Chemistry and Physics of Lipids, 55 (1990) 85--96 Elsevier Scientific Publishers Ireland Ltd.
85
Monopolar-bipolar lipid interactions in model membrane systems Z. M i r g h a n i a*, D . B e r t o i a b, A . G l i o z z P , M . D e R o s a c a n d A . G a m b a c o r t a d aDipartimento di Fisica, Universitb di Genova, 16146 Genova, ~Istituto di Chimica Organica, Universit~ di Genova, 16132 Genova, clstituto di Biochimica delle Macromolecole, P Facoita di Medicina, Universit~ di Napoli, 80138 Napoli and dIstituto per la Chimica di Molecole di Interesse Biologico, 80072 Arco Felice (Na) (Italy) (Received September 22nd, 1989; revision revised December 6th, 1989; accepted January 19th, 1990)
mH-NMR, dynamic light scattering and negative staining electron microscopy have been used to study the formation and physico-chemical properties of aqueous dispersions of mixtures of monopolar lipids with bipolar lipids extracted from Sulfolobus solfataricus. This microorganism is a thermophilic archaeobacterium growing optimally at about 85 °C and pH 3. The two hydrolytic fractions of the membrane complex lipids that have been studied are: the symmetric lipid glycerol dialkyl glycerol tetraether (GDGT) and the asymmetric lipid glycerol dialkyl nonitol tetraether (GDNT). Electron micrographs of pure and mixed GDNT and GDGT dispersions show the formation of complex structures. Only above a critical monopolar/bipolar lipid ratio, typical of the bipolar lipid, could dosed structures be formed and good agreement was obtained in sizing with NMR, electron microscopy and dynamic light scattering. NMR spectra have been carried out at several temperatures from 25 ° to 85°C, to obtain information on the temperature-dependent structural, dynamic and permeability properties of the co-dispersed vesicles. The results are discussed in terms of the steric constraints and the chemico-physicai interactions occurring among the different parts of the molecules and compared with previous studies performed with different physical techniques. Keywords: bipolar lipids; monopolar-bipolar lipid interactions; sonicated vesicles; tH-NMR.
Introduction
The presence of unusual bipolar lipids in the plasma membrane of thermophilic archaebacteria raises questions regarding their organization. We shall refer in this work to the bipolar lipids of Sulfolobus solfataricus, a thermophilic archaeobacterium whose natural habitat is at 90°C with a very low pH (about 1) [1,2]. All the complex bipolar lipid molecules such as phospholipids, glycolipids and sulpholipids, are based on tetraethers, formed by two sn-2,3 glycerol moieties or one glycerol and one nonitol bridged through ether linkages by two C40 biphytanyl chains (see inset of Fig. 2). The first class of compounds is named glycerol Correspondence to: A. Gliozzi. *Permanent address: Faculty of Medicine, University of Khartoum, P.O. Box 102, Sudan.
diaikyl glycerol tetraether (GDGT), the second glycerol dialkyl nonitol tetraether (GDNT). An additional feature is that each biphytanyl chain contains between 0 and 4 pentane rings. The degree of cyclization in the C4o component increases when Sulfoiobus solfataricus is grown at increasing temperatures [4]. Besides the usual role of selective exchange of matter and energy, the cell envelope and the plasma membrane of this microorganism must perform a twofold task. Firstly, they must confer stability on the cell, which is subject to enormous environmental stresses, and secondly the membrane must constitute a barrier against the diffusion of hydrogen ions into the cell, since the cell must withstand a pH gradient of 5 --6 pH units [1,2]. Indirect evidence based on the absence of a preferential fracture plane in the middle of the lipid layer [5] suggests a monolayer organization of the plasma membrane;
0009-3084/90/$03.50 © 1990 Elsevier Scientific Publishers Ireland Ltd. Published and Printed in Ireland
86
moreover, black films from one component of the membrane lipids assumed such a structure [6]. On the other hand, recent X-ray diffraction studies [7,8] have shown the polymorphic nature of the lipid extract, indicating the possible physiological significance of the cubic phase. In our own group, we could not form liposomes or vesicles from the polar lipid extract using standard procedures. By contrast, the addition of small amounts of monopolar lipids favoured the formation of closed vesicles (Cavagnetto et al., in preparation). The aim of this work is to determine, using a lanthanide probe, whether closed vesicular structures are formed with GDGT and GDNT and their mixtures with PC. This information permits us to analyze the thermodynamic factors and the steric constraints hindering or favouring the formation of closed structures. In fact it has been found that only upon addition of PC above a critical concentration could closed structures be formed with GDGT and GDNT. This analysis can be extended to several classes of complex bipolar lipids [2] in order to find a possible combination of these compounds which forms closed structures. Although the final goal of these studies is to prepare vesicles formed entirely of bipolar lipids, the mixed systems presented in this work display very interesting properties which imply far-reaching potential applications also in the biotechnological field. Negative staining electron microscopy [9] has been used to study the morphology of aqueous dispersions of several neutral glycosphingolipids; we have extended this technique to analyse the morphological features of bipolar lipids and their mixtures with PC. Whenever closed structures were formed, light scattering was used to measure their mean size and distribution. In addition, nuclear magnetic resonance and lanthanide shift reagents described by Bystrov et al. [10] could provide information on the permeability properties of such vesicles and the dynamic behaviour of the molecules. The proton NMR spectrum of sonicated choline-containing lipids exhibits resonances belonging to the headgroups and to the hydrocarbon regions of the lipid molecules. The addition of paramagnetic ions in the
external medium of the vesicles was shown to separate the interior and exterior headgroup resonances due to the downfield shift of the outer headgroup signal [11]. This shift is probably due to the spin-lattice electrostatic interactions of the lanthanide ions with the phosphate groups of the lipid [12,13]. A downfield shift of the inner choline headgroup resonance may serve to monitor directly the ionic permeabilities of these vesicles [11]. Furthermore, it is shown that the ratio of the inner to the outer peak area, in conjunction with light scattering experiments, provides information on the composition of the inner and outer layer. It is concluded that both compounds, GDGT and GDNT, mixed with monopolar lipids, are asymmetrically located in the two halves of the membrane. Finally, monitoring the temperature-dependent changes of the line broadening and of the peak amplitude of the signals in the ~H-NMR spectrum of the vesicles helps to examine the thermotropic behaviour of these systems. The results were interpreted in terms of the molecular geometry and chemico-physical properties imposed by the bipolar nature of the molecules and the conclusions reached are discussed with respect to previous studies performed on the same lipids using various techniques [7,8,14-20]. Materials and Methods
Materials GDGT and GDNT bipolar lipids were extracted from Sulfolobus solfataricus as described previously [3]. Typically, these lipids contain an average of 2.3 cyclopentane rings per chain, and can be assigned an average molecular weight of 1290 and 1470 for GDGT and GDNT, respectively. Egg phosphatidylcholine (egg PC) and dipalmitoyl phosphatidylcholine (DPPC) were purchased from Lipid Products (Redhill, U.K.) and diphytanoyl phosphatidylcholine was obtained from Avanti (Birmingham, U.S.A.). Praseodymium chloride (99.9o7o) was obtained from Lancaster Synthesis (Lancaster, U.K.) and deuterium oxide (2H20 , 99.807o) from Aldrich
