ARCHIVES
OF RIOCHEMISTRY
AND
BIOPHYSICS
Vol. 278, No. 1, April, pp. 142-147, 1990
Cholesterol Distribution in Rat Liver and Brain Mitochondria as Determined by Stopped-Flow Kinetics with Filipin G. Crkmel,
D. Filliol,
Centre de Neurochimie
V. Jancsik,’
and A. Rendon”
du CNRS et U44 de I’INSERM.
Received July 25, 1989, and in revised form November
Strasbourg, France
16,1989
Recently, analysis of protein distribution in rat brain mitochondria suggested the existence of distinct cholesterol domains in the outer membrane (Dorbani et al., 1987, Arch. Biochem. Biophys. 252, 188-196) while such domains were not detected in rat liver mitochondria (Jancsik et al., 1988, Arch. Biochem. Biophys. 264, 295-301). We studied cholesterol distribution in both types of mitochondria by analyzing the kinetics of filipin-cholesterol complex formation, using the stopped-flow technique. In liver mitochondria, the kinetics are characterized by a biphasic curve which presumably corresponds to the two membranes. This was confirmed by the finding that pretreatment with digitonin abolished one of the kinetic components. Sonication of the mitochondria increased the rate of the filipin-cholesterol complex formation and also abolished one of the two components. In the case of brain mitochondria, several distinct cholesterol domains could be revealed: one of them was cholesterol-free and it was directly accessible to filipin. Two other domains were revealed by differences found in the rate of the cholesterol-filipin complex formation. It is noteworthy that only a part of the cholesterol is accessible to filipin. Sonication of mitochondria decreased the proportion of cholesterol molecules accessible to filipin. This suggests specific interactions of cholesterol with other mitowhich occur only in brain chondrial components, mitochondria. cc 1990 Academic Press, Inc.
Cholesterol is a widely distributed logical membranes that is important
constituent of biofor their structure
i Permanent address: Institute of Enzymology, BRC, Hungarian Academy of Sciences, Hungary. ’ To whom correspondence should be addressed at Centre de Neurochimie du CNRS, 5, rue Blaise Pascal, 67084 Strasbourg Cedex, France.
and function. It is unevenly distributed among cellular membranes. Plasma membrane has a cholesterol/phospholipid (C/P)” ratio as high as 0.8-0.9 whereas the intracellular membranes have a ratio of less than 0.2 (1, 2). In heart and liver mitochondria the C/P ratio is 0.04 and 0.1, respectively (3, 4), and cholesterol is localized mainly in the outer membrane. Data obtained with rat liver or heart mitochondria were frequently extrapolated to other mitochondria as well, e.g., to brain mitochondria (5, 6). Only limited information is available on the role of cholesterol in the control and regulation of mitochondrial functions. Dependence of protein function was described for the adenine nucleotide translocator (7), ATPase (8), and ,Bhydroxybutyrate dehydrogenase (9). Strong interaction of ergosterol with porin in Neurospora crassa was also reported (10). We suppose (11) that the preferential localization of cholesterol in the outer membrane could regulate interactions between mitochondria and other cell constituents, e.g., in the case of axonal transport of brain mitochondria (12,13). In a previous paper (6), on the basis of sequential release of marker proteins with increasing concentrations of digitonin, three outer membrane domains have been identified in brain mitochondria, one of them devoid of cholesterol, representing the contact points between outer and inner membrane. In contrast to the liver mitochondria, where porin segregation was not observed (14), in the brain mitochondria a part of the porin-hexokinase complex was located in the contact domain. Recent studies of Mannella (15) also suggested a lateral segregation of porin as a function of cholesterol in mitochondrial membranes of N. crassa. In the present study, we analyze the complex formation between filipin and brain or liver mitochondrial cholesterol by stopped-flow experiments. The significance of the different membrane domains in mitochondria is discussed. ” Abbreviation
used: C/P, cholesterol/phospholipid.
142 All
0003.9861/90 $3.00 Copyright 0 1990 by Academic Press, Inc. rights of reproduction in any form reserved.
CHOLESTEROL
MATERIALS
AND
DOMAINS
IN BRAIN
METHODS
Chemicals. The chemicals used were of the highest grade of purity available. Digitonin was obtained from Merck; filipin, egg phosphatidylcholine (Type XI-E), and cholesterol were from Sigma and were used without further purification. Ficoll and Percoll were from Pharmacia. and mitochondrial subfractionaPreparation of mitochondria tion. Rat liver mitochondria were prepared as previously described (16): nonsynaptic rat brain mitochondria were isolated following two methods, one of them using a discontinuous Ficoll gradient (17) and the other using a Percoll gradient procedure (18). Similar results were obtained with both types of mitochondrial preparations. Mitochondria were fractionated by digitonin, according to Dorbani et al. (6). In brief, aliquots of ice-cold digitonin solution at various concentrations were added to equal volumes of mitochondrial suspensions and the reaction mixtures were continuously stirred for 15 min at 0°C and then diluted with 3 vol of isolated medium. After centrifugation for 20 min at 12,5OOg, pellet was resuspended in isolation medium. In the sonication procedure, mitochondria at 10 mg/ml were sonicated with a MSE 150-type sonifier for 3 X 30 s at 0°C (19). Small unilamrllar vesicles preparation. Mixtures of phosphatidylcholine, cholesterol, and dicetylphosphate (7:2:1) dissolved in chloroform were evaporated to dryness under nitrogen. The lipidic film was rehydrated overnight in a buffer containing 5 mM Hepes, 0.1 mM EDTA, pH 7.6, at 4°C. The suspension was vortexed, then sonicated under nitrogen for 3 X 10 min at a frequency of 20 kHz and a 140.em amplitude with a MSE sonifier. During sonication the vial was kept in a water bath at a temperature of 5°C. The resulting clear dispersion was centrifuged for 120 min at 145,OOOgto eliminate multilayers and titanium particles of the probe. For some experiments, the cholesterol amount was modified and varied from 0 to 20%. Kinetic measurements. Stock solutions of filipin were prepared in dimethylformamide according to Blau and Bittman (20). Filipin concentration was determined by its molar absorbance at 360 nm (E = 4.5 X 104M ‘cm ‘). The formation of filipin-cholesterol complex is characterized by a decrease of absorbance at 357 nm. These changes were monitored after (routhe rapid mixing of 30 pM filipin solutions with mitochondria tinely, 0.5 mg/ml), lipid vesicles, or cholesterol solution in a stoppedflow spectrometer (PQ/SF 53 model, Hi-Tech Scientific, England). A slit width of 2 mm was used. Experiments were carried out at 25 t O.l”C. Data were analyzed on a Hewlett-Packard HP 9000 computer.
