Natural membranes are organized sheetlike assemblies consisting mainly of lipids and proteins (figure 3.34). Phospholipids represent the main constituents of cellular membranes. They are relatively small molecules that have both a hydrophilic part, normally consisting of two long alkyl chains, and a hydrophobic moiety. These amphiphiles spontaneously form closed bimolecular sheets in aqueous media where the hydrophobic chains are in contact while the hydrophilic heads are exposed to the aqueous surrounding, thus constituting a barrier to the flow of polar molecules and ions. The proteins are embedded in the lipid bilayers, which creates suitable environments for their action. There are many processes occurring in biological membranes: membrane proteins can serve as ion pumps, gates, receptors, energy transducers and enzymes. The membrane composition in specific proteins varies according to the situation and role of the membrane in the cell. All constituents in membranes are hold together by many noncovalent cooperative interactions. In fact, membranes are fluid structures. Lipid molecules diffuse rapidly in the plane of the membrane, as do proteins. In contrast, they do not rotate across the membranes. Therefore, membranes can be regarded as two-dimensional solutions of oriented proteins and lipids.
Figure 3.34. Schematic diagram of a cellular membrane[94].
Artificial lipid bilayers provide simplified systems to study the complex biological membranes and the processes that are occurring therein. Also, these quasi two-dimensional, partially ordered structures with their unique elastic properties have led to a number of applications in biomedical devices (e.g., biocompatibilization of surfaces, drug delivery)[90]. These properties are a result of a unique double-layer structure and of the frictional drag between the two bilayer leaflets. Usual studies in this field include the introduction of organic molecules in the membranes in order to affect the permeability as well as the physical and mechanical properties of the current system. In this context, [60]fullerene and [70]fullerene have been introduced in lipid bilayers and their ability as both photosensitizers for electron transfer from donor molecules, and mediators for electron transport across the lipid bilayer have been studied[38][95][96]. However, the main limitation for these studies is the low solubility of plain fullerenes in bilayers. [60]Fullerene shows extraordinary lubrication properties[37] and its diameter of 10.5 Å is just one fifth of the average bilayer thickness. The intercalation of [60]fullerene molecules inside a lipid bilayer may affect the motions and flexibility of the whole structure thus giving rise to new outstanding material properties.
To overcome the solubility problem of plain [60]fullerene in bilayers, it was decided to synthesize different fullerene derivatives by attaching long alkyl chains, giving the so-called lipo-fullerenes. It seemed appropriate to synthesize Th-symmetrical [60]fullerene hexakisadducts containing six pairs of dodecyl (85) and octadecyl chains (86) (figure 3.35), corresponding to the chain length of lauric and stearic acid, respectively, using the efficient template method described above. Although this chemical modification certainly alters the rotational dynamics of the resulting molecules compared with plain [60]fullerene, and thus possibly its lubrication properties, it almost completely retains its unique symmetry, and may change the phospholipid bilayer to a composite system with new physical properties. It would also be of interest to study the intercalation of monoaddition products 87 and 88. These compounds have still an almost intact C60 skeleton and thus may also lead to interesting electronical properties of the resulting system.
Bromo malonates 91 and 92 were prepared from the commercially available 1-dodecanol and
1-octadecanol, respectively (scheme 3.19). Both were obtained in good yields after two steps.
[60]Fullerene monoadducts 87 and 88 were easily obtained under classical Bingel reaction conditions in high yields (scheme 3.19). Both targets were purified by flash chromatography using mixtures of hexane/toluene. The spectroscopic characterization of lipo-fullerenes 87 and 88 confirmed their structure.
Figure 3.35. (a) Lipo-fullerenes 85-88, (b) structure of the lipo-fullerene 86 optimized with the MM+ force field implemented in HYPERCHEM[64].
Th-Symmetrical lipo-fullerenes 85 and 86 were synthesized via template activation of [60]fullerene with DMA and subsequent exhaustive cyclopropanation of the [DMA]n-C60 complex with the corresponding bromo malonate 90 and 91 under basic conditions (scheme 3.20).
