Proc. Natl. Acad. Sci. USA Vol. 76, No. 3, pp. 1236-1240, March 1979
Cell Biology
Lateral translational diffusion of cytochrome c oxidase in the mitochondrial energy-transducing membrane (membrane structure/protein diffusion/electron transfer/immunoglobulin probes/freeze-fracture)
MATTHIAS HOCHLI AND CHARLES R. HACKENBROCK Laboratories for Cell Biology, Department of Anatomy, School of Medicine, University of North Carolina at Chapel Hill, Chapel Hill, North Carolina 27514
Communicated by Albert L. Lehninger, November 17, 1978
ABSTRACT The degree of freedom for lateral translational diffusion by cytochrome c oxidase and other integral proteins in the energ-transducing membrane of the mitochondrion was determined by combining the use of an immunoglobulin probe monospecific for the oxidase with thermotropic lipid phase transitions. Lateral mobility of the oxidase was monitored by observing the distribution of the immunoglobulin probe on the membrane surface by deep-etch electron microscopy and by observing the distribution of intramembrane particles (integral proteins) in the hydrophobic interior of the membrane by freeze-fracture electron microscopy. Incubation of the membrane with the immunoglobulin resulted in a time-dependent clustering of predominantly large intramembrane particles. Low temperature-induced lipid phase transitions resulted in the close packing of all intramembrane particles and cytochrome c oxidase by lateral exclusion from domains of gel-state bilayer lipid and was completely reversible. However, when cytochrome c oxidase was crosslinked through an immunoglobulin lattice prior to returning the membrane to above the lipid phase transition temperature, small intramembrane particles rerandomized while the large oxidase-related particles remained clustered. These observations reveal that cytochrome c oxidase can diffuse laterally in the energy-transducing membrane, either independently of all other integral proteins or in physical union with one or more other integral proteins. In addition, many other as yet unidentified smaller integral proteins can diffuse independently of the oxidase.
Whether or not proteins integral to the mitochondrial energy-transducing membrane can undergo two-dimensional lateral diffusion independently of one another relates directly to the molecular events implicit in the mechanisms of electron transfer and energy coupling in oxidative phosphorylation (1). Electron transfer between the respiratory proteins as well as energy coupling between these proteins and the oligomycinsensitive ATPase may occur through direct physical contact, either through a relatively permanent protein-protein complex or by alternating protein-protein collisions. Such direct physical contact is required in coupling mechanisms in which the energy derived from electron transfer is transduced through covalent high-energy bonds or through energized protein conformational states to generate the synthesis of ATP (2, 3). Alternatively, no physical contact, permanent or alternating, is required in coupling mechanisms in which the energy derived from electron transfer is transduced through a proton gradient across or in the mitochondrial membrane to generate the synthesis of ATP (4, 5). Cytochrome c oxidase is one of 30 or so proteins occurring in the energy-transducing membrane and participates in electron-transfer and energy-coupling processes. The oxidase is a completely transmembranous integral protein (6). In the present communication we report on the application of a
metabolic and visual immunoglobulin probe monospecific for cytochrome c oxidase combined with the use of thermotropic lipid phase transitions to examine the degree of freedom for lateral translational diffusion by cytochrome c oxidase and other integral proteins in the plane of the mitochondrial energytransducing membrane. MATERIALS AND METHODS Liver mitochondria were isolated from male Sprague-Dawley rats in an isolation medium containing 70 mM sucrose, 220 mM mannitol, 2 mM Hepes, 0.5 mg of bovine serum albumin per ml, and KOH to pH 7.4 (medium is designated H medium). Subsequent removal of the outer membrane and purification of the inner membrane/matrix (mitoplast) fraction was carried out by use of a controlled digitonin incubation (7, 8). The complex inner membrane was converted to a simple spherical, functional membrane by washing and resuspending in a 1:7.5 diluted (40 mosM) albumin-free H medium