RESULTS
Filipin binds selectively to sterols which have a planar ring system, a free 3 /3-hydroxyl group, and an aliphatic side chain. The stoichiometry of the filipin-sterol complex has been found to be approximately one to one mole (21-25). In order to complex all the mitochondrial cholesterol, we used filipin in excess. The reaction of filipin with cholesterol was analyzed by the stopped-flow method. For cholesterol (20 PM) diluted from an ethanolic stock solution in isolation medium or for lipid vesicles containing various amounts of cholesterol, kinetic curves could be fitted by the sum of two exponentials (Fig. 1). It is reasonable to assume that cholesterol is uniformly distributed in solution or in these model vesicles. We can test cholesterol distribution in other membranes, e.g., mitochondria by this approach. The exactness of a fitting by the sum of two exponentials allows
AND
RAT
143
MITOCHONDRIA
for the conclusion that cholesterol is homogeneously distributed in the membrane tested while failing of a similar fitting is a good indication for inhomogeneous cholesterol distribution within the membranes. Furthermore, under the same experimental conditions, analysis of the interactions of filipin with small unilamellar phospholipid vesicles which did not contain cholesterol or with Triton X-100 micelles also without cholesterol showed an initial absorbance increase (data not shown). Cholesterol-Filipin Interaction Rat Brain Mitochondria
in
Cholesterol-filipin interactions in intact brain mitochondria are characterized by at least three phases (Fig. 2). During the first 100 ms, we observed an increase of absorbance. The second phase is discernible until 600 ms. At this time, a shoulder appears and the third phase, which seems to be of an exponential type, is observable at about 1 s. These results suggest the existence of distinct cholesterol domains according to their filipin accessibility. In order to induce a more homogeneous cholesterol distribution, we attempted to destroy the membrane organization by freezing and thawing the mitochondria followed by sonication. As can be seen, the initial increase of absorbance and the shoulder disappeared and it is possible to fit the curve to a sum of two exponentials. These results suggest the existence of at least three domains in the mitochondria, regarding the cholesterol content: one domain is devoid of cholesterol, corresponding to the increase of absorbance; the two other domains correspond to the two phases of the kinetics. These two domains could correspond to the outer and inner membranes. We studied the rate and extent of filipin-cholesterol complex formation in mitochondria preincubated with digitonin. Digitonin is known to complex cholesterol with a 1:l stoichiometry (26) and it is widely used to separate the two mitochondrial membranes (6, 27). It is believed that digitonin provokes the selective disruption of the outer membrane. We used 1 to 10 pg of digitoninl mg mitochondrial protein. On the other hand, by adding digitonin to lipid vesicles containing cholesterol, we found that filipin could not react with cholesterol complexed to digitonin; in this case, absorbance decrease was abolished. At these digitonin concentrations, no release of soluble mitochondrial protein occured (6,14). If we assume a cholesterol/protein ratio of 20 pg/mg proteins (6), the molar digitonin/cholesterol ratio varies from 0.015 to 0.15. At 1 pg/mg (Fig. 1) or 2 pg/mg digitonin (not shown), the pattern of the kinetics is similar to the intact mitochondria. At 5 pg/mg digitonin, the shoulder disappears, the total decrease of absorbance diminishes, and, comparatively, the initial absorbance increase is more pronounced. At 10 yg/mg digitonin, the
144
CRRMEL
ET AL.
TEROL IN SOLUTION
B
0.015 i
0.6
0
1
I
h
‘s”‘~‘~~‘~*~a~““l Orn' TIME IN SECONDS
CHOLES TEROL IN LIPID VESICLES
0
Om2
" '
TIME IN SECONDS
Om2
FIG. 1 Kinetics of complex formation between filipin and cholesterol. Cholesterol is dissolved in aqueous buffer from ethanolic solution or contained in lipid vesicles (phosphatidylcholine:cholesterol:dicetylphosphate ratio was 7:2:1 for the experiment shown here). (+) Experimental points, (-) fitted curve from the sum of two exponentials. Insets, the residual curve is obtained by the differences between the experimental points and the fitted curves. For comparison, curves are normalized to 1 at the maximal absorbance.
latter pattern is accentuated. At all digitonin concentrations tested, the initial rate of absorbance decrease is persistant, and the cholesterol is not immediately accessible, as suggested by the initial increase of absorbance. It is surprising that the second phase in the kinetics (100-600 ms) corresponds to a cholesterol domain inaccessible to the digitonin. However, we cannot exclude
the possibility that digitonin and/or filipin provokes a redistribution of cholesterol between the two membranes or within one of them. The disappearance of the third phase of the curve (>l s) could correspond to the cholesterol accessible to digitonin, i.e., to the outer membrane. Another intriguing observation is that even at a digitonin/cholesterol molar ratio as low as 0.15 about
IDI6.1=1~g/~g 1
1
1
0.95
a.75
w CJ
0.9
0.9
0.05
0.03 0.95 0.0
1 0.95
‘\i
0
I
0.P b
2
3
‘]:I----0
1
2
3
AFTER SONICATION
BRAIN MITOCHONDRIA
0.0 0
1
2
3
FIG. 2. Kinetics of complex formation between filipin and cholesterol contained in brain mitochondria. Mitochondria are preincubated in absence (upper left) or in presence of digitonin at 1, 5, and 10 Fg/mg protein. Lower right, mitochondria are sonicated before filipin reaction. For comparison, curves are normalized to 1 at the maximal absorbance.
I‘HUL!3S’I’KltUL
UUMAINS
IN
HKAllU
z UY
m
1
2
1
5
00: * 0
2
1
1
[D16*1=10)lq/nq
145
KAI
[DI6.1=1J.!q/11q
I
zm 00: ‘0
ANLJ
3
AFTERSONICATION
0.95
LIVER
0.9
0.95
0.05 -L
MITOCHONDRIA
0.8 0.9 0.75 0.7 0.65 0.05 0.0 i 0
2
1
3
0.b ~ 0 TIME
l IN
S;COND3S
FIG. 3. Kinetics of complex formation between filipin and cholesterol contained in liver mitochondria. Mitochondria are preincubated in are sonicated hefore filipin reaction. absence (upper left) or in presence of digitonin at 1, 5, and 10 rg/mg protein. Lower right, mitochondria For comparison, curves are normalized to 1 at the maximal absorbance.