87 Chemical Company, U.S.A. All other chemicals were analytical grade.
Preparation o f lipid vesicles Dry archaebacterial lipid mixtures of GDGT or GDNT with either egg PC, DPPC or diphytanoyl PC were prepared by adding the required amounts of chloroform solutions of the lipids in a glass sonicating tube. The solvent was then evaporated under a stream of nitrogen and the last traces by evacuating at 3 mmHg. Dry pure choline-containing lipids were prepared in the same way. Multilayer liposomes were prepared by adding 3 ml of 2H20 preheated to 60°C to the dry pure lipids or lipid mixtures at the same temperature to give a final concentration of 37.5 mg/ml of total lipids. Unless otherwise indicated, the proportion of archaebacterial to choline lipids was at a molar ratio of 3:4. Since archaebacterial lipids have a tendency to stick to the surfaces of the vessel the sonicating tube was filled with nitrogen, sealed and shaken vigorously for approximately 30 h in a thermostated water bath at 60°C to obtain homogeneous liposomal preparations. GDNT displays several polymorphic transitions over short ranges of temperature and degrees of hydration [7,16], hence these parameters were kept nearly constant. Vesicular membranes were obtained by sonicaring the liposomes at 60°C (except for pure egg PC where an ice-bath was used) for 20 rain, using a probe-type sonicator (Ultrasonic Ltd.), while passing a stream of nitrogen. Sonication was interrupted every 2 rain for 30 s to avoid overheating. The vesicular solutions were kept in the thermostated water bath at 60°C for 30 rain to anneal.
Electron microscopy The morphology of aqueous dispersions of the pure and mixed lipids was studied by negative staining electror/microscopy. Small amounts of the lipid dispersions were deposited on Forvar carbon-coated grids, stained with 1% phosphotungstic acid solution at pH = 7. A better spreading of the dispersions was obtained by adding 1% bacitracin solution [23]. The excess
solution was blotted out with a f'dter paper and the grids were allowed to air dry. The samples were observed under a Philips EM 400 transmission electron microscope, and the diameters of vesicles or the areas of the various structures were measured by a quantimet 970 Quips/MX (Cambridge Instruments) on randomly taken micrographs.
Light scattering The diameter of vesicles was monitored with dynamic light scattering of a laser beam at 514 nm by use of a Brookaven Instrument BI 2030 digital correlator. Measurements have been performed at various temperatures. The sample volume (3 ml) was placed in the chamber set at constant temperature and data were collected for about 15 min.
IH.NMR Vesicular solutions (0.5 ml) were pipetted into 5-mm diameter NMR tubes. The required quantity of a stock solution of praseodymium chloride in zI-I20 was then added to give an extravesicular Pr 3÷ concentration of 5 raM. IHNMR spectra were obtained in a Varian FTNMR spectrometer, operating at 80 MHz, fitted with a calibrated temperature controller. Typically between 40 and 60 pulses were collected prior to Fourier transformation. For all samples, ~H-NMR spectra were recorded between 25°C and 85°C at intervals of 5°C. Ten minutes were allowed after each 5°C increment to facilitate thermal equilibration of the lipid phase. Results
The results in Fig. 3 show negatively stained electron micrographs and sizing of sonicated dispersions of (a) egg PC, 03) GDNT/egg PC vesicles, (c) GDNT/egg PC liposomes and of sonicated and unsonicated GDGT/egg PC dispersions at two different magnifications (d,e,f). It can be observed that while the GDNT/egg PC sonicated dispersion gives rise to unilamellar vesicles, the same does not occur with GDGT/ egg PC. The unsonicated GDGT/egg PC dispersions show large aggregates which are often
88
Fig. 1. Negatively stained electron micrographs of aqueous dispersions of egg PC and of mixtures of GDNT or GDGT with egg PC at a molar ratio of 1:4. (a) Sonicated bilayer egg PC vesicles; (b) sonicated GDNT/egg PC vesicles; (c) GDNT/egg PC liposomes; (d) GDGT/egg PC dispersions; (e) and (f) GDGT/egg PC sonicated dispersions at two different magnifications. The bars represent 200 nm. The histogram of the size distribution of the diameters measured by light scattering is given in (a) and (b). The other histograms represent the distribution of the areas determined on an average of 200 samples.