Scheme 3.19. Synthesis of lipo-fullerenes 87-88. i) malonyl dichloride, pyridine/CH2Cl2; ii) DBU, Br2/CH2Cl2; iii) C60, NaH/toluene.
Scheme 3.20. Synthesis of Th-symmetrical lipo-fullerenes C60-HAC12 (85) and C60-HAC18 (86).
Both lipo-fullerenes were purified by preparative HPLC on silica gel using mixtures of methylene chloride/hexane as eluent. Their Th-symmetry is clearly proved by their NMR spectra (figure 3.36). For example, the 1H-NMR spectrum of 86 displays a single set of peaks for all twelve aliphatic chains and its 13C-NMR spectrum shows only two types of sp2-carbons (141.17 and 145.83 ppm) and one type of sp3-carbon (69.08 ppm) for the [60]fullerene cage, together with the expected signals corresponding to the malonate moieties.
As expected, the IR and UV/Vis spectra of both hexakisadducts 85 and 86 are almost identical. The low melting point of both compounds is noteworthy: 22 0C for C60-HAC12 (85) and 65 0C for C60-HAC18 (86).
Figure 3.36. 1H-NMR (400 MHz, RT, CDCl3) and 13C-NMR (100.5 MHz, RT, CDCl3) spectra of lipo-fullerene 86 (* water). F denotes the signals for the fullerene carbon atoms.
In order to perform 2H-NMR investigations (see chapter 6.3.4), the perdeuterated analogue to C60-HAC18 (86) was required (scheme 3.21). For this propose, d35-stearic acid was quantitatively reduced with lithium aluminum deuteride and further converted into the symmetrical malonate 94. The target hexakisadduct C60-HAC18-d444 (95) was obtained after template activation of [60]fullerene and its subsequent cyclopropanation with malonate 94 according to the procedure introduced above.
Scheme 3.21. i) LiAlD4, THF (100 %); ii) malonyl dichloride, Pyr, CH2Cl2 (73.3 %); iii) (a) C60, DMA, toluene; (b) 94, CBr4, DBU (22.3 %).
The 13C-NMR spectrum of 95 (figure 3.37) showed five clearly resolved lines in agreement with the five different types of magnetically equivalent carbons not bounded to deuterium atoms. The signals corresponding to the carbons of the perdeuterated chain are not resolved as a consequence of the coupling of the 13C and the 2H nucleus.
Figure 3.37. 13C-NMR (100.5 MHz, RT, CDCl3) of 95. F denotes the fullerene carbon atoms.
No difference is found between the UV/Vis spectra of the d444-lipo-fullerene 95 and that of its non-deuterated analogue 86. In the IR spectra, new intense bands corresponding to the carbon-deuterium stretching vibration appear at 2200 cm-1 instead of 2900 cm-1, which is the normal position for carbon-hydrogen stretching bands.
Multilamellar vesicles (MLVs) of dipalmitoyl-sn-glycero-3-phosphatidylcholine (DPPC) were chosen as membranoic model system due to their easy accessibility (figure 3.38). DPPC consists of an apolar, lipophilic molecule part (two palmitoic ester chains), as well as a polar, hydrophilic part (the phosphatidylcholine moiety). Multilamellar Vesicles of DPPC with onion-like multiple bilayer structures are formed by vigorous stirring or ultra-sonication of DPPC in water. It was planned to carry out the formation of MLVs of DPPC in the presence of the different lipo-fullerenes 85-88 in order to trap them in the lipophilic interior of the bilayers. All these experiments have been performed in collaboration with Prof. T. M. Bayerl and his group at the University of Würzburg.
Figure 3.38. (a) dipalmitoyl-sn-glycero-3-phosphatidylcholine (DPPC); (b) schematic depiction of a multilamellar vesicle (MLV) of DPPC.
