Eur. J. Biochem. 193,921 -925 (1990)
( 3 FEBS 1990
Reconstitution of CFoF into liposomes using a new reconstitution procedure Peter RICHARD ’, Jean-Louis RIGAUD’ and Peter GRABER’ Max-Volmer-Instilut, Technische Universiliit Berlin, Berlin (West) Service de, Biophysique, Departement de Biologie and Unite de Recherche Associee (Centre National de la Recherche Scientifiquc) Centre d’Etudes Nucleaires de Saclay, Gif-sur-Yvette, France Biologischcs Institut, Universitat Stutlgart, Federal Republic of Germany (Received June 1 , 1990)
-
EJB 90 0628
The H+-ATPase (ATP synthase) from chloroplasts was isolated, purified and reconstituted into phosphatidylcholine/phosphatidic-acid liposomes. Liposomes prepared by reverse-phase evaporation were treated with various amounts of Triton X-100 and protein incorporation was studied at each step of the solubilization process. After detergent removal by SM,-Biobeads, the activities of the resulting proteoliposomes were measured indicating that the most efficient reconstitution was obtained by insertion of the protein into preformed, detergent-saturated liposomes. The conditions for the reconstitution were optimized with regard to ATP synthesis driven by an artificially generated ApH/d Y . An important benefit of the new reconstituted CFoFl liposomes is the finding that the rate of ATP synthesis remains constant up to 10 s, indicating a low basal membrane permeability. The H+-ATPasefrom chloroplasts (ATP synthase, CFoF catalyzes proton-transport-coupled ATP synthesis/hydrolysis. The enzyme has been isolated, purified and reconstituted [I]. The purification and the reconstitution into asolectin liposomes have been optimized. The maximal rate of ATP synthesis was 200 s- when the proteoliposomes were energized with an acid-base transition, dpH, combined with a K + / valinomycin diffusion potential, A Y , [2, 31. This is about half the rate obtained with thylakoids under identical reaction conditions [4]. Thus the reconstituted system is well suited for investigation of the mechanism of the enzyme. Kinetic investigations are difficult, however, since the d p H generated by the acid - base transition remains constant for only about 300 ms and, therefore, the rates must be measured with rapid-mixing techniques over this short time range. Presumably, there are several reasons for this disadvantage. (a) Asolectin has an intrinsically relatively high H permeability. (b) The size distribution of the liposomes is inhomogeneous. In small liposomes ApH decreases faster than in large liposomes. (c) The distribution of CFoFl into liposomes is inhomogeneous. Electron micrographs show vesicles with a high number of enzymes and some with no CFoFl at all. The ApH in the liposomes with the highest number of enzymes decreases most rapidly. Here we describe an extension of a reconstitution procedure recently developed for bacteriorhodopsin which allows rapid and easy determination of the optimal experimental conditions for detergent-mediated reconstitution of this protein [ 5 ] . Thus, optimal reconstitution of CFoF1was achieved in the presence of Triton X-100 by a mechanism involving a transfer of the protein from proteinidetergent micelles to +
Correvpondence to P. Richard, Max Volmer Institut, Technische
Universitat Berlin, Strasse des 17. Juni 135, D-1000 Berlin (West) 12 Ahhreviatiuns. CF0F1, proton-translocating ATPase from chloro-
plasts, ATP synthase. Enzyme. ATPase (EC 3.6.1.3).
detergent-saturated liposomes, allowing a good unidirectional orientation of CFoFl into the reconstituted proteoliposomes. Another advantage of this procedure relies on the use of SM,Biobeads as a detergent-removing agent, which provides a reproducible way of achieving impermeable liposomes. The reconstitution procedure was further optimized with regard to lipid/protein ratio and ionic composition of the medium, leading to proteoliposomes where reaction times up to 10 s can be obtained at constant ApH.
MATERIALS AND METHODS CFoF1was isolated and purified as described earlier [l, 31. After purification, the final enzyme solution contained 300 g/1 sucrose, 30 mM Tris/succinate, pH 6.5, 0.5 mM EDTA, 2 g/1 Triton X-100 (Sigma) and 1 gjl asolectin (Associated Concentrates, USA) with a protein concentration of 3 6 g/l. This solution was rapidly frozen and stored under liquid nitrogen until use. The protein concentration was determined as described by Bensadoun et al. [6]. The molar concentration of CFoFl was calculated with a molecular mass of 550 kDa. Phosphatidylcholine was isolated and purified from egg yolk as described by Singelton et al. [7]. Phosphatidic acid was prepared enzymatically from phosphatidylcholine as described by Allgyer and Wells [8]. The purified lipids were stored as chloroform solutions under nitrogen at - 20°C. Liposomes were prepared from phosphatidylcholinej phosphatidic acid (20: 1, by mass), using the reverse-phase method, as described by Paternostre et al. [9]. The final phospholipid concentration was 16 g/l in a buffer containing 40 mM Na2S04, 20 mM sodium Tricine and 20 mM sodium succinate, pH 8.0. These liposomes were passed with a syringe through polycarbonate membranes (Nucleopore) in order to obtain Iiposomes within a narrow size distribution. Two filters, one with 0.4 pm pore diameter and one with 0.2 pm, were used in subsequent steps. The resulting liposomes had a diameter of approximately 150 nm [9].