89
connected by "filament-like" structures. Upon sonication these seem to break, resulting in smaller aggregates of polydispersed sizes. The insets of Fig. 1 a,b, show the diameter distribution of vesicles filtered through a millipore 0.22/~m size filter obtained b y light scattering measurements. The analysis has been done with a multiexponential least square method. Two peaks were frequently obtained; however, no physical meaning can be given to the minor peak of larger vesicles which may represent a few larger particles. In each type of vesicles the predominant peaks represent over 95% of the total population present. The mean diameter, d, of egg PC and GDNT/egg PC vesicles obtained is 40 _+ 2 nm and 76 _ 4 nm, respectively (in agreement with the electron microscopy data). Similar values are obtained substituting DPPC for egg PC. The insets in Fig. 1 c,d,e show the distribution of feature counts versus area performed on an average of 150 structures. IH-NMR • The results in Fig. 2 show the 80 MHz proton spectra at T = 25°C of DPPC, egg PC, diphytanoyl PC, GDNT and GDGT solutions in deutero-chloroform (C2HCI3) together with their assignments. The lateral methyls of the diterpenoid hydrocarbon chain of diphytanoyl phosphatidylcholine exhibit several overlapping doublets, which are at variance with the triplet signals shown by the terminal methyl groups of the hydrocarbon chain of DPPC and egg PC. The methyl branches of both symmetrical and asymmetrical archaebacterial lipids GDGT and GDNT, on the other hand, exhibit a single broad (13 Hz) unresolved peak with dear second order character due to spin-spin coupling with the cyclopentane ring protons. The methylene protons of both DPPC and egg PC exhibit sharp signals indicating narrow differences in chemical shift environments along the acyl chain, while those of diphytanoyl PC exhibit a broader signal which has double the line width (10 Hz) due to the presence of the laterally distributed methyl branches. The methylene protons of both GDGT and GDNT exhibit broad signals at 2.2 ppm because these protons experience two different
chemical environments, neighbouring to the lateral methyl groups and lying between the lateral methyls and the cyclopentane rings. The chemical shifts of these protons cover the range between 2.0 ppm and 2.5 ppm, causing the > CH resonance to appear as a lowfield shoulder of the main methylene peak. The choline resonances of egg PC, DPPC and diphytanoyl PC appear as single sharp peaks at 4.3 ppm. The glycerol moiety signals of both archaebacterial lipids are higher in intensity compared with those obtained from the choline lipids, because the former contain unsubstituted glycerol. Figure 3 shows the proton spectra of the 2H20 sonicated liposomai dispersions of pure egg PC, GDNT/egg PC, GDGT/egg PC, pure DPPC and GDNT/DPPC, respectively, containing 5 mM Pr 3÷ in the external medium. It is well known [11,12,22] that the paramagnetic ions shift the signal of the outer monolayer choline headgroups of egg PC downfield, thus revealing the signal of the inner monolayer choline headgroups. The ratio of the peak area of the outer to inner choline signal averaged upon 10 different experiments is 1.6 showing that the vesicles are 40 nm in diameter if the bilayer thickness is 46 nm [23]. This value is in agreement with that obtained from light scattering experiments performed at room temperature. The signal of the methylene protons of the hydrocarbon core appears in the normal shape with the terminal methyl peak on the highfield region. Comparing this egg PC vesicles spectrum with that obtained from the vesicles prepared from mixing GDNT with egg PC at a molar ratio of 1:4, it is clear that the inner and outer choline signals are not as well separated at equal external concentrations of 5 mM Pr 3÷. The downfield shift of the outer choline signal is 6 Hz smaller in the mixed vesicles. This is due to the shielding of the praseodymium ions by the hydroxyl groups of nonitol present in the outer surface of the vesicles. In fact, increasing the concentration of the bipolar lipid to the molar ratio 1:2, the separation between inner and outer peaks decreases by a further 6 Hz. This behaviour suggests that phase separation had not occurred, a conclusion supported also by other experiments,
90
CH20H H ' ~ ~ O ~ .
H
1 ppm
I
!
CHOH CH~S
CI
R=H
TMS
GDGT
GLyc.roL GD6T~ _
BULK
I
CH~ CH~S TMS I
cy li n
GON.~
I
-N~CH3)3
i Gtyc,rot II oPh.P_.Lc ~L.__~,v,J
/I
BULK
It II
BULK
T,:
.JILl
CHiS
I
CH / v I I ',..._k TMS
-N(CH313
TMS
i
EggPC
Chotin~ CH~ Chotine Gtycerot
oH! CH
-N(CH3)3
CH2
DPPC Fig. 2. Proton NMR (80 MHz) spectra of C2HC13 solutions of egg PC, DPPC, diphytanoyl PC, GDNT and GDGT at T = 25°C. The concentration of the lipid is 37.5 mg/ml. The inset shows the chemical structure of GDGT and GDNT. The number of cyclopentane rings can vary from 0 to 4 per chain [4].