60% of the filipin-accessible cholesterol was complexed by digitonin, as revealed by the total amplitude of the absorbance decrease. Further increase of digitonin did not induce any substantial change of the kinetics (results not shown). Besides, when mitochondria were sonicated, the total amplitude of absorbance decrease was diminished, suggesting that cholesterol could then be masked by interactions with other components of the membranes. Cholesterol-F&pin Interactions Rat Liver Mitochondria
in
In rat liver mitochondria, the kinetic curve is characterized by a break at 600 ms (Fig. 3) and by a slow phase with a plateau at 20 s (not shown). After sonication, a faster kinetics was observed and the curve could be fitted by the sum of two exponentials. This suggests that the sonication permitted the rapid access of filipin to cholesterol which became homogeneously distributed. When mitochondria were preincubated with digitonin at 1 Kg/ mg protein, an absorbance increase appeared during the first 100 ms. Thus, it is possible to assume that digitonin interacts first with the cholesterol present within the outer leaflet of the outer membrane. The break in the curve at 600 ms is still obvious but the kinetics is slower. Increase of the digitonin concentration to 5 and 10 pg/ mg of protein resulted in a decrease of the total absorbance diminution. After neglecting the first 100 ms, the
curves could be fitted to a sum of two exponentials. Since in liver mitochondria cholesterol content is around 3 wg/ mg of protein (4), the molar ratio of digitonin to cholesterol in these experiments varied from 0.1 to 1.0. At the higher concentration of digitonin, absorbance decrease was reduced to about 20% of the total absorbance decrease observed in the control (without digitonin). The cholesterol unaccessible to filipin, but accessible to digitonin, was t,hen only 20% of the total cholesterol, in contrast with 70% for brain mitochondria. In the sonicated liver mitochondria, all the cholesterol could react with filipin. DISCUSSION
Filipin has been used to probe sterols in the biological membranes mainly by electron microscopy (24,25). It is generally assumed that cholesterol is freely accessible to filipin and that filipin does not cause any reorganization of the cholesterol distribution. Another approach is to study the kinetics of the sterol-filipin complex formation by the stopped-flow method (20,28,29). We analyzed the distribution of cholesterol in intact mitochondria by this approach, where four areas of cholesterol could a priori exist, taking into account the two leaflets of the two membranes. Additional domains, limited by membrane components or complex structures such as contact points could be formed if membrane components interact with sterols or
146
CREMEL
provoke a lateral segregation. Such domains were visualized in N. crassa mitochondria (10, 15) and were suggested by us previously (6, 14) for rat brain mitochondria. Cholesterol Distribution
in Rat Liver Mitochondria
In the liver mitochondria, the kinetics of filipin-cholesterol complex formation could be separated into two distinguishable components. The absorbance changes occuring during the two phases are similar in magnitude and we suppose that they correspond to the two membranes. Our suggestion is based upon the following reasoning: In rat liver mitochondria, the outer membrane contains approximately 40 pg cholesterol/mg protein, whereas the cholesterol content of the inner membrane is 3 pg/mg protein (2). Nevertheless, while 50% of the total mitochondrial proteins are found in the inner membrane, only 5% are found in the outer membrane (30). Thus, the total amount of sterol is 150 pg and 200 pg for 100 mg of mitochondrial protein in the inner and outer membranes, respectively; i.e., there is an approximately equivalent distribution between the two membranes. In fact, sterol is more diluted in the inner membrane. In other words, using only the usual cholesterol/ protein or cholesterol/lipid ratio could induce a distorted view of the cholesterol distribution between the two membranes. Sonication of mitochondria, which destroys the membrane organization, induces a more homogeneous distribution of cholesterol with regard to filipin accessibility, since it is noticed that after sonication the kinetic curve is characterized by a single component and a plateau reached at 2 s. Digitonin effect is more difficult to interpret. Logically, we expected that digitonin initially provoked the disappearance of the first part of the curve which probably corresponded to the more accessible outer membrane. However, at the smallest digitonin concentration used, we observed the disappearance of the slow component. This could be due to one of the following effects: (i) The digitonin molecule which forms a complex with a molecule of cholesterol of the outer membrane might be transferred to another sterol molecule in the inner membrane. This could occur during the incubation time, which is very long compared to the rate of the digitonin-cholesterol complex formation. (ii) Digitonin might provoke a redistribution of the sterol by inducing a flip-flop within the outer membrane followed by a transfer between the two membranes. Nevertheless, independently of the intramitochondrial distribution of cholesterol, our results imply that the great majority of the cholesterol of rat liver mitochondria is accessible for both the digitonin and the filipin. Thus, we propose that cholesterol is homogeneously
ET AL.
distributed branes.