922 Reconstitution of CFoFl into liposomes is described under Results. For reconstitution experiments, liposomes prepared by reverse-phase evaporation were resuspended at 8 g/l in the buffer used for their preparation, supplemented with MgSO, to give a final concentration of 2.5 mM. To remove detergent, SM2-Biobeads (extensively washed before use, as described by Holloway [ 101) were added directly to the lipid/detergent/ protein mixture at a concentration of 80 mg wet beads/ml. The mixture was stirred at room temperature. After a 90-min incubation, a second portion of 80 mg Biobeadslml was added for an additional incubation period of 90 min. Then 160 mg Biobeads/ml were added for 30 min to remove residual detergent. Proteoliposomes were usually used directly after reconstitution, however they could be stored at 4 'C for five days without loss of activity. The activity of the reconstituted CFoFl was determined by measurement of the ATP yield after a ApH/d Y jump. One part of the proteoliposome suspension, usually 8 g/l lipid with 0.2 pM CFoF1, was added to five parts of the acidic medium containing 20 mM succinate, 5 mM NaH,PO,, 2.5 mM MgSO,, 0.6mM KOH. The pH was adjusted to 4.7 with NaOH. After mixing, the final pH was 5.0. Valinomycin was added freshly to a final concentration of 1 pM. After a 2-min incubation the same volume of a basic solution was added. The basic solution contains 200 mM Tricine, 5 mM NaH,PO,, 2.5 mM MgSO,, 120 mM KOH and 200 pM ADP was adjusted to pH 8.3 with NaOH. After mixing the final pH was 8.0. The reaction was stopped by addition of trichloroacetic acid (final concentration 2%). For measuring the overall ATP yield, the reaction was stopped after 30 s; for measuring the rate of ATP synthesis the reaction was stopped at different times between 0.5 s and 20 s and the ATP yield was plotted as a function of reaction time. For the zero time control, denaturation was carried out before addition of the basic medium. Throughout this work, the enzyme concentration during reconstitution is given. The CFoFl concentration in the reaction medium is always a factor of 12 lower. ATP was measured with lucifern/luciferase (Pharmacia, LKB) as described by Schmidt and Graber [2]. The sensitivity of this method was limited by the background ATP in ADP. The zero time control contained about 30 nM ATP. The smallest ATP yield shown in Fig. 1 was 30 mol ATP/inol CFoFl. This corresponded to an ATP concentration of 410 nM, i.e. it was a factor of 14 above the zero control. All reagents were of the highest commercially available grade.
RESULTS The standard procedure for reconstitution of CFoF, into liposomes was as follows. Liposomes prepared by the reversephase method, and filtered through Nucleopore filters, were resuspended in 40 inM Na,SO,, 20 mM sodium Tricine, 20 mM sodium succinate, pH 8.0, and 5 mM MgS04 at a phospholipid concentration of 8 g/l. The purified CFoFl preparation was solubilized in the presence of different amounts of detergent (1 -20 mg Triton X-lOO/ml), then the enzyme/ detergent solution was added under vortexing to the liposomes. The detergent/protein/phospholipidmixture was kept at room temperature for 1 h under gentle stirring, and the detergent was then removed by direct contact with SM,-Biobeads. The detergent/lipid ratio has been described as crucial to this reconstitution procedure, determining the mode of protein incorporation [5]. Therefore, the detergent concentration was
150
100 0 In W
6
50
Triton X-100 lipid
];I
0
Fig. 1. Apparent absorbance at 560 n m of the Iipmsome]detergent mixture as a-function o f Triton A'-IVV concentration. Lipid concentration was 4 g/1 (0).ATP yield after complete detergent removal by Biobeads as a function of Triton X-100 concentration during reconstitution. Lipid concentration was 8 g/l and CFoF, concentration was 165 n M ( 0 )
varied at constant lipid concentration (8 g/l) and constant enzyme concentration (165 nM). Fig. 1 shows the influence of the initial detergent/lipid ratio on the enzyme activity of the reconstituted proteoliposomes. This figure indicates the changes in the apparent absorbance of the mixture as a function of Triton X-100 concentration. Actually, there was no absorbance at 560 nm and the changes in transmitted light intensity were due to light scattering, reflecting the interaction between Triton X-100 and liposomes. With increasing detergent concentration, the apparent absorbance first increases, corresponding to detergent incorporation into liposomes without solubilization. At a critical detergent/phospholipid ratio of 0.5 (by mass), the liposomes were saturated with detergent. Increasing detergent concentration above this point produced gradual solubilization of the lipids, leading to a drastic decrease in the apparent absorbance. At a detergent/ phospholipid ratio of 2 (by mass), all the liposomes initially present were transformed into micelles and the suspension became optically transparent. When CFoFl was incubated in the presence of liposomes treated with subsolubilizing Triton X-100 concentrations (detergent/phospholipid ratio < 0.5) no ATP yield was detected after detergent removal. This indicates that the protein was not incorporated. Above this ratio CFoF1 was incorporated into liposomes as indicated by the ATP yield measured. The maximal ATP yield was obtained when the apparent absorbance reached about 50% of its maximal value. This might indicate that about 50% of the liposomes are solubilized. Similar experiments were carried out with other detergents such as n-octyl-fi-D-glucopyranoside (octyglucoside), cholate and dodecyl octa(oxyethy1ene) ether. The ATP yield obtained in these experiments was always lower (unpublished results). Therefore, only the reconstitution with Triton X-100 was optimized. The ATP yield measured in these experiments depended not only on the activity of the reconstituted enzyme, but also on the internal buffering capacity of the liposomes (i. e. on the size of the liposomes) and on the number of enzymes/liposome. With completely solubilized liposomes these parameters might change during detergent removal. Therefore, the initial rate of ATP synthesis was measured as a parameter of optimal reconstitution, since the rate depended only on the magnitude of the ApH/AY and enzyme activity. Fig. 2 A shows the ATP
923
A
(5) 2.0
1.5
1.12
0.87
0
50
100
150
time /(min)
0.62
Fig. 3. ATP yield us a function qf incubation time with Triton X-100 during reconstitution before addition of Biobead?. The ATP yield was measured after complete detergent removal, i.e. 3.5 h after the first addition of Biobeads. Lipid concentration was 4 g/l; CF,F, concentration was 165 n M ; Triton X-I00 concentration was 4.5 g/1
0.37