as described (7, 9) (medium is designated H40 medium). Preparation and analysis of affinity-purified rabbit immunoglobulin monospecific for cytochrome c oxidase, affinity-purified goat anti-rabbit immunoglobulin, and ferritin-conjugated immunoglobulin were as described (6). Inhibition of electron transfer by specific immunoglobulin was monitored with a Clark oxygen electrode (6). Protein determination was by a biuret method (10). Low temperature-induced lipid phase transitions were produced in the mitochondrial membrane as outlined (1, 11). For freeze-fracture, membranes in 30% (wt/vol) glycerol in H40 medium were centrifuged into micropellets and subsequently equilibrated at 25 or -10° C prior to rapid freezing in Freon 22 precooled by liquid nitrogen. For reversibility of the low temperature-induced lipid phase transition, membranes equilibrated at -100C were returned quickly to 250C and then rapidly frozen. For deep etching, after equilibration at 25 or -10°C, membranes were fixed overnight with 1.5% glutaraldehyde in 25% glycerol. The fixed membranes were then washed free of glycerol at 0C with 1:200 diluted albumin-free H medium. Deep etching, freeze-fracturing, and platinum/ carbon replication were carried out at -1000C at 2 X 10-6 torr in a Balzers BA360 freeze-etching apparatus equipped with electron guns. Electron micrographs were taken with a JOEL 100CX electron microscope operated at 80 kV. RESULTS Free Lateral Diffusion of Cytochrome c Oxidase in the
Membrane Lipid Bilayer. Mitochondrial energy-transducing membranes frozen rapidly from 250C revealed intramembrane particles in a dispersed, random distribution (Fig. la). The smallest intramembrane particles measured between 4 and 5 nm in diameter, and 70% of the largest particles measured between 10 and 18 nm in diameter (12, 13). When affinitypurified rabbit immunoglobulin monospecific for cytochrome
The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertisement" in accordance with 18 U. S. C. §1734 solely to indicate this fact.
1236
Cell Biology: H6chli and Hackenbrock
Proc. Natl. Acad. Sci. USA 76 (1979)
1237
IMM
T
70 nmol °2
In
lgG
1 min
FIG. 2. Oxygen electrode traces of effects of rabbit immunoglobulin monospecific for cytochrome c oxidase and rabbit normal immunoglobulin on electron transfer in the energy transducing membrane. Reaction system: 9.3 mM sucrose, 29.3 mM mannitol, 0.26 mM Hepes, 0.33 mM potassium phosphate, 0.13 mM EDTA, 0.66 mM succinate, pH adjusted to 7.4 with KOH, 2 mg of inner membrane matrix protein (IMM). Temperature 250C; volume, 2 ml.
inhibition (Fig. lb) and progressed with increasing incubation time (Fig. lc). The smaller intramembrane particles appeared to remain randomly dispersed. The washing and preparation methods prior to freezing did not permit sampling at times shorter than 20 min. Control nonspecific rabbit immunoglobulin neither inhibited electron transfer (Fig. 2) nor aggregated intramembrane particles. These observations suggest that the divalent immunoglobulin not only binds to and inhibits the oxidase but also crosslinks those oxidases that randomly approach each other through free lateral translational diffusion in the plane of the membrane, resulting in a progressive increase in oxidase clustering (Fig. 3). Thermotropic Lateral Mobility of Cytochrome c Oxidase on the Membrane Surface. Membranes were cooled slowly to -10 C-i.e., to well below the onset temperature of the liquid crystalline-to-gel state lipid phase transition (1)-and then were rapidly frozen. The freeze-fractured membranes clearly revealed the development of large domains of highly ordered gel-state bilayer lipid that exclude integral proteins laterally (Fig. 4). That cytochrome c oxidase was excluded laterally could be clearly demonstrated. Rabbit affinity-purified immunoglobulin monospecific for the oxidase was added to membranes
FIG. 1. Freeze-fractured energy-transducing membrane rapidly frozen from 25°C (X67,500.) (a) Prior to the addition of immunoglobulin monospecific for cytochrome c oxidase, the distribution of large and small intramembrane particles is random. (b) After 20-min incubation with the immunoglobulin, small clusters of large intramembrane particles can be distinguished from regions free of large particles. (c) After 4 hr incubation with the immunoglobulin, large intramembrane particles are aggregated into large clusters resulting in distinct regions containing only small particles.