in each of the two mitochondrial
Cholesterol Distribution
mem-
in Rat Brain Mitochondria
In intact rat brain mitochondria, similarly to lipid vesicles or detergent micelles without cholesterol, an initial absorbance increase occured. This can be interpreted by the presence of filipin aggregates in the aqueous buffer, which dissociate in vesicles or micelles (25, 31). Thus, the increase of the absorbance during the first 100 ms in the case of brain mitochondria could be due to a domain apparently devoid of free cholesterol, corresponding probably to the cytoplasmic layer of the outer membrane. Another interesting point is the plateau which appears at 600 ms. This indicates either the presence of a cholesterol domain not directly accessible to filipin or a reduced diffusion of filipin molecules in certain membrane domains. Moreover, a part of cholesterol remains unaccessible both to filipin and to digitonin. Our results suggest the existence of strong interactions between membrane components and sterol as it was already proposed for other membrane systems (3234). As it is very unlikely that the cytoplasmic layer is completely devoid of cholesterol, we suggest that cholesterol in this layer exists in close association with other membrane components. However, our methods do not allow us to distinguish between a domain devoid of cholesterol and a domain where the cholesterol is inaccessible to filipin and digitonin. In contrast to liver mitochondria, sonication of brain mitochondria rendered cholesterol less accessible to filipin, suggesting again that cholesterol participates in interactions with other molecules. In a previous paper (6), we have proposed the existence of three distinct cholesterol domains in the outer membrane of brain mitochondria, one of them is devoid of cholesterol and could correspond to the contact points between the two membranes. The present results are compatible with this model, if we assume that the contact points are either devoid of cholesterol or the cholesterol in this region is embedded in complexes with membrane components. Such a lack of cholesterol has already been described for gap or tight junctions (35), but other evidence has shown that cholesterol was present in these domains but in a complexed form (36). It is important that in brain mitochondria an equilibrium could exist between the complexed and uncomplexed forms of cholesterol. Displacement of this equilibrium could possibly provide a way to modulate the contact point organization and thus presumably modify interactions of the mitochondria with cytoskeletal components (11). ACKNOWLEDGMENTS The authors are grateful Vera Jan&k by a fellowship
to Prof. Guy Vincendon for supporting and to the European Science Foundation
CHOLESTEROL
DOMAINS
IN BRAIN
for a twinning grant (87/44). We are greatly indebted to Drs. C. Staedel, C. Burgun, and P. Hubert for critical discussions and to Dr. S. Felter for her helpful assistance in biological preparations. We also thank Dr. Joseph Bieth and Dr. Bernard Faller (INSERM U237) for help in the stopped-8ow experiments and Ms. C. Thomassin-Orphanides for typing the manuscript. Note added in proof. Very recently De Pinto, Benz, and Palmieri reported in Eur. J. Rio&em. 183,179-187 (1989) the lipid content of purified porin from bovine heart mitochondria. They showed that the active pore-forming complex contained no phospholipids but rather five molecules of cholesterol/polypeptide chain. These data fit well with our suggestion that the mitochondrial-membrane cholesterol interacts with specific membrane components.
REFERENCES 1. Yeagle, P. L. (1985) Biochim.
Biophys. Acta 822,267-287.
2. Daum, G. (1985) Biochim. Biophys. Acta 822, l-42. 3. Comte, J., Maisterrena, B., and Gautheron Biophys. Acta 419,271-284.
D. C. (1976) Biochim.
4. Colbeau, A., Nachbaur, ,J., and Vignais, P. M. (1971) Biochim. Biophys. Acta 249,462%492.
P. ,J., and Basford, R. E. (1969) Biochem-
5. Craven, A. P., Goldblatt, istry 8.3525-3532.
6. Dorbani, L., Jancsik, V., Linden, M., Leterrier, J. F., Nelson, B. D., and Rendon, A. (1987) Arch. Biochem. Biophys. 252,188196. 7. Kramer,
R. (1982) Biochim. Biophys. Acta 693, 296-304.
8. Pitotti, A., Dabbeni-Sala, phys. Acta 600, 79-90.
F., and Bruni,
9. Levy, M., Joncourt, M., and Thiessard, phys. Acta 424,57-65. 10. Freitag, H., Neupert, 123,629-636. 11. Rendon, A., Filliol,
A. (1980) Biochim.
Bio-
J. (1976) Biochim.
Bio-
V. (1987) Biochem. Biophys.
Kes. ('ommun. 149, 776-783. 12. Smith, D. S., Jarlfors, U., and Cameron, Acad. Sci. 253.472-506.
RAT
MITOCHONDRIA
147
14. Jancsik, V., Linden, M., Dorbani, L., Rendon, A., and Nelson, B. D. (1988) Arch. Biochem. Biophys. 264,295-301. 15. Mannella, C. A. (1988) J. Ultrastruc. Mol. Strut. Rex 98, 212216. 16. Johnson, D., and Lardy, H. A. (1967) in Methods in Enzymology (Estabrook, R. W., and Pullman, H. E., Eds.), Vol. 10, pp. 94496, Academic Press, New York. 17. Clark, J. B., and Nicklas, W. J. (1970) J. Riol. Chem. 245, 47244731. 18. Rendon, A., and Masmoudi, A. (1985) J. Neurosci. Methods 14, 41-51. 19. Muscatello, V., and Carafoli, E. (1969) J. Cell Biol. 40, 6022621. 20. Blau, L., and Bittman, R. (1977) Biochemistry 16,4139-4144. 21. De Kruijff, B., and Demel, R. A. (1974) Biochim. Biophys. Acta 339,57-70. 22. Bittman, R. (1978) Lipids 13,686-691. 23. Bolard d. (1986) Biochim. Biophys. Acta 864, 257-304. J., and van Deurs, B. (1984) E’ur. J. 24. Behnke O., Tranum-Jensen, Cell Biol. 35, 1899199. J., and van Deurs, B. (1984) Eur. J. 25. Behnke O., Tranum-Jensen, Cell Biol. 35, 2OOG215. 26. Nishikawa, M., Nojima, S., Akiyama, T., Sankawa, U., and Inoue, K. (1984) J. Biochem. 96,1231-1239. 27. Schnaitman, C., and Greenawalt, J. W. (1968) J. Cell Biol. 38, 158-175. 28. Clejan, S., and Bittman, R. (1985) J. Biol. Chem. 260,2884-2889. 29. Blau, L., and Bittman, R. (1978) J. Biol. Chem. 253,8366-8368. 30. Waksman, A., and Rendon, A. (1974) Biochimie 56,907-924. 31. Milhaud, J., Bolard, J., Benveniste, P., and Hartmann, M. A. (1988) Biochin. Biophys. Acta 943,315-325. 32. Feltkamp, C. A., Van der Waerden, A. W. M. (198’2) Exp. Cell Rex
140,289-297.
W., and Benz, R. (1982) Eur. J. Biochem. D., and Jancsik,
AND
B. F. (1975) Ann. N. Y.
13. Martz, D., Lasek, R. J., Brady, S. T., and Allen, R. D. (1984) Cell Motil. 4,899109.
33. Steer, C. J., Bisher, M., Blumenthal, R., and Steven, A. C. (1984) J. Cell Biol. 99, 315-319. 34. Severs, N. .J., and Simons, H. L. (1983) Nature (London) 303, 637-638. 35. Robenek, M., Jung, W., and Gebhardt, R. (1982) J. Ultrastruct. Res. 78,95-106. 36. Hertzberg, E. I,., and Gilula, N. B. (1979) J. Biol. C’hem. 254, 2138-2147.