of its maximal value. We conclude from this result that a Triton X-lOO/lipid ratio of about one is the optimal value for reconstitution of CFoF1.The higher ATP yields obtained with higher detergent concentrations reflect the formation of bigger 0.12 liposomes i. e. the size distribution of the vesicles changed during reconstitution. Therefore, we always used a Triton 0 1 2 3 4 X-lOO/lipid ratio of about one. time 4 s ) Another feature of our results is the critical dependence of the activity of proteoliposomes at the time of incubation of CFoFl with detergent/phospholipid mixtures before detergent removal Fig. 3 shows ATP yield as a function of the incubation 2 time before addition of the Biobeads to the mixture. Since -LO 0, removal of the detergent by the Biobeads was slow, the actual incubation time was longer. It can be seen that even addition 5 -30 of Biobeads directly after mixing of the liposomes with the Y ln enzyme-detergent solution led to some reconstitution. How-20 z7 ever, the curve indicates that incubation for l h before addition of the Biobeads was optimal and a longer incubation had no - 10 t. effect. Therefore, in the following we always used an incubation time of 1 h. Finally, we analyzed, in detail, the influence of the lipid/ protein ratio on the efficiency of the reconstitution. The Triton X-100 amount of CF,F, for reconstitution was varied at constant lipid lipid concentration (8 g/l) and a Triton X-lOO/lipid ratio of Fig. 2. ATP yield and apparent absorbance at 560 nm as a function of one. After detergent removal, the ATP yield was measured. In Triton X-100 concentration. (A) ATP yield as a function of reaction this case, a reaction time of 30 s was used for measuring time at different Triton X-100 concentrations during reconstitution. ATP yield. Fig. 4 shows the result: with decreasing enzyme Lipid concentration was 8 g/1 and CFoFl concentration was 175 nM. concentration, the amount of ATP/CFoF1increased, reaching (B) Apparent absorbance a t 560 nm of the liposome/detergent mixture as a function of Triton X-100 concentration. Lipid concentration was a maximum at about 60 nM CFoF1. The constant value of the ATP yield implies that below this enzyme concentration 4 g/l (n).Rate of ATP synthesis after complete detergent removal by Biobeads as a function of Triton X-100 concentration during there was only one CF,Fl/liposome, or less (see Discussion). reconstitution ( 0 ) . The number of enzymes/vesicle obviously had a marked effect on the ATP yield, however, no effect was expected on yield as a function of reaction time for different detergent the initial rate of ATP synthesis. In Fig. 5 the ATP yield is concentrations during reconstitution under similar conditions plotted as a function of the reaction time for two different to those in Fig. 1, i.e. lipid concentration was 8 g/1 and CFOF, enzyme concentrations. The initial slope of both curves i.e. concentration was 175 nM. The slopes of these plots are the the rate of ATP synthesis was the same in both cases. At a rates of ATP synthesis. In Fig. 2B, the rate of ATP synthesis protein concentration of 175 nM, the rate was constant up to is shown as a function of the Triton X-lOO/lipid ratio together a reaction time of about 4 s and there was no further increase with the apparent absorbance of the initial mixture (data from in the yield after 10 s. At a CFoFl concentration of 43 nM, Fig. 2A). The maximal rate of ATP synthesis was obtained at the rate was constant up to about 10 s and no further increase a much lower detergent concentration than the maximal ATP was observed after 20 s. As long as the rate was constant, the yield i.e. when the absorption was decreased by about 20% initial ApH/AY was also constant. This result shows that at
[;I
924
T
150
-
100
-
a
j o t I
I
I
10
100
1000
enzyme concentration
0 ' 0
I
2
3
I
4
t i m e /(min)
I(nM)
Fig. 4. ATP jield as a ,firnetion of the CFoFl concentration. Lipid conccntration was 8 g/l, Triton X-100 concentration was 8 g/l. The reaction time for mcasuring the ATP yield was 30 s
I
1
Fig. 6. ATP yield as a,function of incubation time in the acidic bujrer. CFuF, concentration was 165 n M ; other conditions were as in Fig. 4
and basic media. The ATP yield did not depend on the sulfate concentration under our conditions. Only at low phosphate concentrations was inhibition observed. Additionally, there was no difference in the ATP yield when liposomes are prepared in the presence of chloride or sulfate. m
DISCUSSION
5 x
100
In this work a new procedure for reconstitution of CFoFl into liposomes containing phosphatidylcholine, and phos2 phatidic acid was optimized in order to obtain long reaction times at constant ApH. Preparation of liposomes and reconsti0 tution of CFoFl were carried out in two different steps. This 5 10 15 20 0 allowed optimization of liposome preparation without worrytime / ( s ) ing about denaturation of the enzyme. First, the liposomes Fig. 5. ATPj'ield us a,function of tlw reuction time with 175 nM CFoF, were prepared using the reverse-phase method and ( 0 )und43 nM CFoFl (M). Other conditions were as in Fig. 4 homogenized by passing the suspension through polycarbonate membranes, resulting in a considerable narrowing of the size distribution. With a pore size of 0.2 pm, liposomes low enzyme concentrations (one CF,F,/liposome or less) the of diameter approximately 150 30 nm were obtained. Sereaction time can be extended up to 10 s and the reaction cond, the liposomes were treated with different amounts of conditions (especially ApH/d Y ) remain constant. Triton X-100 to achieve the desired step in the lamellarThis long reaction time implies that the proton per- inicellar transition of the lipid. meability of the membrane of the reconstituted proteoSince solubilized CFoFl forms string-like structures, the liposome was very low. This low permeability might lead to a length of which depend on detergent concentration [13], the slow equilibration of the internal phase of the proteoliposomes protein was first mixed with detergent then added as a monowith the external phase during the acid incubation. Therefore, meric solution to the liposome suspension. The results from the ATP yield was measured after different incubation times reconstitution studies with Triton X-100 demonstrate that in the acidic medium. Fig. 6 shows the result. After about optimal reconstitution of CFoFl is detected in samples recon2 min the ATP yield became constant, and we conclude that stituted from detergent/lipid mixtures where about 70% of after this time, internal and external phases were completely the lipids were still present as Triton-X-100-saturated lipoequilibrated. At this point, it should be recalled that in the somes. A time-dependent incorporation of CFoFl into the cholate dialysis method the equilibration during acidic incu- preformed liposomes was observed, leading to the formation bation was already complete in about 15 s [2]. of homogeneous proteoliposomes. Such a reconstitution Preparation of liposomes, reconstitution of CFoFl and mechanism was already reported for Triton-X-100-mediated measurement of the ATP yield was carried out in the presence reconstitution of bacteriorodopsin [5]. An important conseof sulfate. However, it has been shown that the presence of quence of such a mechanism of lipid/protein association is sulfate inhibits ATP synthesis in chloroplasts [ I l , 121. There- related to the final orientation of the protein in the membrane. fore, we have carried out the following experiment. Liposomes It is generally admitted that the insertion of a protein into were prepared and CFoFl was reconstituted as described preformed liposomes leads to proteoliposomes with better above, except that in all solutions 40mM Na2S0, was re- asymmetric protein Orientation than when proteoliposomes placed by 80 mM NaCl, and MgS04 was replaced by MgC12. are formed by detergent removal from ternary phospholipid,/ Also, in the acidic and the basicmedia, the sulfate was replaced detergent/protein micelles [5, 141. A possiblc mechanism of by chloride. The ATP yicld after a ApH/AY jump was then CFoFl incorporation into preformed liposomes is that the measured as a function of th sulfate concentration in the acidic proteins are inserted through the hydrophobic domain of the L