c oxidase was added to the mitochondrial membrane, an abrupt
and complete inhibition of electron transfer occurred (Fig. 2). After inhibition by specific immunoglobulin, freeze-fractured membranes revealed a nonrandom distribution of intramembrane particles characterized by a general clustering of the larger particles. Such clustering was apparent after 20 min of
FIG. 3. Result of immunoglobulin binding to cytochrome c oxidase. (A) Rabbit immunoglobulin monospecific for cytochrome c oxidase (R cyt ox IgG) binds to the oxidase (cyt ox) of the intact membrane at 250C. (B) During incubation at 25°C, divalent immunoglobulin crosslinks those oxidases that approach each other through free lateral diffusion, resulting in oxidase clusters. IP, unidentified integral proteins; LPL, liquid crystalline phospholipid.
1238
Cell Biology: Hochli and Hackenbrock
Proc. Natl. Acad. Sci. USA 76 (1979)
'K
. .
FIG. 4. Freeze-fractured energy-transducing membrane rapidly frozen from -100C. G, gel-state lipid domains. (X67,500.)
above the lipid phase transition temperature as well as to membranes below the lipid phase transition temperature. In both cases this was followed by latticing the rabbit immunoglobulin that was bound to the oxidase on the membrane surface by the further addition of a ferritin-conjugated affinity-purified goat anti-rabbit immunoglobulin. Visualization of this latticed immunoglobulin-ferritin probe was used to approximate the location of the oxidase on the membrane surface. Deep etching of the true surface of the membrane frozen from above the transition temperature revealed a random although slightly clustered distribution of the immunoglobulin-ferritin probe (Fig. 5a). Below the transition temperature the conjugate probe had a nonrandom distribution, appearing to be crowded between large, smooth membrane surface domains free of the oxidase (Fig. 5b). Freeze-fractured membranes revealed that the latticed immunoglobulin-ferritin probe was bound to the membrane surface over domains rich in intramembrane particles but not over domains of gel-state lipid free of intramembrane particles (Fig. 6). Lateral Mobility of Integral Proteins Independent of Cytochrome c Oxidase. The low temperature-induced lateral separations between domains of gel-state bilayer lipid and integral proteins (Fig. 4) were completely reversible. When membranes were returned to above their transition temperature
.-
FIG. 6. Freeze-fractured energy-transducing membrane, showing distribution of ferritin-conjugated immunoglobulin on the membrane surface (arrows) in register with intramembrane particles. Equilibration of membranes at -100C was followed by addition of rabbit immunoglobulin monospecific for cytochrome c oxidase and then ferritin-conjugated goat anti-rabbit immunoglobulin (as in Fig. 5b). G, gel-state lipid domains. (X82,500.)
for a few seconds, the intramembrane particles were observed to be completely and randomly dispersed (Fig. 7). This result demonstrates that the integral proteins, including the cytochrome c oxidase-related large particle, can diffuse and rerandomize rapidly in the plane of the membrane during melting-i.e., during the gel state-to-liquid crystalline lipid phase transition. Such reversible thermotropic transitions are not disruptive to the structure or normal metabolic functions of the mitochondrial membrane (14). Thermotropic lipid phase transitions were used to induce lateral diffusion of integral proteins in the plane of the membrane, and immunoglobulin latticing was used to prevent the free lateral diffusion of cytochrome c oxidase. Rabbit immunoglobulin monospecific for the -oxidase and ferritin-conjugated goat anti-rabbit immunoglobulin were added in sequence to membranes below the lipid phase transition temperature (Figs. 5b and 6). After such immunoglobulin latticing of the oxidase on the membrane surface, freeze-fracture revealed that lateral diffusion and rerandomization of the large intramembrane particles did not occur when the membranes were returned to well above their lipid phase transition temperature (Fig. 8). However, under such latticing conditions specific for cytochrome c oxidase, small intramembrane particles (4.4-7.4 nm in diameter) were not inhib-
.TAeX * .iS-Ar..................b
FIG. 5. Deep-etched surface of energy-transducing membrane, showing general distribution of cytochrome c oxidase revealed by ferritin-
conjugated immunoglobulin. Rabbit immunoglobulin monospecific for the oxidase was added to membranes followed by ferritin-conjugated goat anti-rabbit immunoglobulin. (X132,000.) (a) Membranes equilibrated at 250C followed by addition of immunoglobulins. (b) Membranes equilibrated at -10C followed by addition of immunoglobulins. G, gel-state lipid domains.