91
which will be discussed later. The ratio of the outer to the inner choline resonance is 1.3. In the Discussion section it will be shown that this measurement allows a very interesting conclusion. The hydrocarbon signal of egg PC in the mixed vesicles, shown in Fig. 3, collapses to approximately half its intensity and broadens due to the interactions with GDNT. The hydrocarbon of GDNT appears at the chemical shift of the terminal methyl signal of egg PC due to the different chemical composition in the hydrocarbon region of GDNT compared with that of egg PC. It can be deduced from the GDGT/egg PC mixed vesicle spectrum of Fig. 3 that at a molar ratio of 1:4, closed structures were not formed
HDO
I
~ydrocarbong '
v
-
EggPC
IA ¸
Fig. 3. Proton NMR (80 MHz) spectra of D20 sonicated dispersions of egg PC, GDNT/egg PC, GDGT/egg PC, DPPC and GDNT/DPPC containing 5 mM Pr s÷ in the external medium. Concentration of total lipid is 37.5 mg/ml and T = 60°C.
since the added praseodymium had immediately been completely internalized, hence a single broad resonance which belongs to both choline heads was observed. This is in agreement with electron microscopy results shown in Fig. 1 d,e,f. In spite of the same chemical composition of GDGT and GDNT in their hydrocarbon regions, their mixtures with egg PC give different packings, due to the presence of nonitol in the polar region of GDNT. Further, contrary to the results obtained with DPPC/GDNT vesicles where closed structures were obtained, we found it impossible even to form liposomes from GDNT and diphytanoyl PC mixed in the usual molar ratio of 1:4. This fact suggests that the lateral methyl groups of both lipids do not intercalate, but orient themselves opposite each other thus hindering hydrophilic interactions of the heads to take place on the outer side of the vesicle. If we now compare the spectrum of GDNT/ DPPC vesicles with that of pure DPPC vesicles at 60°C (Fig. 3) two major differences are observed. The ratio of the outer to the inner choline resonances is 1.3 instead of 1.6 and the hydrocarbon peak of DPPC collapses to one third of its original intensity in the mixed lipid vesicles, due to the decrease in motional freedom brought by the thermophilic lipid. However, both these vesicles have stable barriers for praseodymium ions, in spite of the shift of the outer monolayer choline resonance being up to I0 Hz smaller than for pure DPPC. It is interesting to examine the thermotropic properties of the vesicles prepared from mixtures of egg PC with the archaebacteriai lipids (Fig. 4). Scanning the ~H-NMR spectrum of pure egg PC vesicles at several temperatures between 25°C and 85°(2, the width at half height of the main hydrocarbon peak is reduced progressively, showing that both the motionai freedom of the lipid and the tumbling rate of the vesicles have increased. GDNT/egg PC vesicles spectrum at 27 °C show broad hydrocarbon spectrum, which changes with increase in temperature. At 72°C the spectrum shows equal intensities for both egg PC and GDNT hydrocarbon signals, and at 85°C egg PC signal intensity is higher than that of
92
CH2
A]
B)
lppm I~ppm w B
7z~/.A.////~\~..
1
PCGDNT
PC lppm C]/] }~DGT / B5oc
i
[,z_~,j
Fig. 4. Proton NMR (g0 MHz) spectra of the hydrocarbon region of the vesicles prepared from (A) egg PC, (B) GDNT/egg and (C) GDGT/egg PC at selected temperatures between T = 25°C and T = 85°C. Total lipid concentration is 37.5 mg/ml.
GDNT. GDGT/egg PC vesicles spectrum shows in comparison much less, but similar, changes with temperature (Fig. 4C). Figure 5A shows the NMR spectrum of the hydrocarbon region of pure DPPC vesicles below and at several degrees above its phase transition. Below the phase transition the hydrocarbon signal of the vesicles is broadened to the base line and, as expected, the signal sharpens with increase in temperature. The vesicles formed from GDNT/DPPC (Fig. 5B) show, in comparison, higher chain mobility at the temperature below the Tc of DPPC. GDNT, on the other hand, shows higher hydrocarbon signal intensity up to 70°C, while DPPC shows a sharper resonance at 85 °C. The graph in Fig. 6 shows a plot of the relative amplitudes of the IH-NMR signals of the hydrocarbon chains of the pure and the mixed
vesicles versus temperature; the inner choline signal was taken as a reference for the signal intensities. DPPC chains in the unmixed vesicles below and above the phase transition show a non-linear saturating behaviour (Fig. 6A) compared to the linear graph obtained when the vesicles were made from mixtures of DPPC with GDNT (Fig. 6C). This linearity was also observed for the hydrocarbon chains of GDNT in the mixed vesicles (Fig. 6D). The same linear behaviour was displayed by pure egg PC vesicles, which do not have any phase transition in the explored range of temperatures. A completely analogous result was obtained by plotting the line width Acol/2 at which the signal drops to half its maximal value. It is evident that no phase separation of the lipids had occurred in the mixed vesicles, as also indicated by the downfield shift of the outer choline signal and
93 CH
O!