925 membrane always with their more hydrophobic moiety first. Thus, the CFo part will be the first to penetrate the membrane, leading to almost unidirectional orientation, with the CF1 part to the outside of the resulting proteoliposome. The results in Fig. 2 B are in accordance with this mechanism. At a Triton X-I00/lipid ratio of one (by mass), i.e. when only unidirectional insertion occurs, a rate of 45 s - ' is observed. At a Triton X-IOOjlipid ratio of two (by mass), the liposomes are completely solubilized ;upon detergent removal, liposomes are formed again and the enzyme is inserted into the membrane. Under these conditions, only half of the enzymes can synthesize ATP and therefore a rate of 22 s-' should be observed. The observed rate was 35 s-'. This indicates that the enzyme was incorporated preferentially with the CF1 part to the outside when reconstitution was performed after complete solubilization of the liposomes. A vesicle with a diameter of 150 nm has a surface of 1.4 x 10' nm2 (inner pluse outer surface). Assuming an area of 0.6 nm2/lipid molecule [I51 each vesicle contains about 2.4 x 10' lipid molecules. Using a molecular mass of 800 Da for the lipids, a concentration of 8 g/1 is equal to 10 mM, and this corresponds to a vesicle concentration of 43 nM. If the enzyme concentration is varied at constant lipid concentration, it is expected that at about 50 nM CFoFl an average distribution of one CFnFl/vesicle is obtained. The results shown in Fig. 4 indicate that the ATP yield becomes constant at approximately this enzyme concentration. This implies that at lower enzyme concentrations, the same amount of internally buffered protons can always be used for ATP synthesis. This is possible only when the number of enzymes/vesicle is lower than one. This is in accordance with the calculation presented above. If the enzyme concentration is doubled, i.e. two CFoFl/vesicle,the ATP yield should drop by a Factor of two, since the same amount internally buffered protons is now distributed between two enzymes. This prediction is also fulfilled (see Fig. 4). It should be mentioned that the initial rate of ATP synthesis depends only on ApH and AY. It does not depend on the amount ofprotons available for the enzyme. This can be seen very clearly in Fig. 5 where the initial rates do not depend on enzyme concentration; the final yield, however, is considerably increased at low enzyme concentration. The method used here and the results differ in some respect from our earlier results where CFoFl was reconstituted into asolectin using the dialysis method [2, 161. With this method, vesicle formation and reconstitution occur simultaneously. Therefore, proteoliposomes can be formed where the enzyme is incorporated with the hydrophilic part directed to the internal aqueous phase. The size distribution of the proteoliposomes is broad and depends on the amount of incorporated protein [17]. Additionally, the protein is distributed inhomogeneously : some vesicles (mostly small ones) contain a high number of enzymes, other vesicles none at all 1181. The apparent proton permeability of the phosphatidylcholine/phosphatidic acid liposomes is much lower than that observed with asolectin liposomes. This can be seen from two experiments: first, it takes about 2 min incubation in the acidic medium for complete equilibration between the internal and external phases (see Fig. 6) compared to 30 s necessary for
asolectin liposomes. Secondly, the ApH during ATP synthesis remains constant up to 10 s, whereas with the asolectin liposomes the time is maximally 300ms. With lipids from the thermophilic cyanobacteria, even lower permeabilities have been observed. In this case it has been shown that proton efflux can be measured up to 60 min [19]. The maximal rate observed here after energization with ApHlAY is only 40 s p l . This is much lower than the rate of 200 s-' obtained with reconstitution into asolectin liposomes [2, 207. The reason for this difference is under investigation (nature of lipids, surface charge, activation, etc.). The reconstitution procedure in this work, however, has the following advantages. The liposomes have a narrow size distribution and a homogeneous distribution of the enzymes between the liposomes. The reaction time can be extended up to 10 s with the transmembrane energization remaining constant. This facilitates kinetic investigations since the measurements do not require a rapid-mixing machine. In this work phosphatidylcholine and phosphatidic acid were used at a molar ratio of 20:1, i.e. the vesicle membrane contains one negative charge/20 lipid molcules. This ratio can be easily changed. This allows the investigation of the effect of the surface charge on reconstitution and on ATP synthesis. This work was supported by the Deutsche Forschungsgemeinschuft (Sfb 312).
REFERENCES 1. Pick, U. & Racker, E. (1979) J . Biol. Chem. 254, 2793-2799. 2. Schmidt, G. & Grlber, P. (1 985) Biochim. Biophys. Acta 808,46 51. 3. Fromme. P., Boekema, E. J. & Griiber, P. (1987) Z. Nufurjbrsch. 42c, 1239- 1245. 4. Graber, P., Junesch, U. & Schatz, G. H. (1984) Ber. Bunsen-ges. Phys. Chem. 88, 599 -608. 5. Rigaud, J.-L., Paternostre, M.-T. & Bluzat, A. (1988) Biochemistry 27, 2671 - 2688. 6. Bensadoun, A. & Weinstein, P. (1976) Anal. Biocliem. 70, 241 250. 7. Singleton, W. S., Gray, M. S., Brown, M. L. &White, J. L. (2965) J . Am. Oil.Chem. Soc. 42, 53-51. 8. Allgyer, T. & Wells, M. A. (1 979) Biochemistry 18, 5348 - 5353. 9. Paternostre, M.-T., Roux, M. & Rigaud, J.-L. (1988) Biochemistry 27, 2668-2677. 10. Holloway, P. W. (1973) Anal. Biochem. 53, 304-308. 11. Ryrie, I. J. & Jagendorf, A. T. (1971) J . Biol. Chem. 246, 582588. 12. Uribe. E. G. & Li, B., C., Y. (1973) Bioenergetics 4,435-444. 13. Boekema, E. J., Fromme, P. & Graber, P. (1988) Ber. Bensen-ges. Phys. Cliem. 92, 1031 -1036. 14. Eytan, G . D. (1982) Biochim. Biophys. Actu 694, 185-202. 15. McLaughlin, S. (1977) Curr. Top. Memhr. Trans. 9, 71 - 144. 16. Sone, N., Yoshida, M., Hirata, H. & Kagawa, Y. (1977) J . Biochem. (Tokyo) 81, 519-528. 17. Fromme, P. (1 988) Thesis, Technische Universitlt Berlin. 18. Schmidt, G. (1987) Thesis, Technische Universitlt Berlin. 19. van Walraven, H. S., Hagendorn, M. J. M.. Krab. K., Haak, N. P. & Kraayenhof, R. (1985) Biochim.Biophys. Acta. 809,236244. 20. Schmidt, G. & Graber, P. (1987) Biocliim. Biophys. Acta 890, 392- 394.
