to the enthalpy difference at the transition temperature in excess water. Second, the partial molar enthalpy is large in the gel phase, as demonstrated in Figure 4b, and the enthalpy increase in going from a water content of 3-4 molecules per lipid to full hydration of the gel phase is much larger than that in going from 7 or 10 waters per lipid to full hydration of the liquid crystalline phase. In Figure 5, we present a scheme that shows that the enthalpies measured for the transition from gel to liquid crystalline phase in the sorption process are consistent with the previously known values measured at full hydration. We note that it is necessary to have a large positive value of ∆sorpH h in the gel phase at low water contents to obtain consistency, which further supports the results of Figure 4b and d. In the scheme in Figure 5, the heat capacity difference between the liquid crystalline and gel phases of DLPC was approximated as twice the heat capacity difference for the liquid and crystalline phase of dodecane [∆Cp(C12H26) ) 69.6 J K-1 mol-1].29 The enthalpy of hydration of the DLPC gel phase, 23 ( 5 kJ/mol, was estimated by taking an average of the integrated partial molar enthalpies of water for the gel phases of DPPC and DMPC, 25 °C, extrapolated to the limit of the full hydration. The same argument applied to the DMPC case demonstrates the consistency between the transition enthalpy obtained by sorption calorimetry at 27 °C (61.5 kJ/mol) and the value obtained in excess water by DSC [Lβ′ f Pβf LR, ∆H ) 31.2

8058 J. Phys. Chem. B, Vol. 104, No. 33, 2000

Markova et al.

kJ/mol(DMPC)]. The contribution from the heat capacity term to the discrepancy between the DSC and the sorption calorimetry data is rather small ∆Cp∆T ) 0.6 kJ/mol(DMPC) [∆Cp(C14H30) ) 89.0 J K-1 mol-1]. The enthalpy of the hydration of the gel phase approximated from the measurements on DMPC at 25C° is 28.5 kJ/mol (DMPC). At the completion of the transition, the partial molar enthalpy has small positive values of 1.0 ( 0.5 kJ/mol and 2.0 ( 0.5 kJ/mol for DLPC and DMPC, respectively. The positive value of the partial molar enthalpy has also been indirectly confirmed by solution calorimetry. Two DLPC samples equilibrated at 0.81 and 0.94 relative humidity were immersed into water at 25 °C. The difference between the measured heats of immersion gave the integral enthalpy of sorption for the interval of the two water-lipid compositions.30 The obtained value was endothermic and was of the same order of magnitude as the corresponding sorption calorimetric value. The chemical potential at these transitions corresponds to values of the partial molar free energy of -0.58 kJ/mol and -0.18 kJ/mol, respectively. Primarily, this demonstrates that the free energy and the partial molar enthalpy have opposite signs also in the liquid crystalline phase (Figure 4a, inset), thus implying a positive value of the partial molar entropy. At further sorption in the liquid crystalline phase, the partial molar enthalpy decays toward zero. From the osmotic stress measurements, it is established that the partial free energy continues to decay up to a sorption of approximately 25 water molecules, and we thus expect a qualitatively similar behavior of the partial molar enthalpy. However, in this regime, the magnitude of the enthalpy effect is too small to be measured with our present accuracy. An entropy-driven swelling in these systems could, in principle, also be revealed by studying the temperature dependence of the interaction force. It follows, in an analogy with the Clausius-Claperyon equation, that the relative change in RH in a temperature interval from T1 to T0 is given by

RH(T1) RH(T0)

[

]

∆H h w(T1 - T0) RT0T1

) exp -

(8)

In the gel phase, ∆H h w is measured to be approximately 3 kJ/ mol. As an example, a temperature change of 10 °C thus gives a relative change in RH of ca. 4%, which is within the experimental error limit. Therefore, a temperature dependence can possibly only be detected in the swelling limit where the system is very sensitive to small changes.31 The negative value of the free energy reflects the fact that these systems swell spontaneously when exposed to pure water, and this interaction has been termed the “hydration force”. There exists a debate over the molecular mechanism behind this interaction.8,12 One line of thought attributes the forces to structural effects in the solvent water, and this is essentially an enthalpic source of the interaction.32,33 The other main suggestion attributes the forces to thermal excitations of the lipidwater interface providing an entropic source of the repulsive interaction.9,34 Our experimental observation clearly is most compatible with the latter type of mechanism. Thus, the calorimetric measurements have provided an essentially new type of data for resolving a long-standing controversy. We can also estimate the total enthalpy for the swelling of a gel phase from three water molecules too full hydration by extrapolating the enthalpies of Figure 4b. As the lipid approaches the completely hydrated state, the partial molar enthalpy should decrease drastically, making the extrapolation described in Figure 5 possible. The value of 23 ( 5 kJ/mol (lipid) can be