OF RIOCHEMISTRY
AND
BIOPHYSICS
Vol. 278, No. 1, April, pp. 142-147, 1990
Cholesterol Distribution in Rat Liver and Brain Mitochondria as Determined by Stopped-Flow Kinetics with Filipin G. Crkmel,
D. Filliol,
Centre de Neurochimie
V. Jancsik,’
and A. Rendon”
du CNRS et U44 de I’INSERM.
Received July 25, 1989, and in revised form November
Strasbourg, France
16,1989
Recently, analysis of protein distribution in rat brain mitochondria suggested the existence of distinct cholesterol domains in the outer membrane (Dorbani et al., 1987, Arch. Biochem. Biophys. 252, 188-196) while such domains were not detected in rat liver mitochondria (Jancsik et al., 1988, Arch. Biochem. Biophys. 264, 295-301). We studied cholesterol distribution in both types of mitochondria by analyzing the kinetics of filipin-cholesterol complex formation, using the stopped-flow technique. In liver mitochondria, the kinetics are characterized by a biphasic curve which presumably corresponds to the two membranes. This was confirmed by the finding that pretreatment with digitonin abolished one of the kinetic components. Sonication of the mitochondria increased the rate of the filipin-cholesterol complex formation and also abolished one of the two components. In the case of brain mitochondria, several distinct cholesterol domains could be revealed: one of them was cholesterol-free and it was directly accessible to filipin. Two other domains were revealed by differences found in the rate of the cholesterol-filipin complex formation. It is noteworthy that only a part of the cholesterol is accessible to filipin. Sonication of mitochondria decreased the proportion of cholesterol molecules accessible to filipin. This suggests specific interactions of cholesterol with other mitowhich occur only in brain chondrial components, mitochondria. cc 1990 Academic Press, Inc.
Cholesterol is a widely distributed logical membranes that is important
constituent of biofor their structure
i Permanent address: Institute of Enzymology, BRC, Hungarian Academy of Sciences, Hungary. ’ To whom correspondence should be addressed at Centre de Neurochimie du CNRS, 5, rue Blaise Pascal, 67084 Strasbourg Cedex, France.
and function. It is unevenly distributed among cellular membranes. Plasma membrane has a cholesterol/phospholipid (C/P)” ratio as high as 0.8-0.9 whereas the intracellular membranes have a ratio of less than 0.2 (1, 2). In heart and liver mitochondria the C/P ratio is 0.04 and 0.1, respectively (3, 4), and cholesterol is localized mainly in the outer membrane. Data obtained with rat liver or heart mitochondria were frequently extrapolated to other mitochondria as well, e.g., to brain mitochondria (5, 6). Only limited information is available on the role of cholesterol in the control and regulation of mitochondrial functions. Dependence of protein function was described for the adenine nucleotide translocator (7), ATPase (8), and ,Bhydroxybutyrate dehydrogenase (9). Strong interaction of ergosterol with porin in Neurospora crassa was also reported (10). We suppose (11) that the preferential localization of cholesterol in the outer membrane could regulate interactions between mitochondria and other cell constituents, e.g., in the case of axonal transport of brain mitochondria (12,13). In a previous paper (6), on the basis of sequential release of marker proteins with increasing concentrations of digitonin, three outer membrane domains have been identified in brain mitochondria, one of them devoid of cholesterol, representing the contact points between outer and inner membrane. In contrast to the liver mitochondria, where porin segregation was not observed (14), in the brain mitochondria a part of the porin-hexokinase complex was located in the contact domain. Recent studies of Mannella (15) also suggested a lateral segregation of porin as a function of cholesterol in mitochondrial membranes of N. crassa. In the present study, we analyze the complex formation between filipin and brain or liver mitochondrial cholesterol by stopped-flow experiments. The significance of the different membrane domains in mitochondria is discussed. ” Abbreviation
used: C/P, cholesterol/phospholipid.
142 All
0003.9861/90 $3.00 Copyright 0 1990 by Academic Press, Inc. rights of reproduction in any form reserved.
CHOLESTEROL
MATERIALS
AND
DOMAINS
IN BRAIN
METHODS
Chemicals. The chemicals used were of the highest grade of purity available. Digitonin was obtained from Merck; filipin, egg phosphatidylcholine (Type XI-E), and cholesterol were from Sigma and were used without further purification. Ficoll and Percoll were from Pharmacia. and mitochondrial subfractionaPreparation of mitochondria tion. Rat liver mitochondria were prepared as previously described (16): nonsynaptic rat brain mitochondria were isolated following two methods, one of them using a discontinuous Ficoll gradient (17) and the other using a Percoll gradient procedure (18). Similar results were obtained with both types of mitochondrial preparations. Mitochondria were fractionated by digitonin, according to Dorbani et al. (6). In brief, aliquots of ice-cold digitonin solution at various concentrations were added to equal volumes of mitochondrial suspensions and the reaction mixtures were continuously stirred for 15 min at 0°C and then diluted with 3 vol of isolated medium. After centrifugation for 20 min at 12,5OOg, pellet was resuspended in isolation medium. In the sonication procedure, mitochondria at 10 mg/ml were sonicated with a MSE 150-type sonifier for 3 X 30 s at 0°C (19). Small unilamrllar vesicles preparation. Mixtures of phosphatidylcholine, cholesterol, and dicetylphosphate (7:2:1) dissolved in chloroform were evaporated to dryness under nitrogen. The lipidic film was rehydrated overnight in a buffer containing 5 mM Hepes, 0.1 mM EDTA, pH 7.6, at 4°C. The suspension was vortexed, then sonicated under nitrogen for 3 X 10 min at a frequency of 20 kHz and a 140.em amplitude with a MSE sonifier. During sonication the vial was kept in a water bath at a temperature of 5°C. The resulting clear dispersion was centrifuged for 120 min at 145,OOOgto eliminate multilayers and titanium particles of the probe. For some experiments, the cholesterol amount was modified and varied from 0 to 20%. Kinetic measurements. Stock solutions of filipin were prepared in dimethylformamide according to Blau and Bittman (20). Filipin concentration was determined by its molar absorbance at 360 nm (E = 4.5 X 104M ‘cm ‘). The formation of filipin-cholesterol complex is characterized by a decrease of absorbance at 357 nm. These changes were monitored after (routhe rapid mixing of 30 pM filipin solutions with mitochondria tinely, 0.5 mg/ml), lipid vesicles, or cholesterol solution in a stoppedflow spectrometer (PQ/SF 53 model, Hi-Tech Scientific, England). A slit width of 2 mm was used. Experiments were carried out at 25 t O.l”C. Data were analyzed on a Hewlett-Packard HP 9000 computer.