( 3 FEBS 1990
Reconstitution of CFoF into liposomes using a new reconstitution procedure Peter RICHARD ’, Jean-Louis RIGAUD’ and Peter GRABER’ Max-Volmer-Instilut, Technische Universiliit Berlin, Berlin (West) Service de, Biophysique, Departement de Biologie and Unite de Recherche Associee (Centre National de la Recherche Scientifiquc) Centre d’Etudes Nucleaires de Saclay, Gif-sur-Yvette, France Biologischcs Institut, Universitat Stutlgart, Federal Republic of Germany (Received June 1 , 1990)
-
EJB 90 0628
The H+-ATPase (ATP synthase) from chloroplasts was isolated, purified and reconstituted into phosphatidylcholine/phosphatidic-acid liposomes. Liposomes prepared by reverse-phase evaporation were treated with various amounts of Triton X-100 and protein incorporation was studied at each step of the solubilization process. After detergent removal by SM,-Biobeads, the activities of the resulting proteoliposomes were measured indicating that the most efficient reconstitution was obtained by insertion of the protein into preformed, detergent-saturated liposomes. The conditions for the reconstitution were optimized with regard to ATP synthesis driven by an artificially generated ApH/d Y . An important benefit of the new reconstituted CFoFl liposomes is the finding that the rate of ATP synthesis remains constant up to 10 s, indicating a low basal membrane permeability. The H+-ATPasefrom chloroplasts (ATP synthase, CFoF catalyzes proton-transport-coupled ATP synthesis/hydrolysis. The enzyme has been isolated, purified and reconstituted [I]. The purification and the reconstitution into asolectin liposomes have been optimized. The maximal rate of ATP synthesis was 200 s- when the proteoliposomes were energized with an acid-base transition, dpH, combined with a K + / valinomycin diffusion potential, A Y , [2, 31. This is about half the rate obtained with thylakoids under identical reaction conditions [4]. Thus the reconstituted system is well suited for investigation of the mechanism of the enzyme. Kinetic investigations are difficult, however, since the d p H generated by the acid - base transition remains constant for only about 300 ms and, therefore, the rates must be measured with rapid-mixing techniques over this short time range. Presumably, there are several reasons for this disadvantage. (a) Asolectin has an intrinsically relatively high H permeability. (b) The size distribution of the liposomes is inhomogeneous. In small liposomes ApH decreases faster than in large liposomes. (c) The distribution of CFoFl into liposomes is inhomogeneous. Electron micrographs show vesicles with a high number of enzymes and some with no CFoFl at all. The ApH in the liposomes with the highest number of enzymes decreases most rapidly. Here we describe an extension of a reconstitution procedure recently developed for bacteriorhodopsin which allows rapid and easy determination of the optimal experimental conditions for detergent-mediated reconstitution of this protein [ 5 ] . Thus, optimal reconstitution of CFoF1was achieved in the presence of Triton X-100 by a mechanism involving a transfer of the protein from proteinidetergent micelles to +
Correvpondence to P. Richard, Max Volmer Institut, Technische
Universitat Berlin, Strasse des 17. Juni 135, D-1000 Berlin (West) 12 Ahhreviatiuns. CF0F1, proton-translocating ATPase from chloro-
plasts, ATP synthase. Enzyme. ATPase (EC 3.6.1.3).
detergent-saturated liposomes, allowing a good unidirectional orientation of CFoFl into the reconstituted proteoliposomes. Another advantage of this procedure relies on the use of SM,Biobeads as a detergent-removing agent, which provides a reproducible way of achieving impermeable liposomes. The reconstitution procedure was further optimized with regard to lipid/protein ratio and ionic composition of the medium, leading to proteoliposomes where reaction times up to 10 s can be obtained at constant ApH.
MATERIALS AND METHODS CFoF1was isolated and purified as described earlier [l, 31. After purification, the final enzyme solution contained 300 g/1 sucrose, 30 mM Tris/succinate, pH 6.5, 0.5 mM EDTA, 2 g/1 Triton X-100 (Sigma) and 1 gjl asolectin (Associated Concentrates, USA) with a protein concentration of 3 6 g/l. This solution was rapidly frozen and stored under liquid nitrogen until use. The protein concentration was determined as described by Bensadoun et al. [6]. The molar concentration of CFoFl was calculated with a molecular mass of 550 kDa. Phosphatidylcholine was isolated and purified from egg yolk as described by Singelton et al. [7]. Phosphatidic acid was prepared enzymatically from phosphatidylcholine as described by Allgyer and Wells [8]. The purified lipids were stored as chloroform solutions under nitrogen at - 20°C. Liposomes were prepared from phosphatidylcholinej phosphatidic acid (20: 1, by mass), using the reverse-phase method, as described by Paternostre et al. [9]. The final phospholipid concentration was 16 g/l in a buffer containing 40 mM Na2S04, 20 mM sodium Tricine and 20 mM sodium succinate, pH 8.0. These liposomes were passed with a syringe through polycarbonate membranes (Nucleopore) in order to obtain Iiposomes within a narrow size distribution. Two filters, one with 0.4 pm pore diameter and one with 0.2 pm, were used in subsequent steps. The resulting liposomes had a diameter of approximately 150 nm [9].