AI
lppm I
OPPC CH3
40"C
Fig. 5. Proton NMR (80 MHz) spectra of the hydrocarbon regions of sonicated dispersions of: (A) DPPC and (B) DPPC/GNDT at selected temperatures between T = 40°C and T = 85 °C. Lipid concentration is 37.5 mg/ml. 1.8
by electron microscopy studies. This conclusion is in agreement with microcalorimetric studies of GDNT/DPPC mixtures [27]. In Fig. 7 the downfield shift of the inner choline resonance was measured immediately after addition of 5 mM Pr 3+ and 10 h later at 85°C to check for Pr 3. ion diffusion to the interior of the vesicles. It can be deduced that GDNT/egg PC vesicles have a barrier against Pr 3. ions as that o f pure egg PC. The permeability properties were not affected by varying the GDNT/egg PC ratio from 10 to 50 moW0. In fact, a downfield shift of 2--4 Hz of the inner choline signal was measured for both types o f vesicles, 10 h after the addition of 5ram Pr 3. in the external medium. G D N T / D P P C exhibited similar behaviour. By contrast, in G D G T / e g g PC sonicated dispersions the choline resonance was not shifted, in agreement with the fact that no closed structures were formed with a molar ratio of 1:4. Closed vesicles, 50 nm in diameter, were obtained only with a molar ratio of 1:6, a condition in which the inner and outer choline peaks were separated after addition o f Pr 3* ions. The
3.8
A
3.6
1A o
o
85
Monopolar-bipolar lipid interactions in model membrane systems Z. M i r g h a n i a*, D . B e r t o i a b, A . G l i o z z P , M . D e R o s a c a n d A . G a m b a c o r t a d aDipartimento di Fisica, Universitb di Genova, 16146 Genova, ~Istituto di Chimica Organica, Universit~ di Genova, 16132 Genova, clstituto di Biochimica delle Macromolecole, P Facoita di Medicina, Universit~ di Napoli, 80138 Napoli and dIstituto per la Chimica di Molecole di Interesse Biologico, 80072 Arco Felice (Na) (Italy) (Received September 22nd, 1989; revision revised December 6th, 1989; accepted January 19th, 1990)
mH-NMR, dynamic light scattering and negative staining electron microscopy have been used to study the formation and physico-chemical properties of aqueous dispersions of mixtures of monopolar lipids with bipolar lipids extracted from Sulfolobus solfataricus. This microorganism is a thermophilic archaeobacterium growing optimally at about 85 °C and pH 3. The two hydrolytic fractions of the membrane complex lipids that have been studied are: the symmetric lipid glycerol dialkyl glycerol tetraether (GDGT) and the asymmetric lipid glycerol dialkyl nonitol tetraether (GDNT). Electron micrographs of pure and mixed GDNT and GDGT dispersions show the formation of complex structures. Only above a critical monopolar/bipolar lipid ratio, typical of the bipolar lipid, could dosed structures be formed and good agreement was obtained in sizing with NMR, electron microscopy and dynamic light scattering. NMR spectra have been carried out at several temperatures from 25 ° to 85°C, to obtain information on the temperature-dependent structural, dynamic and permeability properties of the co-dispersed vesicles. The results are discussed in terms of the steric constraints and the chemico-physicai interactions occurring among the different parts of the molecules and compared with previous studies performed with different physical techniques. Keywords: bipolar lipids; monopolar-bipolar lipid interactions; sonicated vesicles; tH-NMR.
Introduction
The presence of unusual bipolar lipids in the plasma membrane of thermophilic archaebacteria raises questions regarding their organization. We shall refer in this work to the bipolar lipids of Sulfolobus solfataricus, a thermophilic archaeobacterium whose natural habitat is at 90°C with a very low pH (about 1) [1,2]. All the complex bipolar lipid molecules such as phospholipids, glycolipids and sulpholipids, are based on tetraethers, formed by two sn-2,3 glycerol moieties or one glycerol and one nonitol bridged through ether linkages by two C40 biphytanyl chains (see inset of Fig. 2). The first class of compounds is named glycerol Correspondence to: A. Gliozzi. *Permanent address: Faculty of Medicine, University of Khartoum, P.O. Box 102, Sudan.
diaikyl glycerol tetraether (GDGT), the second glycerol dialkyl nonitol tetraether (GDNT). An additional feature is that each biphytanyl chain contains between 0 and 4 pentane rings. The degree of cyclization in the C4o component increases when Sulfoiobus solfataricus is grown at increasing temperatures [4]. Besides the usual role of selective exchange of matter and energy, the cell envelope and the plasma membrane of this microorganism must perform a twofold task. Firstly, they must confer stability on the cell, which is subject to enormous environmental stresses, and secondly the membrane must constitute a barrier against the diffusion of hydrogen ions into the cell, since the cell must withstand a pH gradient of 5 --6 pH units [1,2]. Indirect evidence based on the absence of a preferential fracture plane in the middle of the lipid layer [5] suggests a monolayer organization of the plasma membrane;
0009-3084/90/$03.50 © 1990 Elsevier Scientific Publishers Ireland Ltd. Published and Printed in Ireland
86
moreover, black films from one component of the membrane lipids assumed such a structure [6]. On the other hand, recent X-ray diffraction studies [7,8] have shown the polymorphic nature of the lipid extract, indicating the possible physiological significance of the cubic phase. In our own group, we could not form liposomes or vesicles from the polar lipid extract using standard procedures. By contrast, the addition of small amounts of monopolar lipids favoured the formation of closed vesicles (Cavagnetto et al., in preparation). The aim of this work is to determine, using a lanthanide probe, whether closed vesicular structures are formed with GDGT and GDNT and their mixtures with PC. This information permits us to analyze the thermodynamic factors and the steric constraints hindering or favouring the formation of closed structures. In fact it has been found that only upon addition of PC above a critical concentration could closed structures be formed with GDGT and GDNT. This analysis can be extended to several classes of complex bipolar lipids [2] in order to find a possible combination of these compounds which forms closed structures. Although the final goal of these studies is to prepare vesicles formed entirely of bipolar lipids, the mixed systems presented in this work display very interesting properties which imply far-reaching potential applications also in the biotechnological field. Negative staining electron microscopy [9] has been used to study the morphology of aqueous dispersions of several neutral glycosphingolipids; we have extended this technique to analyse the morphological features of bipolar lipids and their mixtures with PC. Whenever closed structures were formed, light scattering was used to measure their mean size and distribution. In addition, nuclear magnetic resonance and lanthanide shift reagents described by Bystrov et al. [10] could provide information on the permeability properties of such vesicles and the dynamic behaviour of the molecules. The proton NMR spectrum of sonicated choline-containing lipids exhibits resonances belonging to the headgroups and to the hydrocarbon regions of the lipid molecules. The addition of paramagnetic ions in the
external medium of the vesicles was shown to separate the interior and exterior headgroup resonances due to the downfield shift of the outer headgroup signal [11]. This shift is probably due to the spin-lattice electrostatic interactions of the lanthanide ions with the phosphate groups of the lipid [12,13]. A downfield shift of the inner choline headgroup resonance may serve to monitor directly the ionic permeabilities of these vesicles [11]. Furthermore, it is shown that the ratio of the inner to the outer peak area, in conjunction with light scattering experiments, provides information on the composition of the inner and outer layer. It is concluded that both compounds, GDGT and GDNT, mixed with monopolar lipids, are asymmetrically located in the two halves of the membrane. Finally, monitoring the temperature-dependent changes of the line broadening and of the peak amplitude of the signals in the ~H-NMR spectrum of the vesicles helps to examine the thermotropic behaviour of these systems. The results were interpreted in terms of the molecular geometry and chemico-physical properties imposed by the bipolar nature of the molecules and the conclusions reached are discussed with respect to previous studies performed on the same lipids using various techniques [7,8,14-20]. Materials and Methods
Materials GDGT and GDNT bipolar lipids were extracted from Sulfolobus solfataricus as described previously [3]. Typically, these lipids contain an average of 2.3 cyclopentane rings per chain, and can be assigned an average molecular weight of 1290 and 1470 for GDGT and GDNT, respectively. Egg phosphatidylcholine (egg PC) and dipalmitoyl phosphatidylcholine (DPPC) were purchased from Lipid Products (Redhill, U.K.) and diphytanoyl phosphatidylcholine was obtained from Avanti (Birmingham, U.S.A.). Praseodymium chloride (99.9o7o) was obtained from Lancaster Synthesis (Lancaster, U.K.) and deuterium oxide (2H20 , 99.807o) from Aldrich