RESULTS
Filipin binds selectively to sterols which have a planar ring system, a free 3 /3-hydroxyl group, and an aliphatic side chain. The stoichiometry of the filipin-sterol complex has been found to be approximately one to one mole (21-25). In order to complex all the mitochondrial cholesterol, we used filipin in excess. The reaction of filipin with cholesterol was analyzed by the stopped-flow method. For cholesterol (20 PM) diluted from an ethanolic stock solution in isolation medium or for lipid vesicles containing various amounts of cholesterol, kinetic curves could be fitted by the sum of two exponentials (Fig. 1). It is reasonable to assume that cholesterol is uniformly distributed in solution or in these model vesicles. We can test cholesterol distribution in other membranes, e.g., mitochondria by this approach. The exactness of a fitting by the sum of two exponentials allows
AND
RAT
143
MITOCHONDRIA
for the conclusion that cholesterol is homogeneously distributed in the membrane tested while failing of a similar fitting is a good indication for inhomogeneous cholesterol distribution within the membranes. Furthermore, under the same experimental conditions, analysis of the interactions of filipin with small unilamellar phospholipid vesicles which did not contain cholesterol or with Triton X-100 micelles also without cholesterol showed an initial absorbance increase (data not shown). Cholesterol-Filipin Interaction Rat Brain Mitochondria
in
Cholesterol-filipin interactions in intact brain mitochondria are characterized by at least three phases (Fig. 2). During the first 100 ms, we observed an increase of absorbance. The second phase is discernible until 600 ms. At this time, a shoulder appears and the third phase, which seems to be of an exponential type, is observable at about 1 s. These results suggest the existence of distinct cholesterol domains according to their filipin accessibility. In order to induce a more homogeneous cholesterol distribution, we attempted to destroy the membrane organization by freezing and thawing the mitochondria followed by sonication. As can be seen, the initial increase of absorbance and the shoulder disappeared and it is possible to fit the curve to a sum of two exponentials. These results suggest the existence of at least three domains in the mitochondria, regarding the cholesterol content: one domain is devoid of cholesterol, corresponding to the increase of absorbance; the two other domains correspond to the two phases of the kinetics. These two domains could correspond to the outer and inner membranes. We studied the rate and extent of filipin-cholesterol complex formation in mitochondria preincubated with digitonin. Digitonin is known to complex cholesterol with a 1:l stoichiometry (26) and it is widely used to separate the two mitochondrial membranes (6, 27). It is believed that digitonin provokes the selective disruption of the outer membrane. We used 1 to 10 pg of digitoninl mg mitochondrial protein. On the other hand, by adding digitonin to lipid vesicles containing cholesterol, we found that filipin could not react with cholesterol complexed to digitonin; in this case, absorbance decrease was abolished. At these digitonin concentrations, no release of soluble mitochondrial protein occured (6,14). If we assume a cholesterol/protein ratio of 20 pg/mg proteins (6), the molar digitonin/cholesterol ratio varies from 0.015 to 0.15. At 1 pg/mg (Fig. 1) or 2 pg/mg digitonin (not shown), the pattern of the kinetics is similar to the intact mitochondria. At 5 pg/mg digitonin, the shoulder disappears, the total decrease of absorbance diminishes, and, comparatively, the initial absorbance increase is more pronounced. At 10 yg/mg digitonin, the
144
CRRMEL
ET AL.
TEROL IN SOLUTION
B
0.015 i
0.6
0
1
I
h
‘s”‘~‘~~‘~*~a~““l Orn' TIME IN SECONDS
CHOLES TEROL IN LIPID VESICLES
0
Om2
" '
TIME IN SECONDS
Om2
FIG. 1 Kinetics of complex formation between filipin and cholesterol. Cholesterol is dissolved in aqueous buffer from ethanolic solution or contained in lipid vesicles (phosphatidylcholine:cholesterol:dicetylphosphate ratio was 7:2:1 for the experiment shown here). (+) Experimental points, (-) fitted curve from the sum of two exponentials. Insets, the residual curve is obtained by the differences between the experimental points and the fitted curves. For comparison, curves are normalized to 1 at the maximal absorbance.
latter pattern is accentuated. At all digitonin concentrations tested, the initial rate of absorbance decrease is persistant, and the cholesterol is not immediately accessible, as suggested by the initial increase of absorbance. It is surprising that the second phase in the kinetics (100-600 ms) corresponds to a cholesterol domain inaccessible to the digitonin. However, we cannot exclude
the possibility that digitonin and/or filipin provokes a redistribution of cholesterol between the two membranes or within one of them. The disappearance of the third phase of the curve (>l s) could correspond to the cholesterol accessible to digitonin, i.e., to the outer membrane. Another intriguing observation is that even at a digitonin/cholesterol molar ratio as low as 0.15 about
IDI6.1=1~g/~g 1
1
1
0.95
a.75
w CJ
0.9
0.9
0.05
0.03 0.95 0.0
1 0.95
‘\i
0
I
0.P b
2
3
‘]:I----0
1
2
3
AFTER SONICATION
BRAIN MITOCHONDRIA
0.0 0
1
2
3
FIG. 2. Kinetics of complex formation between filipin and cholesterol contained in brain mitochondria. Mitochondria are preincubated in absence (upper left) or in presence of digitonin at 1, 5, and 10 Fg/mg protein. Lower right, mitochondria are sonicated before filipin reaction. For comparison, curves are normalized to 1 at the maximal absorbance.
I‘HUL!3S’I’KltUL
UUMAINS
IN
HKAllU
z UY
m
1
2
1
5
00: * 0
2
1
1
[D16*1=10)lq/nq
145
KAI
[DI6.1=1J.!q/11q
I
zm 00: ‘0
ANLJ
3
AFTERSONICATION
0.95
LIVER
0.9
0.95
0.05 -L
MITOCHONDRIA
0.8 0.9 0.75 0.7 0.65 0.05 0.0 i 0
2
1
3
0.b ~ 0 TIME
l IN
S;COND3S
FIG. 3. Kinetics of complex formation between filipin and cholesterol contained in liver mitochondria. Mitochondria are preincubated in are sonicated hefore filipin reaction. absence (upper left) or in presence of digitonin at 1, 5, and 10 rg/mg protein. Lower right, mitochondria For comparison, curves are normalized to 1 at the maximal absorbance.