922 Reconstitution of CFoFl into liposomes is described under Results. For reconstitution experiments, liposomes prepared by reverse-phase evaporation were resuspended at 8 g/l in the buffer used for their preparation, supplemented with MgSO, to give a final concentration of 2.5 mM. To remove detergent, SM2-Biobeads (extensively washed before use, as described by Holloway [ 101) were added directly to the lipid/detergent/ protein mixture at a concentration of 80 mg wet beads/ml. The mixture was stirred at room temperature. After a 90-min incubation, a second portion of 80 mg Biobeadslml was added for an additional incubation period of 90 min. Then 160 mg Biobeads/ml were added for 30 min to remove residual detergent. Proteoliposomes were usually used directly after reconstitution, however they could be stored at 4 'C for five days without loss of activity. The activity of the reconstituted CFoFl was determined by measurement of the ATP yield after a ApH/d Y jump. One part of the proteoliposome suspension, usually 8 g/l lipid with 0.2 pM CFoF1, was added to five parts of the acidic medium containing 20 mM succinate, 5 mM NaH,PO,, 2.5 mM MgSO,, 0.6mM KOH. The pH was adjusted to 4.7 with NaOH. After mixing, the final pH was 5.0. Valinomycin was added freshly to a final concentration of 1 pM. After a 2-min incubation the same volume of a basic solution was added. The basic solution contains 200 mM Tricine, 5 mM NaH,PO,, 2.5 mM MgSO,, 120 mM KOH and 200 pM ADP was adjusted to pH 8.3 with NaOH. After mixing the final pH was 8.0. The reaction was stopped by addition of trichloroacetic acid (final concentration 2%). For measuring the overall ATP yield, the reaction was stopped after 30 s; for measuring the rate of ATP synthesis the reaction was stopped at different times between 0.5 s and 20 s and the ATP yield was plotted as a function of reaction time. For the zero time control, denaturation was carried out before addition of the basic medium. Throughout this work, the enzyme concentration during reconstitution is given. The CFoFl concentration in the reaction medium is always a factor of 12 lower. ATP was measured with lucifern/luciferase (Pharmacia, LKB) as described by Schmidt and Graber [2]. The sensitivity of this method was limited by the background ATP in ADP. The zero time control contained about 30 nM ATP. The smallest ATP yield shown in Fig. 1 was 30 mol ATP/inol CFoFl. This corresponded to an ATP concentration of 410 nM, i.e. it was a factor of 14 above the zero control. All reagents were of the highest commercially available grade.
RESULTS The standard procedure for reconstitution of CFoF, into liposomes was as follows. Liposomes prepared by the reversephase method, and filtered through Nucleopore filters, were resuspended in 40 inM Na,SO,, 20 mM sodium Tricine, 20 mM sodium succinate, pH 8.0, and 5 mM MgS04 at a phospholipid concentration of 8 g/l. The purified CFoFl preparation was solubilized in the presence of different amounts of detergent (1 -20 mg Triton X-lOO/ml), then the enzyme/ detergent solution was added under vortexing to the liposomes. The detergent/protein/phospholipidmixture was kept at room temperature for 1 h under gentle stirring, and the detergent was then removed by direct contact with SM,-Biobeads. The detergent/lipid ratio has been described as crucial to this reconstitution procedure, determining the mode of protein incorporation [5]. Therefore, the detergent concentration was
150
100 0 In W
6
50
Triton X-100 lipid
];I
0
Fig. 1. Apparent absorbance at 560 n m of the Iipmsome]detergent mixture as a-function o f Triton A'-IVV concentration. Lipid concentration was 4 g/1 (0).ATP yield after complete detergent removal by Biobeads as a function of Triton X-100 concentration during reconstitution. Lipid concentration was 8 g/l and CFoF, concentration was 165 n M ( 0 )
varied at constant lipid concentration (8 g/l) and constant enzyme concentration (165 nM). Fig. 1 shows the influence of the initial detergent/lipid ratio on the enzyme activity of the reconstituted proteoliposomes. This figure indicates the changes in the apparent absorbance of the mixture as a function of Triton X-100 concentration. Actually, there was no absorbance at 560 nm and the changes in transmitted light intensity were due to light scattering, reflecting the interaction between Triton X-100 and liposomes. With increasing detergent concentration, the apparent absorbance first increases, corresponding to detergent incorporation into liposomes without solubilization. At a critical detergent/phospholipid ratio of 0.5 (by mass), the liposomes were saturated with detergent. Increasing detergent concentration above this point produced gradual solubilization of the lipids, leading to a drastic decrease in the apparent absorbance. At a detergent/ phospholipid ratio of 2 (by mass), all the liposomes initially present were transformed into micelles and the suspension became optically transparent. When CFoFl was incubated in the presence of liposomes treated with subsolubilizing Triton X-100 concentrations (detergent/phospholipid ratio < 0.5) no ATP yield was detected after detergent removal. This indicates that the protein was not incorporated. Above this ratio CFoF1 was incorporated into liposomes as indicated by the ATP yield measured. The maximal ATP yield was obtained when the apparent absorbance reached about 50% of its maximal value. This might indicate that about 50% of the liposomes are solubilized. Similar experiments were carried out with other detergents such as n-octyl-fi-D-glucopyranoside (octyglucoside), cholate and dodecyl octa(oxyethy1ene) ether. The ATP yield obtained in these experiments was always lower (unpublished results). Therefore, only the reconstitution with Triton X-100 was optimized. The ATP yield measured in these experiments depended not only on the activity of the reconstituted enzyme, but also on the internal buffering capacity of the liposomes (i. e. on the size of the liposomes) and on the number of enzymes/liposome. With completely solubilized liposomes these parameters might change during detergent removal. Therefore, the initial rate of ATP synthesis was measured as a parameter of optimal reconstitution, since the rate depended only on the magnitude of the ApH/AY and enzyme activity. Fig. 2 A shows the ATP
923
A
(5) 2.0
1.5
1.12
0.87
0
50
100
150
time /(min)
0.62
Fig. 3. ATP yield us a function qf incubation time with Triton X-100 during reconstitution before addition of Biobead?. The ATP yield was measured after complete detergent removal, i.e. 3.5 h after the first addition of Biobeads. Lipid concentration was 4 g/l; CF,F, concentration was 165 n M ; Triton X-I00 concentration was 4.5 g/1
0.37