87 Chemical Company, U.S.A. All other chemicals were analytical grade.
Preparation o f lipid vesicles Dry archaebacterial lipid mixtures of GDGT or GDNT with either egg PC, DPPC or diphytanoyl PC were prepared by adding the required amounts of chloroform solutions of the lipids in a glass sonicating tube. The solvent was then evaporated under a stream of nitrogen and the last traces by evacuating at 3 mmHg. Dry pure choline-containing lipids were prepared in the same way. Multilayer liposomes were prepared by adding 3 ml of 2H20 preheated to 60°C to the dry pure lipids or lipid mixtures at the same temperature to give a final concentration of 37.5 mg/ml of total lipids. Unless otherwise indicated, the proportion of archaebacterial to choline lipids was at a molar ratio of 3:4. Since archaebacterial lipids have a tendency to stick to the surfaces of the vessel the sonicating tube was filled with nitrogen, sealed and shaken vigorously for approximately 30 h in a thermostated water bath at 60°C to obtain homogeneous liposomal preparations. GDNT displays several polymorphic transitions over short ranges of temperature and degrees of hydration [7,16], hence these parameters were kept nearly constant. Vesicular membranes were obtained by sonicaring the liposomes at 60°C (except for pure egg PC where an ice-bath was used) for 20 rain, using a probe-type sonicator (Ultrasonic Ltd.), while passing a stream of nitrogen. Sonication was interrupted every 2 rain for 30 s to avoid overheating. The vesicular solutions were kept in the thermostated water bath at 60°C for 30 rain to anneal.
Electron microscopy The morphology of aqueous dispersions of the pure and mixed lipids was studied by negative staining electror/microscopy. Small amounts of the lipid dispersions were deposited on Forvar carbon-coated grids, stained with 1% phosphotungstic acid solution at pH = 7. A better spreading of the dispersions was obtained by adding 1% bacitracin solution [23]. The excess
solution was blotted out with a f'dter paper and the grids were allowed to air dry. The samples were observed under a Philips EM 400 transmission electron microscope, and the diameters of vesicles or the areas of the various structures were measured by a quantimet 970 Quips/MX (Cambridge Instruments) on randomly taken micrographs.
Light scattering The diameter of vesicles was monitored with dynamic light scattering of a laser beam at 514 nm by use of a Brookaven Instrument BI 2030 digital correlator. Measurements have been performed at various temperatures. The sample volume (3 ml) was placed in the chamber set at constant temperature and data were collected for about 15 min.
IH.NMR Vesicular solutions (0.5 ml) were pipetted into 5-mm diameter NMR tubes. The required quantity of a stock solution of praseodymium chloride in zI-I20 was then added to give an extravesicular Pr 3÷ concentration of 5 raM. IHNMR spectra were obtained in a Varian FTNMR spectrometer, operating at 80 MHz, fitted with a calibrated temperature controller. Typically between 40 and 60 pulses were collected prior to Fourier transformation. For all samples, ~H-NMR spectra were recorded between 25°C and 85°C at intervals of 5°C. Ten minutes were allowed after each 5°C increment to facilitate thermal equilibration of the lipid phase. Results
The results in Fig. 3 show negatively stained electron micrographs and sizing of sonicated dispersions of (a) egg PC, 03) GDNT/egg PC vesicles, (c) GDNT/egg PC liposomes and of sonicated and unsonicated GDGT/egg PC dispersions at two different magnifications (d,e,f). It can be observed that while the GDNT/egg PC sonicated dispersion gives rise to unilamellar vesicles, the same does not occur with GDGT/ egg PC. The unsonicated GDGT/egg PC dispersions show large aggregates which are often
88
Fig. 1. Negatively stained electron micrographs of aqueous dispersions of egg PC and of mixtures of GDNT or GDGT with egg PC at a molar ratio of 1:4. (a) Sonicated bilayer egg PC vesicles; (b) sonicated GDNT/egg PC vesicles; (c) GDNT/egg PC liposomes; (d) GDGT/egg PC dispersions; (e) and (f) GDGT/egg PC sonicated dispersions at two different magnifications. The bars represent 200 nm. The histogram of the size distribution of the diameters measured by light scattering is given in (a) and (b). The other histograms represent the distribution of the areas determined on an average of 200 samples.