Cell Biology: H6chli and Hackenbrock
Proc. Natl. Acad. Sci. USA 76 (1979)
1239
"I V
I Cool
to
%-WII"to
r7Q
FIG. 7. Freeze-fractured energy-transducing membrane equilibrated at -100C, returned to 250C, and then rapidly frozen. Intramembrane particles are random. (X67,500.)
At
-nInor IL
ox IgG 10Add RF-Gcytanti-R
A1C4Add
ited from rerandomizing when membranes were returned to above the transition temperature. In control experiments, treatment with nonspecific rabbit immunoglobulin followed by ferritin-conjugated goat anti-rabbit immunoglobulin had no effect on the lateral diffusion of intramembrane particles of any size. These results demonsf rate that lateral diffusion of cytochrome c oxidase in the membrane bilayer lipid can be inhibited by immunoglobulin latticing of the oxidase on the membrane surface and that many smaller, as yet unidentified, integral proteins can diffuse laterally in the hydrophobic interior of the mitochondrial membrane independently of the oxidase (Fig. 9).
IgG
G anti-R IgG
GPL
LPL
-10 C
0B
1
ox
I
Warm to 25 C
DISCUSSION
The combined use of an immunoglobulin probe monospecific for cytochrome c oxidase and thermotropic lipid phase transitions reported here reveals that cytochrome c oxidase can be observed, by freeze-fracture electron microscopy, to be related to large intramembrane particles in the hydrophobic interior of the highly complex energy-transducing membrane of the mitochondrion. From monitoring the distribution of the monospecific immunoglobulin on the membrane surface by deep-etch electron microscopy, and the distribution of large
anti-R IgG
F
\ \ LPL
R
250C
FIG. 9. Result of combining thermotropic lipid phase transitions with immunoglobulin latticing of cytochrome c oxidase. (A) Membrane above the lipid phase transition at 250C. (B) Membrane below ;.S .... Ha .... * t --+W @-> monospecific for cytochrome c oxidase (R cyt ox IgG) binds to the n sat~-. Ir f-i.ra \ oxidase (cyt ox) at -100C. Ferritin-conjugated goat anti-rabbit imI' ... *-Ai l (F-G anti-RIgG) binds to the immunoglobulin momunoglobulin ~. -i-. -.1~4~ .~" ospecific for the oxidase, resulting in a cross-atcnofm bre zoidases. (D) Membrane temperature raised above the lipid phase -. transition to 250C. Lateral diffusion of crosslinked oxidases does not toccur but other unidentified, small, integral proteins (IP) diffuse POW!,"-laterally, independently of the oxidase. LPL, liquid crystalline .~~.~N ~ phospholipid; GPL, gel-state phospholipid.
-
a
FIG. 8. Freeze-fractured energy-transducing membrane equilibrated at -100C, followed by addition of rabbit immunoglobulin monospecific for cytochrome c oxidase followed by addition of goat antirabbit immunoglobulin (as in Fig. 6) and finally warmed to 25C before rapid freezing. Large intramembrane particles are clustered but small intramembrane particles are random (cf. Fig. 7).