of its maximal value. We conclude from this result that a Triton X-lOO/lipid ratio of about one is the optimal value for reconstitution of CFoF1.The higher ATP yields obtained with higher detergent concentrations reflect the formation of bigger 0.12 liposomes i. e. the size distribution of the vesicles changed during reconstitution. Therefore, we always used a Triton 0 1 2 3 4 X-lOO/lipid ratio of about one. time 4 s ) Another feature of our results is the critical dependence of the activity of proteoliposomes at the time of incubation of CFoFl with detergent/phospholipid mixtures before detergent removal Fig. 3 shows ATP yield as a function of the incubation 2 time before addition of the Biobeads to the mixture. Since -LO 0, removal of the detergent by the Biobeads was slow, the actual incubation time was longer. It can be seen that even addition 5 -30 of Biobeads directly after mixing of the liposomes with the Y ln enzyme-detergent solution led to some reconstitution. How-20 z7 ever, the curve indicates that incubation for l h before addition of the Biobeads was optimal and a longer incubation had no - 10 t. effect. Therefore, in the following we always used an incubation time of 1 h. Finally, we analyzed, in detail, the influence of the lipid/ protein ratio on the efficiency of the reconstitution. The Triton X-100 amount of CF,F, for reconstitution was varied at constant lipid lipid concentration (8 g/l) and a Triton X-lOO/lipid ratio of Fig. 2. ATP yield and apparent absorbance at 560 nm as a function of one. After detergent removal, the ATP yield was measured. In Triton X-100 concentration. (A) ATP yield as a function of reaction this case, a reaction time of 30 s was used for measuring time at different Triton X-100 concentrations during reconstitution. ATP yield. Fig. 4 shows the result: with decreasing enzyme Lipid concentration was 8 g/1 and CFoFl concentration was 175 nM. concentration, the amount of ATP/CFoF1increased, reaching (B) Apparent absorbance a t 560 nm of the liposome/detergent mixture as a function of Triton X-100 concentration. Lipid concentration was a maximum at about 60 nM CFoF1. The constant value of the ATP yield implies that below this enzyme concentration 4 g/l (n).Rate of ATP synthesis after complete detergent removal by Biobeads as a function of Triton X-100 concentration during there was only one CF,Fl/liposome, or less (see Discussion). reconstitution ( 0 ) . The number of enzymes/vesicle obviously had a marked effect on the ATP yield, however, no effect was expected on yield as a function of reaction time for different detergent the initial rate of ATP synthesis. In Fig. 5 the ATP yield is concentrations during reconstitution under similar conditions plotted as a function of the reaction time for two different to those in Fig. 1, i.e. lipid concentration was 8 g/1 and CFOF, enzyme concentrations. The initial slope of both curves i.e. concentration was 175 nM. The slopes of these plots are the the rate of ATP synthesis was the same in both cases. At a rates of ATP synthesis. In Fig. 2B, the rate of ATP synthesis protein concentration of 175 nM, the rate was constant up to is shown as a function of the Triton X-lOO/lipid ratio together a reaction time of about 4 s and there was no further increase with the apparent absorbance of the initial mixture (data from in the yield after 10 s. At a CFoFl concentration of 43 nM, Fig. 2A). The maximal rate of ATP synthesis was obtained at the rate was constant up to about 10 s and no further increase a much lower detergent concentration than the maximal ATP was observed after 20 s. As long as the rate was constant, the yield i.e. when the absorption was decreased by about 20% initial ApH/AY was also constant. This result shows that at
[;I
924
T
150
-
100
-
a
j o t I
I
I
10
100
1000
enzyme concentration
0 ' 0
I
2
3
I
4
t i m e /(min)
I(nM)
Fig. 4. ATP jield as a ,firnetion of the CFoFl concentration. Lipid conccntration was 8 g/l, Triton X-100 concentration was 8 g/l. The reaction time for mcasuring the ATP yield was 30 s
I
1
Fig. 6. ATP yield as a,function of incubation time in the acidic bujrer. CFuF, concentration was 165 n M ; other conditions were as in Fig. 4
and basic media. The ATP yield did not depend on the sulfate concentration under our conditions. Only at low phosphate concentrations was inhibition observed. Additionally, there was no difference in the ATP yield when liposomes are prepared in the presence of chloride or sulfate. m
DISCUSSION
5 x
100
In this work a new procedure for reconstitution of CFoFl into liposomes containing phosphatidylcholine, and phos2 phatidic acid was optimized in order to obtain long reaction times at constant ApH. Preparation of liposomes and reconsti0 tution of CFoFl were carried out in two different steps. This 5 10 15 20 0 allowed optimization of liposome preparation without worrytime / ( s ) ing about denaturation of the enzyme. First, the liposomes Fig. 5. ATPj'ield us a,function of tlw reuction time with 175 nM CFoF, were prepared using the reverse-phase method and ( 0 )und43 nM CFoFl (M). Other conditions were as in Fig. 4 homogenized by passing the suspension through polycarbonate membranes, resulting in a considerable narrowing of the size distribution. With a pore size of 0.2 pm, liposomes low enzyme concentrations (one CF,F,/liposome or less) the of diameter approximately 150 30 nm were obtained. Sereaction time can be extended up to 10 s and the reaction cond, the liposomes were treated with different amounts of conditions (especially ApH/d Y ) remain constant. Triton X-100 to achieve the desired step in the lamellarThis long reaction time implies that the proton per- inicellar transition of the lipid. meability of the membrane of the reconstituted proteoSince solubilized CFoFl forms string-like structures, the liposome was very low. This low permeability might lead to a length of which depend on detergent concentration [13], the slow equilibration of the internal phase of the proteoliposomes protein was first mixed with detergent then added as a monowith the external phase during the acid incubation. Therefore, meric solution to the liposome suspension. The results from the ATP yield was measured after different incubation times reconstitution studies with Triton X-100 demonstrate that in the acidic medium. Fig. 6 shows the result. After about optimal reconstitution of CFoFl is detected in samples recon2 min the ATP yield became constant, and we conclude that stituted from detergent/lipid mixtures where about 70% of after this time, internal and external phases were completely the lipids were still present as Triton-X-100-saturated lipoequilibrated. At this point, it should be recalled that in the somes. A time-dependent incorporation of CFoFl into the cholate dialysis method the equilibration during acidic incu- preformed liposomes was observed, leading to the formation bation was already complete in about 15 s [2]. of homogeneous proteoliposomes. Such a reconstitution Preparation of liposomes, reconstitution of CFoFl and mechanism was already reported for Triton-X-100-mediated measurement of the ATP yield was carried out in the presence reconstitution of bacteriorodopsin [5]. An important conseof sulfate. However, it has been shown that the presence of quence of such a mechanism of lipid/protein association is sulfate inhibits ATP synthesis in chloroplasts [ I l , 121. There- related to the final orientation of the protein in the membrane. fore, we have carried out the following experiment. Liposomes It is generally admitted that the insertion of a protein into were prepared and CFoFl was reconstituted as described preformed liposomes leads to proteoliposomes with better above, except that in all solutions 40mM Na2S0, was re- asymmetric protein Orientation than when proteoliposomes placed by 80 mM NaCl, and MgS04 was replaced by MgC12. are formed by detergent removal from ternary phospholipid,/ Also, in the acidic and the basicmedia, the sulfate was replaced detergent/protein micelles [5, 141. A possiblc mechanism of by chloride. The ATP yicld after a ApH/AY jump was then CFoFl incorporation into preformed liposomes is that the measured as a function of th sulfate concentration in the acidic proteins are inserted through the hydrophobic domain of the L