89
connected by "filament-like" structures. Upon sonication these seem to break, resulting in smaller aggregates of polydispersed sizes. The insets of Fig. 1 a,b, show the diameter distribution of vesicles filtered through a millipore 0.22/~m size filter obtained b y light scattering measurements. The analysis has been done with a multiexponential least square method. Two peaks were frequently obtained; however, no physical meaning can be given to the minor peak of larger vesicles which may represent a few larger particles. In each type of vesicles the predominant peaks represent over 95% of the total population present. The mean diameter, d, of egg PC and GDNT/egg PC vesicles obtained is 40 _+ 2 nm and 76 _ 4 nm, respectively (in agreement with the electron microscopy data). Similar values are obtained substituting DPPC for egg PC. The insets in Fig. 1 c,d,e show the distribution of feature counts versus area performed on an average of 150 structures. IH-NMR • The results in Fig. 2 show the 80 MHz proton spectra at T = 25°C of DPPC, egg PC, diphytanoyl PC, GDNT and GDGT solutions in deutero-chloroform (C2HCI3) together with their assignments. The lateral methyls of the diterpenoid hydrocarbon chain of diphytanoyl phosphatidylcholine exhibit several overlapping doublets, which are at variance with the triplet signals shown by the terminal methyl groups of the hydrocarbon chain of DPPC and egg PC. The methyl branches of both symmetrical and asymmetrical archaebacterial lipids GDGT and GDNT, on the other hand, exhibit a single broad (13 Hz) unresolved peak with dear second order character due to spin-spin coupling with the cyclopentane ring protons. The methylene protons of both DPPC and egg PC exhibit sharp signals indicating narrow differences in chemical shift environments along the acyl chain, while those of diphytanoyl PC exhibit a broader signal which has double the line width (10 Hz) due to the presence of the laterally distributed methyl branches. The methylene protons of both GDGT and GDNT exhibit broad signals at 2.2 ppm because these protons experience two different
chemical environments, neighbouring to the lateral methyl groups and lying between the lateral methyls and the cyclopentane rings. The chemical shifts of these protons cover the range between 2.0 ppm and 2.5 ppm, causing the > CH resonance to appear as a lowfield shoulder of the main methylene peak. The choline resonances of egg PC, DPPC and diphytanoyl PC appear as single sharp peaks at 4.3 ppm. The glycerol moiety signals of both archaebacterial lipids are higher in intensity compared with those obtained from the choline lipids, because the former contain unsubstituted glycerol. Figure 3 shows the proton spectra of the 2H20 sonicated liposomai dispersions of pure egg PC, GDNT/egg PC, GDGT/egg PC, pure DPPC and GDNT/DPPC, respectively, containing 5 mM Pr 3÷ in the external medium. It is well known [11,12,22] that the paramagnetic ions shift the signal of the outer monolayer choline headgroups of egg PC downfield, thus revealing the signal of the inner monolayer choline headgroups. The ratio of the peak area of the outer to inner choline signal averaged upon 10 different experiments is 1.6 showing that the vesicles are 40 nm in diameter if the bilayer thickness is 46 nm [23]. This value is in agreement with that obtained from light scattering experiments performed at room temperature. The signal of the methylene protons of the hydrocarbon core appears in the normal shape with the terminal methyl peak on the highfield region. Comparing this egg PC vesicles spectrum with that obtained from the vesicles prepared from mixing GDNT with egg PC at a molar ratio of 1:4, it is clear that the inner and outer choline signals are not as well separated at equal external concentrations of 5 mM Pr 3÷. The downfield shift of the outer choline signal is 6 Hz smaller in the mixed vesicles. This is due to the shielding of the praseodymium ions by the hydroxyl groups of nonitol present in the outer surface of the vesicles. In fact, increasing the concentration of the bipolar lipid to the molar ratio 1:2, the separation between inner and outer peaks decreases by a further 6 Hz. This behaviour suggests that phase separation had not occurred, a conclusion supported also by other experiments,
90
CH20H H ' ~ ~ O ~ .
H
1 ppm
I
!
CHOH CH~S
CI
R=H
TMS
GDGT
GLyc.roL GD6T~ _
BULK
I
CH~ CH~S TMS I
cy li n
GON.~
I
-N~CH3)3
i Gtyc,rot II oPh.P_.Lc ~L.__~,v,J
/I
BULK
It II
BULK
T,:
.JILl
CHiS
I
CH / v I I ',..._k TMS
-N(CH313
TMS
i
EggPC
Chotin~ CH~ Chotine Gtycerot
oH! CH
-N(CH3)3
CH2
DPPC Fig. 2. Proton NMR (80 MHz) spectra of C2HC13 solutions of egg PC, DPPC, diphytanoyl PC, GDNT and GDGT at T = 25°C. The concentration of the lipid is 37.5 mg/ml. The inset shows the chemical structure of GDGT and GDNT. The number of cyclopentane rings can vary from 0 to 4 per chain [4].