converted to an estimate of the interfacial surface enthalpy of the gel lipid bilayer-water system. Using an area of 58 × 10-20 m2 per lipid, the estimated interfacial enthalpy is 34 mJ/m2, which is, in fact, a typical magnitude for an interfacial enthalpy. This type of argument provides yet another aspect of the importance of entropy effects in explaining the spontaneous swelling.35 Conclusions We have demonstrated the versatility of a novel method for measuring vapor sorption, which provides simultaneous values of the partial molar free energies and enthalpies of water. It was shown that, for the chemical potential, the method gives results that are consistent with those obtained using a commercial sorption microbalance and also with previous osmotic stress and climatic chamber measurements. The enthalpies for the sorption process have not previously been measured in these systems simultaneously with the partial molar free energies. The basic findings are that there is (i) an exothermic water uptake of the first three to four water molecules, (ii) an endothermic swelling in the gel phase containing more than four water per lipid, (iii) a sorption-induced gel-to-liquid crystal transition with high positive values of the enthalpy change, and (iv) an endothermic swelling in the more concentrated region of the liquid crystalline phase with a ∆H h w of an opposite sign from the ∆µw and up to a factor of 10 times larger in magnitude. Thus, the first few water molecules develop a strong interaction with atoms of the lipid headgroup. The most likely sorption sites are the oxygen atoms of the choline phosphate groups. The subsequent sorption is driven by entropy effects, and the water molecules at this stage of the sorption process are more likely to interact with the quaternary ammonium group and the less polar ether-like oxygen atoms. The thermodynamic observation that the partial molar free energy and enthalpy have opposite signs provides strong support to the interpretation that the swelling of the lipids is due to increased thermal excitations in the lipid headgroup as water is added.34,36 It is much more difficult to reconcile the experimental findings with the idea that the swelling is caused by structuring effects in the water itself, as in the Marc¸ elja theory.32 In more qualitative discussions of lipid hydration,8,37 the free energy is often discussed as if it were enthalpy-dominated, and the magnitude is related to the strength of a hydrogen bond. The measurements demonstrate that such arguments are misleading. We have obtained unique thermodynamic data for the water sorption of some lipid systems. There is no strict procedure for transforming this set of data into a molecular picture. It seems that, for a system as complex as a swollen lipid phase, the detailed molecular description has ultimately to be accomplished using computer simulations.13,38,39 It is our opinion that, in this area, there are still severe technical problems in properly describing the molecular interactions in the polar region. The data presented above could turn out to be very useful for testing molecular force fields, because the enthalpies are more readily extracted from simulations than are the chemical potentials. Acknowledgment. We thank Katarina Ekelund for help with the X-ray measurements. Dr. Burkhard Ma¨dler and Dr. KlausPeter Schneider, University of Leipzig, are gratefully acknowledged for providing us with the DLPC sample and for cooperation in the early stage of the project. Appendix: Calculation of the Phase Diagram The condition of equilibrium between two phases, I and II, requires that the chemical potentials, µ, of all components are

Calorimetric Study of Phospholipid Hydration

J. Phys. Chem. B, Vol. 104, No. 33, 2000 8059 TABLE A1: Parameters40 Used in the Calculations of the Phase Diagrams

equal in the two phases.

µI(H2O) ) µII(H2O) and µI(lipid) ) µII(lipid)

(A1)

To be able to calculate the phase equilibrium, one needs to know the chemical potentials of both components. At fixed temperature, T, the changes in chemical potentials of lipid (l) and water (w) in phase i are related by the Gibbs-Duhem relation

nw dµl(i,T) ) dµw(i,T) ) Xi dµw(i,T) nl

(A2)

where Xi represents the number of water molecules per lipid in phase i. Integration of eq A2 gives θ ) µlip,i[Xi(µw)] - µlip,i[Xi(µθw)] ) µlip,i[Xi(µw)] - µlip,i

-

∫µµ

w

w

X dµw ≡ -Ii(µw) (A3)

If we define the fully swollen lamellar system as the standard θ state, then µθw is the chemical potential for pure water, and µl,i is determined by the free energy per lipid at the composition where the lamellar phase, i, is in equilibrium with pure water. From eqs A1 and A3, it follows that the difference between the standard chemical potential of the lipid in phases I and II at equilibrium can be written as a function of µw.