925 membrane always with their more hydrophobic moiety first. Thus, the CFo part will be the first to penetrate the membrane, leading to almost unidirectional orientation, with the CF1 part to the outside of the resulting proteoliposome. The results in Fig. 2 B are in accordance with this mechanism. At a Triton X-I00/lipid ratio of one (by mass), i.e. when only unidirectional insertion occurs, a rate of 45 s - ' is observed. At a Triton X-IOOjlipid ratio of two (by mass), the liposomes are completely solubilized ;upon detergent removal, liposomes are formed again and the enzyme is inserted into the membrane. Under these conditions, only half of the enzymes can synthesize ATP and therefore a rate of 22 s-' should be observed. The observed rate was 35 s-'. This indicates that the enzyme was incorporated preferentially with the CF1 part to the outside when reconstitution was performed after complete solubilization of the liposomes. A vesicle with a diameter of 150 nm has a surface of 1.4 x 10' nm2 (inner pluse outer surface). Assuming an area of 0.6 nm2/lipid molecule [I51 each vesicle contains about 2.4 x 10' lipid molecules. Using a molecular mass of 800 Da for the lipids, a concentration of 8 g/1 is equal to 10 mM, and this corresponds to a vesicle concentration of 43 nM. If the enzyme concentration is varied at constant lipid concentration, it is expected that at about 50 nM CFoFl an average distribution of one CFnFl/vesicle is obtained. The results shown in Fig. 4 indicate that the ATP yield becomes constant at approximately this enzyme concentration. This implies that at lower enzyme concentrations, the same amount of internally buffered protons can always be used for ATP synthesis. This is possible only when the number of enzymes/vesicle is lower than one. This is in accordance with the calculation presented above. If the enzyme concentration is doubled, i.e. two CFoFl/vesicle,the ATP yield should drop by a Factor of two, since the same amount internally buffered protons is now distributed between two enzymes. This prediction is also fulfilled (see Fig. 4). It should be mentioned that the initial rate of ATP synthesis depends only on ApH and AY. It does not depend on the amount ofprotons available for the enzyme. This can be seen very clearly in Fig. 5 where the initial rates do not depend on enzyme concentration; the final yield, however, is considerably increased at low enzyme concentration. The method used here and the results differ in some respect from our earlier results where CFoFl was reconstituted into asolectin using the dialysis method [2, 161. With this method, vesicle formation and reconstitution occur simultaneously. Therefore, proteoliposomes can be formed where the enzyme is incorporated with the hydrophilic part directed to the internal aqueous phase. The size distribution of the proteoliposomes is broad and depends on the amount of incorporated protein [17]. Additionally, the protein is distributed inhomogeneously : some vesicles (mostly small ones) contain a high number of enzymes, other vesicles none at all 1181. The apparent proton permeability of the phosphatidylcholine/phosphatidic acid liposomes is much lower than that observed with asolectin liposomes. This can be seen from two experiments: first, it takes about 2 min incubation in the acidic medium for complete equilibration between the internal and external phases (see Fig. 6) compared to 30 s necessary for
asolectin liposomes. Secondly, the ApH during ATP synthesis remains constant up to 10 s, whereas with the asolectin liposomes the time is maximally 300ms. With lipids from the thermophilic cyanobacteria, even lower permeabilities have been observed. In this case it has been shown that proton efflux can be measured up to 60 min [19]. The maximal rate observed here after energization with ApHlAY is only 40 s p l . This is much lower than the rate of 200 s-' obtained with reconstitution into asolectin liposomes [2, 207. The reason for this difference is under investigation (nature of lipids, surface charge, activation, etc.). The reconstitution procedure in this work, however, has the following advantages. The liposomes have a narrow size distribution and a homogeneous distribution of the enzymes between the liposomes. The reaction time can be extended up to 10 s with the transmembrane energization remaining constant. This facilitates kinetic investigations since the measurements do not require a rapid-mixing machine. In this work phosphatidylcholine and phosphatidic acid were used at a molar ratio of 20:1, i.e. the vesicle membrane contains one negative charge/20 lipid molcules. This ratio can be easily changed. This allows the investigation of the effect of the surface charge on reconstitution and on ATP synthesis. This work was supported by the Deutsche Forschungsgemeinschuft (Sfb 312).
REFERENCES 1. Pick, U. & Racker, E. (1979) J . Biol. Chem. 254, 2793-2799. 2. Schmidt, G. & Grlber, P. (1 985) Biochim. Biophys. Acta 808,46 51. 3. Fromme. P., Boekema, E. J. & Griiber, P. (1987) Z. Nufurjbrsch. 42c, 1239- 1245. 4. Graber, P., Junesch, U. & Schatz, G. H. (1984) Ber. Bunsen-ges. Phys. Chem. 88, 599 -608. 5. Rigaud, J.-L., Paternostre, M.-T. & Bluzat, A. (1988) Biochemistry 27, 2671 - 2688. 6. Bensadoun, A. & Weinstein, P. (1976) Anal. Biocliem. 70, 241 250. 7. Singleton, W. S., Gray, M. S., Brown, M. L. &White, J. L. (2965) J . Am. Oil.Chem. Soc. 42, 53-51. 8. Allgyer, T. & Wells, M. A. (1 979) Biochemistry 18, 5348 - 5353. 9. Paternostre, M.-T., Roux, M. & Rigaud, J.-L. (1988) Biochemistry 27, 2668-2677. 10. Holloway, P. W. (1973) Anal. Biochem. 53, 304-308. 11. Ryrie, I. J. & Jagendorf, A. T. (1971) J . Biol. Chem. 246, 582588. 12. Uribe. E. G. & Li, B., C., Y. (1973) Bioenergetics 4,435-444. 13. Boekema, E. J., Fromme, P. & Graber, P. (1988) Ber. Bensen-ges. Phys. Cliem. 92, 1031 -1036. 14. Eytan, G . D. (1982) Biochim. Biophys. Actu 694, 185-202. 15. McLaughlin, S. (1977) Curr. Top. Memhr. Trans. 9, 71 - 144. 16. Sone, N., Yoshida, M., Hirata, H. & Kagawa, Y. (1977) J . Biochem. (Tokyo) 81, 519-528. 17. Fromme, P. (1 988) Thesis, Technische Universitlt Berlin. 18. Schmidt, G. (1987) Thesis, Technische Universitlt Berlin. 19. van Walraven, H. S., Hagendorn, M. J. M.. Krab. K., Haak, N. P. & Kraayenhof, R. (1985) Biochim.Biophys. Acta. 809,236244. 20. Schmidt, G. & Graber, P. (1987) Biocliim. Biophys. Acta 890, 392- 394.