DPPC/lipo-fullerene MLV aggregates were prepared by mixing the two components in chloroform at room temperature, followed by lyophilization at liquid nitrogen temperature and a pressure of 15 hPa and overnight vacuum desiccation. The powder was then hydrated in water (5 mg of DPPC/lipo-fullerene mixture in 1 mL H2O) at 50 0C and allowed to swell at this temperature for 1 h under gentle vortexing. The formation of MLVs containing either
C60-HAC12 (85) or C60-HAC18 (86) was successfully achieved. The resulting MLVs appeared yellow colored, as the parent lipo-fullerenes, and homogeneous by optical inspection. No precipitation of lipo-fullerenes was observed when up to 25 mol % of lipo-fullerene were employed, even after the samples were stored in the refrigerator for several weeks. Unfortunately, the formation of MLVs containing monoaddition lipo-fullerenes 87 or 88 was not accomplished under the same conditions. The samples appeared heterogeneous due to the presence of a brownish precipitate consisting of lipo-fullerenes.
Differential scanning calorimetry (DSC)[98] is a convenient method to study the thermodynamic properties of membrane model systems[99]. In that way it is possible to measure directly or indirectly parameters corresponding to the transition behavior of the sample (phase transition temperature, transition enthalpy, heat capacity, etc.), its purity, or the interaction between several components. Briefly, the calorimeter takes advantage of the heat-leak principle. DSC is based on the measurement of the difference in power requirements for a sample in comparison with a reference, both heating (or cooling) at the same constant rate.
Multilamellar vesicles (MLVs) of DPPC containing C60-HAC12 (85) and C60-HAC18 (86), respectively, were studied at different mixing ratios by DSC[100]. These results were compared with those obtained for pure samples of C60-HAC12 (85) and C60-HAC18 (86) (figure 3.39).
The DSC curve obtained for pure C60-HAC18 (86) (figure 3.39.b) shows a peculiar profile. Heating the sample up from 20 0C to 70 0C causes the sample to undergo two major structural transitions. At 55 0C an exothermic transition, and at 64 0C an endothermic melting transition are observed. This phenomenon has been extensively studied by 2H-NMR and X-ray scattering[101]. At 55 0C an exothermic transition from a low-temperature, hard sphere fashion packing state of the molecules with their separation distances (61 Å) slightly above the maximum diameter of the molecules, to a condensed one involving partial interdigitation of the alkyl chains belonging to adjacent molecules, is observed. This is preceded by a local melting of the chains, in order to tilt chains to bunches as a prerequisite for interdigitation. The interdigitation allows a denser packing with an average separation distance of 48 Å. The driving force for this process is the optimization of strong van der Waals interactions between the chains, which is hampered by the stiffness of the chains leading to a hard sphere packing at temperatures below 50 0C. At a temperature of 64 0C an endothermic melting transition from the interdigitated to a viscous fluid state with average separation distances of 28 Å is observed. Cooling the sample from 70 0C causes a direct transition from the fluid into the low temperature state with no interdigitation in between.
Figure 3.39. Endotherms (heating scans) and exotherms (cooling scans) corresponding to (a) pure 85, (b) pure 86, (c) MLVs of pure DPPC, (d) MLVs of DPPC containing 15 mol % of lipo-fullerene 85, and (e) MLVs of DPPC containing 15 mol % of lipo-fullerene 86. Solid curves correspond to the heating and the dashed ones to the cooling scans.
Comparing the results obtained from DSC measurements it was concluded that the effect of both lipo-fullerenes 85 and 86 on the phase transition behavior of the MLVs is surprisingly low. A broadening of the main phase transition corresponding to the melting of DPPC (Tm = 410C) by a factor of 2-3 and a slight reduction of the phase transition temperature by 0.50C were observed in MLVs containing lipo-fullerenes in comparison with homogeneous DPPC MLVs. Even the pre-transition of DPPC which can be observed for heating scans at
Tp = 30.50C was not completely eliminated by the presence of the lipo-fullerenes. In the case of MLVs of DPPC containing 15 mol % of lipo-fullerene C60-HAC18 (86) (figure 3.39.e) a second big peak was observed at 710C (heating scan) and at 380C (cooling scan). Since plain 86 exhibits a transition at nearly the same temperature and with similar hysteresis (figure 3.39.b), this peak was assigned to the chain melting transition of the lipo-fullerene 86. In contrast, plain C60-HAC12 (85) (figure 3.39.a) has a phase transition at 21 0C (heating scan) and -7 0C (cooling scan). Therefore, no lipo-fullerene transition was observed since the sample corresponding to MLVs of DPPC containing 15 mol % of lipo-fullerene 85 was not cooled below 0 0C and hence its chains were always fluid during the measurements.