µθlip,II(µw) - µθlip,I(µw) ) III(µw) - II(µw)

(A4)

The difference in the standard chemical potential of the lipid in phases I and II is also dependent on the temperature. Assuming the enthalpy, Hθ, and the entropy, Sθ, to be temperature-independent over a limited temperature range, the difference in standard chemical potential can be expressed as θ θ - T∆SIfII ) µθlip,II(T) - µθlip,I(T) ) ∆HIfII θ (A5) (1 - T/Tc)∆HIfII

By combining eqs A4 and A5, a relation between T and µw is obtained. To perform calculations of the phase diagram, we need a relation between µw and Xi. The chemical potential of water is simply related to the osmotic pressure for the solvent between the bilayers, which is the same as the interaction force per unit area.

∆µ(H2O) ) -VH2OΠosm ) - VH2O

F area

(A6)

The interaction force can be described either by the sorption data (Figure 1) extrapolated to the metastable region or by older osmotic stress data.25 The latter requires a conversion from h to the molar ratio of water per lipid, X. If the system is considered as incompressible, h and X are related by hiai ) 2VH2OXi, where ai is the average area per lipid headgroup and VH2O is the molar volume for water. For the considered concentration region, the force between the bilayers is dominated by a repulsive hydration interaction, Fr, which is counterbalanced by an attractive dispersion force, Fa, and the total force per unit area is given by

F ) F r + Fa ) Fr +

(

)

H 1 2 1 + 6π h3 (h + l)3 (h + 2l)3

(A7)

Finally, the water activity/relative humidity at the phase

Hamacker constant, H amplitude of hydration force (Pβ); F0 decay length of hydration force (Pβ); λ0 fully hydrated phase (LR); nH2O/nlipid fully hydrated phase (Lβ′); nH2O/nlipid fully hydrated phase (Pβ); nH2O/nlipid

6 × 10-21 J 1.5 × 109 N/m2 2.1 Å 23 mol/mol 15 mol/mol 18 mol/mol

TABLE A2: Parameters40 Used in the Calculations of the DLPC and DMPC Phase Diagrams DLPC ∆HLβfPβ ∆HPβfLR ∆HLβfLR Tc,LβfPβ Tc,PβfLR Tc,LβfLR l (Pβ); l

18 kJ/mol -1 °C 33 Å

DMPC 5 kJ/mol 26 kJ/mol 31 kJ/mol 14 °C 23 °C 21.5 °C 37 Å

transition is related to the chemical potential of water by

RH% ) 100e-µw/RT

(A*)

For the numerical calculations, sorption data were used to describe the interaction force in the LR (DLPC, 25 °C, Figure 1a) and the Lβ′ (DPPC, 25 °C, Figure 1d) phases. These data were chosen in the calculations of both the DLPC and DMPC phase diagrams because they give the most complete sets of data for each phase. No major differences in the interaction force are expected upon variation of the hydrocarbon chain lengths for either of the two phases, which was also confirmed by our results. No experimental results describing the interaction force in the Pβ phase are available from the literature. For this phase the repulsive force was assumed to decay exponentially as Fr ) F0e-h/λ0, and the interaction force is calculated from eq A7. However, we have no theoretical basis for this assumption. The phase boundaries are very sensitive to the exact expression for the interaction force and the calculated results for the Pβ phase can therefore only be seen as a qualitative description of the phase behavior. Parameters used for the numerical calculations are summarized in Tables A1 and A2. References and Notes (1) Giron, D. Thermochim. Acta 1995, 248, 1-59. (2) Khankari, R. K.; Grant, D. J. W. Thermochim. Acta 1995, 278, 61-79. (3) Ahlneck, C.; Zografi, G. Int. J. Pharm. 1990, 62, 87-95. (4) The Role of Water in Intermolecular Interactions in Food. In Water Management in the Design and Distribution of Quality Foods; Tolstoguzov, V., Ed.; Technomic Publishing Co., Inc.: Basel, Switzerland, 1999. (5) Alonso, A.; Meirelles, N. C.; Tabak, M. Biochim. Biophys. Acta 1995, 1237, 6-15. (6) Elworthy, P. H. J. Chem. Soc. 1961, 5385-5389. (7) Binder, H.; Kohlstrunk, B.; Heerklotz, H. H. J. Phys. Lett. 1999, 304, 329-335. (8) Rand, R. P.; Parsegian, V. A. Biochim. Biophys. Acta 1989, 988, 351-376. (9) McIntosh, T. J.; Magid, A. D.; Simon, S. A. Biochemistry 1987, 26, 7325-7332. (10) Wadso¨, L.; Markova, N. Thermochim. Acta 2000, in press. (11) Wennerstro¨m, H. Interfacial Interactions. In Physical Chemistry at Biological Interfaces; Baszkin, A., Norde, W., Eds.; Marcel Dekker Inc.: New York, 2000; pp 85-114. (12) Israelachvili, J.; Wennerstro¨m, H. Nature 1996, 379, 219-225. (13) Perera, L.; U., E.; Berkowitz, M. L. Langmuir 1996, 12, 26252629. (14) Nilsson, A.; Holmgren, A.; Lindblom, G. Biochemistry 1991, 30, 2126-2133. (15) Cevc, G.; Marsh, D. Phosphophilipid Bilayers: Physical Principles and Models; Wiley & Sons: New York, 1987; Chapter 8.