60% of the filipin-accessible cholesterol was complexed by digitonin, as revealed by the total amplitude of the absorbance decrease. Further increase of digitonin did not induce any substantial change of the kinetics (results not shown). Besides, when mitochondria were sonicated, the total amplitude of absorbance decrease was diminished, suggesting that cholesterol could then be masked by interactions with other components of the membranes. Cholesterol-F&pin Interactions Rat Liver Mitochondria
in
In rat liver mitochondria, the kinetic curve is characterized by a break at 600 ms (Fig. 3) and by a slow phase with a plateau at 20 s (not shown). After sonication, a faster kinetics was observed and the curve could be fitted by the sum of two exponentials. This suggests that the sonication permitted the rapid access of filipin to cholesterol which became homogeneously distributed. When mitochondria were preincubated with digitonin at 1 Kg/ mg protein, an absorbance increase appeared during the first 100 ms. Thus, it is possible to assume that digitonin interacts first with the cholesterol present within the outer leaflet of the outer membrane. The break in the curve at 600 ms is still obvious but the kinetics is slower. Increase of the digitonin concentration to 5 and 10 pg/ mg of protein resulted in a decrease of the total absorbance diminution. After neglecting the first 100 ms, the
curves could be fitted to a sum of two exponentials. Since in liver mitochondria cholesterol content is around 3 wg/ mg of protein (4), the molar ratio of digitonin to cholesterol in these experiments varied from 0.1 to 1.0. At the higher concentration of digitonin, absorbance decrease was reduced to about 20% of the total absorbance decrease observed in the control (without digitonin). The cholesterol unaccessible to filipin, but accessible to digitonin, was t,hen only 20% of the total cholesterol, in contrast with 70% for brain mitochondria. In the sonicated liver mitochondria, all the cholesterol could react with filipin. DISCUSSION
Filipin has been used to probe sterols in the biological membranes mainly by electron microscopy (24,25). It is generally assumed that cholesterol is freely accessible to filipin and that filipin does not cause any reorganization of the cholesterol distribution. Another approach is to study the kinetics of the sterol-filipin complex formation by the stopped-flow method (20,28,29). We analyzed the distribution of cholesterol in intact mitochondria by this approach, where four areas of cholesterol could a priori exist, taking into account the two leaflets of the two membranes. Additional domains, limited by membrane components or complex structures such as contact points could be formed if membrane components interact with sterols or
146
CREMEL
provoke a lateral segregation. Such domains were visualized in N. crassa mitochondria (10, 15) and were suggested by us previously (6, 14) for rat brain mitochondria. Cholesterol Distribution
in Rat Liver Mitochondria
In the liver mitochondria, the kinetics of filipin-cholesterol complex formation could be separated into two distinguishable components. The absorbance changes occuring during the two phases are similar in magnitude and we suppose that they correspond to the two membranes. Our suggestion is based upon the following reasoning: In rat liver mitochondria, the outer membrane contains approximately 40 pg cholesterol/mg protein, whereas the cholesterol content of the inner membrane is 3 pg/mg protein (2). Nevertheless, while 50% of the total mitochondrial proteins are found in the inner membrane, only 5% are found in the outer membrane (30). Thus, the total amount of sterol is 150 pg and 200 pg for 100 mg of mitochondrial protein in the inner and outer membranes, respectively; i.e., there is an approximately equivalent distribution between the two membranes. In fact, sterol is more diluted in the inner membrane. In other words, using only the usual cholesterol/ protein or cholesterol/lipid ratio could induce a distorted view of the cholesterol distribution between the two membranes. Sonication of mitochondria, which destroys the membrane organization, induces a more homogeneous distribution of cholesterol with regard to filipin accessibility, since it is noticed that after sonication the kinetic curve is characterized by a single component and a plateau reached at 2 s. Digitonin effect is more difficult to interpret. Logically, we expected that digitonin initially provoked the disappearance of the first part of the curve which probably corresponded to the more accessible outer membrane. However, at the smallest digitonin concentration used, we observed the disappearance of the slow component. This could be due to one of the following effects: (i) The digitonin molecule which forms a complex with a molecule of cholesterol of the outer membrane might be transferred to another sterol molecule in the inner membrane. This could occur during the incubation time, which is very long compared to the rate of the digitonin-cholesterol complex formation. (ii) Digitonin might provoke a redistribution of the sterol by inducing a flip-flop within the outer membrane followed by a transfer between the two membranes. Nevertheless, independently of the intramitochondrial distribution of cholesterol, our results imply that the great majority of the cholesterol of rat liver mitochondria is accessible for both the digitonin and the filipin. Thus, we propose that cholesterol is homogeneously
ET AL.
distributed branes.