Increasing the lipo-fullerene content of the sample up to 25 mol % did not change the DSC results significantly. The only remarkable changes were a gradual attenuation of the DPPC main transition (by a factor of 3 - 4 at 25 mol % compared to pure DPPC transitions).
From these results it is possible to conclude that the phase transitions of DPPC and of 85 or 86 are completely decoupled, indicating pronounced microscopic demixing. This is, the interaction between lipo-fullerenes and DPPC in the MLVs seems to be at least very small. The unusually large hysteresis of the C60-HAC18 (86) transition is most likely a result of the rapid rotation of the [60]fullerene core in a microenvironment made up mainly of fluid octadecyl chains.
Freeze fracture electron microscopy was employed to study the morphological changes in the MLVs caused by the lipo-fullerenes. Figure 3.40 shows freeze fracture replicas of DPPC-d62 MLVs without (figure 3.40.a) and with (figure 3.40.b) 15 mol % of lipo-fullerene C60-HAC12 (85) under fluid phase conditions of the bilayer (samples quenched at 600C). In both cases predominantly large MLVs with diameters up to 5 
m with the typical onion like structure were observed. However, while the plain MLVs showed smooth fracture faces in the bilayer plane, the inclusion of C60-HAC12 (85) caused the formation of a rod-like surface structure that may look, at first glance, similar to the so called ripple phase of DPPC. Ripple phase was observed in plain DPPC MLVs at temperatures between 34 and 41 0C, i.e. below the main phase transition, and is shown in figure 3.40.c. However, it is clear for the following reasons that figure 3.40.b does not represent a ripple phase:
1) The bilayer exhibits a fluid state at 60 0C, and a ripple phase has never been observed under such conditions. Quenching the sample at even higher temperatures (70 0C) gave a similar result, thus ruling out the possibility that temperature artifacts during the freezing procedure were responsible for this behavior.
Figure 3.40. Freeze fracture micrographs of multilamellar vesicles of DPPC-d62 without (a) and (c), and with 15 mol % of the lipo-fullerene C60-HAC12 (b) and (d). Samples (a) and (b) were quenched at 60 0C, (c) at 40 0C, and (d) at 25 0C respectively. The bars represent a size of 0.5
2) The diameter of the individual rods was 10-30 nm and thus does not agree with the known dimensions of ripples[102]. The diameter increased with increasing the lipo-fullerene concentration in the bilayer.
3) The structures in figure 3.40.b were not coherently in phase over the bilayers, as is typical for ripples in MLVs.
No significant differences to the above mentioned results were observed when lipo-fullerene C60-HAC12 (85) was replaced by lipo-fullerene C60-HAC18 (86); this holds also true when the quenching of the sample was carried out at 60 0C under conditions where the octadecyl chains were in all-trans conformation. A further interesting feature is the (reversible) long range ordering of the above mentioned rod-like structures under gel phase conditions of the bilayer (quenching the samples at 25 0C). Here, for both lipo-fullerenes a stratification of the rods over distances of several m and the formation of superstructures that looked like bundles of individual rods are observed (figure 3.40.d). Again, the apparent rod diameter (up to 30 nm) increased with increasing the lipo-fullerene concentration.