91
which will be discussed later. The ratio of the outer to the inner choline resonance is 1.3. In the Discussion section it will be shown that this measurement allows a very interesting conclusion. The hydrocarbon signal of egg PC in the mixed vesicles, shown in Fig. 3, collapses to approximately half its intensity and broadens due to the interactions with GDNT. The hydrocarbon of GDNT appears at the chemical shift of the terminal methyl signal of egg PC due to the different chemical composition in the hydrocarbon region of GDNT compared with that of egg PC. It can be deduced from the GDGT/egg PC mixed vesicle spectrum of Fig. 3 that at a molar ratio of 1:4, closed structures were not formed
HDO
I
~ydrocarbong '
v
-
EggPC
IA ¸
Fig. 3. Proton NMR (80 MHz) spectra of D20 sonicated dispersions of egg PC, GDNT/egg PC, GDGT/egg PC, DPPC and GDNT/DPPC containing 5 mM Pr s÷ in the external medium. Concentration of total lipid is 37.5 mg/ml and T = 60°C.
since the added praseodymium had immediately been completely internalized, hence a single broad resonance which belongs to both choline heads was observed. This is in agreement with electron microscopy results shown in Fig. 1 d,e,f. In spite of the same chemical composition of GDGT and GDNT in their hydrocarbon regions, their mixtures with egg PC give different packings, due to the presence of nonitol in the polar region of GDNT. Further, contrary to the results obtained with DPPC/GDNT vesicles where closed structures were obtained, we found it impossible even to form liposomes from GDNT and diphytanoyl PC mixed in the usual molar ratio of 1:4. This fact suggests that the lateral methyl groups of both lipids do not intercalate, but orient themselves opposite each other thus hindering hydrophilic interactions of the heads to take place on the outer side of the vesicle. If we now compare the spectrum of GDNT/ DPPC vesicles with that of pure DPPC vesicles at 60°C (Fig. 3) two major differences are observed. The ratio of the outer to the inner choline resonances is 1.3 instead of 1.6 and the hydrocarbon peak of DPPC collapses to one third of its original intensity in the mixed lipid vesicles, due to the decrease in motional freedom brought by the thermophilic lipid. However, both these vesicles have stable barriers for praseodymium ions, in spite of the shift of the outer monolayer choline resonance being up to I0 Hz smaller than for pure DPPC. It is interesting to examine the thermotropic properties of the vesicles prepared from mixtures of egg PC with the archaebacteriai lipids (Fig. 4). Scanning the ~H-NMR spectrum of pure egg PC vesicles at several temperatures between 25°C and 85°(2, the width at half height of the main hydrocarbon peak is reduced progressively, showing that both the motionai freedom of the lipid and the tumbling rate of the vesicles have increased. GDNT/egg PC vesicles spectrum at 27 °C show broad hydrocarbon spectrum, which changes with increase in temperature. At 72°C the spectrum shows equal intensities for both egg PC and GDNT hydrocarbon signals, and at 85°C egg PC signal intensity is higher than that of
92
CH2
A]
B)
lppm I~ppm w B
7z~/.A.////~\~..
1
PCGDNT
PC lppm C]/] }~DGT / B5oc
i
[,z_~,j
Fig. 4. Proton NMR (g0 MHz) spectra of the hydrocarbon region of the vesicles prepared from (A) egg PC, (B) GDNT/egg and (C) GDGT/egg PC at selected temperatures between T = 25°C and T = 85°C. Total lipid concentration is 37.5 mg/ml.
GDNT. GDGT/egg PC vesicles spectrum shows in comparison much less, but similar, changes with temperature (Fig. 4C). Figure 5A shows the NMR spectrum of the hydrocarbon region of pure DPPC vesicles below and at several degrees above its phase transition. Below the phase transition the hydrocarbon signal of the vesicles is broadened to the base line and, as expected, the signal sharpens with increase in temperature. The vesicles formed from GDNT/DPPC (Fig. 5B) show, in comparison, higher chain mobility at the temperature below the Tc of DPPC. GDNT, on the other hand, shows higher hydrocarbon signal intensity up to 70°C, while DPPC shows a sharper resonance at 85 °C. The graph in Fig. 6 shows a plot of the relative amplitudes of the IH-NMR signals of the hydrocarbon chains of the pure and the mixed
vesicles versus temperature; the inner choline signal was taken as a reference for the signal intensities. DPPC chains in the unmixed vesicles below and above the phase transition show a non-linear saturating behaviour (Fig. 6A) compared to the linear graph obtained when the vesicles were made from mixtures of DPPC with GDNT (Fig. 6C). This linearity was also observed for the hydrocarbon chains of GDNT in the mixed vesicles (Fig. 6D). The same linear behaviour was displayed by pure egg PC vesicles, which do not have any phase transition in the explored range of temperatures. A completely analogous result was obtained by plotting the line width Acol/2 at which the signal drops to half its maximal value. It is evident that no phase separation of the lipids had occurred in the mixed vesicles, as also indicated by the downfield shift of the outer choline signal and
93 CH
O!
AI
lppm I
OPPC CH3
40"C
Fig. 5. Proton NMR (80 MHz) spectra of the hydrocarbon regions of sonicated dispersions of: (A) DPPC and (B) DPPC/GNDT at selected temperatures between T = 40°C and T = 85 °C. Lipid concentration is 37.5 mg/ml. 1.8
by electron microscopy studies. This conclusion is in agreement with microcalorimetric studies of GDNT/DPPC mixtures [27]. In Fig. 7 the downfield shift of the inner choline resonance was measured immediately after addition of 5 mM Pr 3+ and 10 h later at 85°C to check for Pr 3. ion diffusion to the interior of the vesicles. It can be deduced that GDNT/egg PC vesicles have a barrier against Pr 3. ions as that o f pure egg PC. The permeability properties were not affected by varying the GDNT/egg PC ratio from 10 to 50 moW0. In fact, a downfield shift of 2--4 Hz of the inner choline signal was measured for both types o f vesicles, 10 h after the addition of 5ram Pr 3. in the external medium. G D N T / D P P C exhibited similar behaviour. By contrast, in G D G T / e g g PC sonicated dispersions the choline resonance was not shifted, in agreement with the fact that no closed structures were formed with a molar ratio of 1:4. Closed vesicles, 50 nm in diameter, were obtained only with a molar ratio of 1:6, a condition in which the inner and outer choline peaks were separated after addition o f Pr 3* ions. The
3.8
A
3.6
1A o
o