8060 J. Phys. Chem. B, Vol. 104, No. 33, 2000 (16) Small, D. M. Handbook of Lipid Research 4. The Physical Chemistry of Lipids; Plenum Press: New York, 1986; Chapter 12. (17) Jendrasiak, G. L.; Hasty, J. H. Biochim. Biophys. Acta 1974, 337, 79-91. (18) Phospholipid Hydration; McIntosh, T. J., Magid, A. D., Eds.; Marcel Dekker: New York, 1993. (19) Janiak, M. J.; Small, D. M.; Shipley, G. G. J. Biol. Chem. 1979, 254, 6068-6078. (20) Ulmius, J.; Wennerstro¨m, H.; Lindblom, B.; Arvidson, G. Biochemistry 1977, 16, 5742. (21) Grabrielle-Madelmont, C.; Perron, R. J. Colloid Interface Sci. 1983, 95, 471-482. (22) Guldbrand, L.; Jo¨nsson, B.; Wennerstro¨m, H. J. Colloid Interface Sci. 1982, 89, 532-541. (23) Goldstein, R. E.; Leibler, S. Phys. ReV. Lett. 1988, 61, 2213-2216. (24) LeNeveu, D. M.; Rand, R. P.; Parsegian, V. A. Nature 1976, 259, 601-603. (25) Lis, L. J.; McAlister, M.; Fuller, N.; Rand, R. P.; Persegian, V. A. Biophys. J. 1982, 37, 657-666. (26) Smith, G. S.; E. B., S.; Safinya, C. R.; Plano, R. J.; Clark, N. A. J. Chem. Phys. 1990, 92, 4519-4529. (27) Pohle, W.; Selle, C.; Fritzsche, H.; Binder, H. Biospectroscopy 1998, 4, 267-280.

Markova et al. (28) Caffrey, M.; Hogan, J. Chem. Phys. Lipids 1992, 61, 1-109. (29) Finke, H. L.; M. E., G.; Waddington, G.; Huffman, H. M. J. Am. Chem. Soc. 1954, 76, 333. (30) Harkins, W. D.; Jura, G. J. Pharm. Sci. 1944, 66, 919-927. (31) Gordeliy, V. I.; Cherezov, V. G.; Teixeira, J. J. Molecular Structure 1996, 383, 117-124. (32) Marcelja, S.; Radic, N. Chem. Phys. Lett. 1976, 42, 129-130. (33) Parsegian, V. A.; Rand, R. P. Langmuir 1991, 7, 1299-1301. (34) Israelachvili, J.; Wennerstro¨m, H. J. Phys. Chem. 1992, 96, 520531. (35) Wennerstro¨m, H. Colloids Surf., A 2000, 167, 209-214. (36) McIntosh, T. J.; Simon, S. A. Biochemistry 1993, 32, 8374-8384. (37) McIntosh, T. J.; Stewart, M. E.; Downing, D. T. Biochemistry 1996, 35, 3649-3653. (38) Berger, O.; Edholm, O.; Ja¨hnig, F. Biophys. J. 1997, 72, 20022013. (39) Tieleman, D. P.; Marrink, S. J.; Berendsen, H. J. C. Biochim. Biophys. Acta-ReV. Biomembr. 1997, 1331, 235-270. (40) Cevc, G. Phospholipids Handbook; Marcel Dekker: New York, 1993.