The dynamic of a lipid bilayer consists of several types of movements of the individual molecules composing the system. Intercalation of lipo-fullerenes between the two monolayers of a lipid bilayer may affect the hierarchy of motions in bilayer, which covers nearly twelve orders of magnitude in time from hertz to terahertz[103][104]. These motions include: internal chain movements (frequency range 10-9 to 10-14 s), flip-flop movements of the polar heads (10-9 to 10-12 s), lipid rotations (10-9 s), lateral diffusions (10-6 to 10-1 s), and collective undulations and related movements (10-6 to 101 s). 2H-NMR is a powerful tool to study changes to the molecular order in the bilayer resulting from the intercalation of lipo fullerenes[103][104]. Several studies were performed combining the use of C60-HAC18-d444 (95), C60-HAC12 (85), C60-HAC18 (86), DPPC, and DPPC with selectively (DPPC-d8) and completely deuterated chains (DPPC-d62). From these studies it was possible to conclude that:
1) The presence of the lipo-fullerene does not cause significant changes in the distribution of the methylene groups of the lipid chains from the DPPC, but the methyl group reveals some significant effect. The interactions between lipo-fullerene and DPPC in the MLVs seem to take place mainly in the extremities, i.e., the lipo-fullerenes are in contact between them and the methyl group of DPPC, while the rest of the DPPC molecules remains unaffected.
2) Plain DPPC MLVs exhibit a macroscopic magnetic field orientation due to the diamagnetic susceptibility anisotropy of the lipid[105][106], which results in an elliptical MLV shape, with its long axis parallel to the magnetic field. MLVs containing lipo-fullerenes do not experience this effect. The typical MLVs spherical shape remains unaltered in front of an external magnetic field. The lipo-fullerenes seem to stiffen the fluid bilayer without altering its molecular order.
3) The effect of lipo-fullerenes on bilayer dynamics was studied by measuring the 2H-NMR longitudinal (T1z) and transverse (T2e) relaxation of the deuterated DPPC MLVs, with and without the lipo-fullerene in the fluid phase bilayer. The high frequency dynamics of the DPPC are not affected by the motion of the lipo-fullerenes, but some motions of the DPPC, which are slow on the NMR time scale (10-5 s), change significantly according to the reduction of the values of T2e[107]-[109]. The most likely reason for the T2e reduction, also in agreement with the microscopic results, is the formation of a wavy surface structure with a higher local curvature of the lipo-fullerene MLVs compared to the control experiment. It can be assumed that the lateral diffusion of the DPPC over this surface, and the resulting modulation of the molecular director axis, contribute to the ‘slow motion’ transverse relaxation mechanisms discussed previously.
An arrangement of the lipo-fullerenes in the bilayer as depicted in figure 3.41 is suggested. In this model, which is capable of explaining all the experimental observations, the lipo-fullerenes form rod-like structures on their own, with the rods sandwiched between the two layers of the bilayer. The rods appear rather disordered in the fluid bilayer (figure 3.41.A), but become stratified and long range ordered (rod axis parallel to adjacent rods in the same plane) in the gel phase bilayer (figure 3.41.B). This is probably induced by the high order of the lipids in the host bilayer under gel phase conditions. The lipo-fullerene rods would stabilize the bilayer against deformation in a magnetic field in a similar fashion to the way a membrane cytoskeleton does, with the important difference that the former seems to be located between the two layer leaflets, while the cytoskeleton is attached at the outside of the membrane.
The data do not permit to distinguish directly between a hollow rod and a massive rod filled by lipo-fullerene molecules. However, the obviously high lipo-fullerene capacity of the bilayers, together with the observation of increasing rod diameter with increasing the lipo-fullerene concentration, strongly suggests massive rods. This would allow the bilayer to take up tremendous amounts of lipo-fullerene without enhancing significantly the DPPC-rod contact area. Considering that rods have a diameter of about 30 nm and a length of 1.5 m according to the microscopy results, and that the volume of a molecule of lipo-fullerene
C60-HAC18 (86) is about 5 nm3, it is possible to conclude that when the concentration of lipo-fullerene in the MLVs is 15 %, a maximum of 10-15 % of the bilayer is in close contact with the rods, while the remaining bilayer area is left undisturbed.
Figure 3.41. Schematic depiction of the suggested arrangement of lipo-fullerenes: (A) in a fluid bilayer, and (B) in a gel phase bilayer.