in each of the two mitochondrial
Cholesterol Distribution
mem-
in Rat Brain Mitochondria
In intact rat brain mitochondria, similarly to lipid vesicles or detergent micelles without cholesterol, an initial absorbance increase occured. This can be interpreted by the presence of filipin aggregates in the aqueous buffer, which dissociate in vesicles or micelles (25, 31). Thus, the increase of the absorbance during the first 100 ms in the case of brain mitochondria could be due to a domain apparently devoid of free cholesterol, corresponding probably to the cytoplasmic layer of the outer membrane. Another interesting point is the plateau which appears at 600 ms. This indicates either the presence of a cholesterol domain not directly accessible to filipin or a reduced diffusion of filipin molecules in certain membrane domains. Moreover, a part of cholesterol remains unaccessible both to filipin and to digitonin. Our results suggest the existence of strong interactions between membrane components and sterol as it was already proposed for other membrane systems (3234). As it is very unlikely that the cytoplasmic layer is completely devoid of cholesterol, we suggest that cholesterol in this layer exists in close association with other membrane components. However, our methods do not allow us to distinguish between a domain devoid of cholesterol and a domain where the cholesterol is inaccessible to filipin and digitonin. In contrast to liver mitochondria, sonication of brain mitochondria rendered cholesterol less accessible to filipin, suggesting again that cholesterol participates in interactions with other molecules. In a previous paper (6), we have proposed the existence of three distinct cholesterol domains in the outer membrane of brain mitochondria, one of them is devoid of cholesterol and could correspond to the contact points between the two membranes. The present results are compatible with this model, if we assume that the contact points are either devoid of cholesterol or the cholesterol in this region is embedded in complexes with membrane components. Such a lack of cholesterol has already been described for gap or tight junctions (35), but other evidence has shown that cholesterol was present in these domains but in a complexed form (36). It is important that in brain mitochondria an equilibrium could exist between the complexed and uncomplexed forms of cholesterol. Displacement of this equilibrium could possibly provide a way to modulate the contact point organization and thus presumably modify interactions of the mitochondria with cytoskeletal components (11). ACKNOWLEDGMENTS The authors are grateful Vera Jan&k by a fellowship
to Prof. Guy Vincendon for supporting and to the European Science Foundation
CHOLESTEROL
DOMAINS
IN BRAIN
for a twinning grant (87/44). We are greatly indebted to Drs. C. Staedel, C. Burgun, and P. Hubert for critical discussions and to Dr. S. Felter for her helpful assistance in biological preparations. We also thank Dr. Joseph Bieth and Dr. Bernard Faller (INSERM U237) for help in the stopped-8ow experiments and Ms. C. Thomassin-Orphanides for typing the manuscript. Note added in proof. Very recently De Pinto, Benz, and Palmieri reported in Eur. J. Rio&em. 183,179-187 (1989) the lipid content of purified porin from bovine heart mitochondria. They showed that the active pore-forming complex contained no phospholipids but rather five molecules of cholesterol/polypeptide chain. These data fit well with our suggestion that the mitochondrial-membrane cholesterol interacts with specific membrane components.
REFERENCES 1. Yeagle, P. L. (1985) Biochim.
Biophys. Acta 822,267-287.
2. Daum, G. (1985) Biochim. Biophys. Acta 822, l-42. 3. Comte, J., Maisterrena, B., and Gautheron Biophys. Acta 419,271-284.
D. C. (1976) Biochim.
4. Colbeau, A., Nachbaur, ,J., and Vignais, P. M. (1971) Biochim. Biophys. Acta 249,462%492.
P. ,J., and Basford, R. E. (1969) Biochem-
5. Craven, A. P., Goldblatt, istry 8.3525-3532.
6. Dorbani, L., Jancsik, V., Linden, M., Leterrier, J. F., Nelson, B. D., and Rendon, A. (1987) Arch. Biochem. Biophys. 252,188196. 7. Kramer,
R. (1982) Biochim. Biophys. Acta 693, 296-304.
8. Pitotti, A., Dabbeni-Sala, phys. Acta 600, 79-90.
F., and Bruni,
9. Levy, M., Joncourt, M., and Thiessard, phys. Acta 424,57-65. 10. Freitag, H., Neupert, 123,629-636. 11. Rendon, A., Filliol,
A. (1980) Biochim.
Bio-
J. (1976) Biochim.
Bio-
V. (1987) Biochem. Biophys.
Kes. ('ommun. 149, 776-783. 12. Smith, D. S., Jarlfors, U., and Cameron, Acad. Sci. 253.472-506.
RAT
MITOCHONDRIA
147
14. Jancsik, V., Linden, M., Dorbani, L., Rendon, A., and Nelson, B. D. (1988) Arch. Biochem. Biophys. 264,295-301. 15. Mannella, C. A. (1988) J. Ultrastruc. Mol. Strut. Rex 98, 212216. 16. Johnson, D., and Lardy, H. A. (1967) in Methods in Enzymology (Estabrook, R. W., and Pullman, H. E., Eds.), Vol. 10, pp. 94496, Academic Press, New York. 17. Clark, J. B., and Nicklas, W. J. (1970) J. Riol. Chem. 245, 47244731. 18. Rendon, A., and Masmoudi, A. (1985) J. Neurosci. Methods 14, 41-51. 19. Muscatello, V., and Carafoli, E. (1969) J. Cell Biol. 40, 6022621. 20. Blau, L., and Bittman, R. (1977) Biochemistry 16,4139-4144. 21. De Kruijff, B., and Demel, R. A. (1974) Biochim. Biophys. Acta 339,57-70. 22. Bittman, R. (1978) Lipids 13,686-691. 23. Bolard d. (1986) Biochim. Biophys. Acta 864, 257-304. J., and van Deurs, B. (1984) E’ur. J. 24. Behnke O., Tranum-Jensen, Cell Biol. 35, 1899199. J., and van Deurs, B. (1984) Eur. J. 25. Behnke O., Tranum-Jensen, Cell Biol. 35, 2OOG215. 26. Nishikawa, M., Nojima, S., Akiyama, T., Sankawa, U., and Inoue, K. (1984) J. Biochem. 96,1231-1239. 27. Schnaitman, C., and Greenawalt, J. W. (1968) J. Cell Biol. 38, 158-175. 28. Clejan, S., and Bittman, R. (1985) J. Biol. Chem. 260,2884-2889. 29. Blau, L., and Bittman, R. (1978) J. Biol. Chem. 253,8366-8368. 30. Waksman, A., and Rendon, A. (1974) Biochimie 56,907-924. 31. Milhaud, J., Bolard, J., Benveniste, P., and Hartmann, M. A. (1988) Biochin. Biophys. Acta 943,315-325. 32. Feltkamp, C. A., Van der Waerden, A. W. M. (198’2) Exp. Cell Rex
140,289-297.
W., and Benz, R. (1982) Eur. J. Biochem. D., and Jancsik,
AND
B. F. (1975) Ann. N. Y.
13. Martz, D., Lasek, R. J., Brady, S. T., and Allen, R. D. (1984) Cell Motil. 4,899109.
33. Steer, C. J., Bisher, M., Blumenthal, R., and Steven, A. C. (1984) J. Cell Biol. 99, 315-319. 34. Severs, N. .J., and Simons, H. L. (1983) Nature (London) 303, 637-638. 35. Robenek, M., Jung, W., and Gebhardt, R. (1982) J. Ultrastruct. Res. 78,95-106. 36. Hertzberg, E. I,., and Gilula, N. B. (1979) J. Biol. C’hem. 254, 2138-2147.