(X67,500-)
intramembrane particles in the hydrophobic interior of the membrane by freeze-fracture electron microscopy, it is apparent that cytochrome c oxidase can diffuse laterally in the
offteenergy-transducing membrane. Whether the idase diffuses entirely independently of all other integral proteins in the membrane or diffuses in physical union with one or more integral proteins is not known at present. In either case, ane
ox-
1240
Cell Biology: Hochli and Hackenbrock
our use of immunoglobulin latticing to prevent lateral diffusion of cytochrome c oxidase clearly demonstrates that smaller, as yet unidentified, integral proteins can diffuse laterally independently of the oxidase. The high potential for free lateral translational diffusion by cytochrome c oxidase and other unidentified, smaller integral proteins in the mitochondrial membrane reported here is consistent with compositional and metabolic characteristics indicative of a highly fluid mitochondrial membrane. Although the mass of the energy-transducing membrane is approximately 75% protein (15), only 50% of the protein is integral to the membrane (16, 17). Thus, although an unusually large number of specific metabolically active proteins are associated with the mitochondrial membrane compared to most other membranes of eukaryotic cells, the total protein integral to this membrane is not significantly different from that of most other cell membranes. Our results indicate that the lipid bilayer of the mitochondrial membrane is considerably less crowded with protein than recognized heretofore and that the lateral space requirement for protein diffusion in this membrane is unrestrictive. The unique composition of bilayer lipid in the mitochondrial membrane offers potentially less resistance to the lateral mobility of integral proteins compared to any other membrane of eukaryotic cells. The high percentage of unsaturated phospholipids together with the virtual lack of cholesterol (15) most likely accounts for the relatively low onset temperature of the lipid crystalline-to-gel state lipid phase transition (1), the exceptionally high mobility in the phosphilipid hydrocarbon chains (18), and the intrinsically low microviscosities (in the range of 0.1-0.9 P) reported for the mitochondrial lipid component (19, 20). Because the integral proteins of the energy-transducing membrane do not appear to be latticed or anchored through strong molecular interactions (11), the potential for lateral diffusion by such proteins should be approximately proportional to the degree of fluidity of the lipid bilayer, which in this membrane is quite high. Indeed, in the cholesterol-containing, higher-viscosity, lower-fluidity plasma membranes, lateral diffusion coefficients for integral proteins are as high as 1-5 X 10-9 cm2_sec-1 at 20'C (21-23). These coefficients are equal to an approximate lateral linear displacement of 40 nm in 1 msec. Thus, the potential rate for free lateral translational diffusion of integral proteins in the cholesterol-free, low-viscosity, high-fluidity lipid bilayer of the mitochondrial energy-transducing membrane is well within the time structure of the known rates of electron transfer and energy transduction at physiological temperature. Whether or not cytochrome c oxidase undergoes independent diffusion or diffuses in physical union with the oligomycinsensitive ATPase, its reductant cytochrome c, or indeed, with the remainder of the respiratory chain through cytochrome cl
Proc. Natl. Acad. Sci. USA 76 (1979)
has not been determined. Available data suggest that cytochrome c reacts alternately through its heme group with the hemes of cytochrome cl and cytochrome c oxidase through a weak complementary charge-mediated interaction (24). Such alternate interaction may require two-dimensional lateral diffusion of cytochrome c on the surface of the membrane or diffusion of cytochrome cl and cytochrome c oxidase in the lipid bilayer of the membrane or both. This investigation was supported by Research Grants BMS75-02372 and PCM77-20689 from The National Science Foundation to C.R.H. and research fellowships to M.H. from the Swiss National Foundation and The Muscular Distrophy Association of America. 1. Hackenbrock, C. R., Hochli, M. & Chau, R. M. (1976) Biochim. Biophys. Acta 455,466-484. 2. Slater, E. C. (1953) Nature (London) 172, 975-978. 3. Boyer, P. D. (1965) in Oxidases and Related Redox Systems, eds. King, T. E., Mason, S. & Morrison, M. (Wiley, New York), Vol. 2, pp. 994-1008. 4. Mitchell, P. (1961) Nature (London) 191, 144-148. 5. Williams, R. J. P. (1961) J. Theor. Biol. 1, 1-17. 6. Hackenbrock, C. R. & Miller Hammon, K. (1975) J. Biol. Chem. 250,9185-9197. 7. Hackenbrock, C. R. (1973) J. Cell. Biol. 53, 450-465. 8. Schnaitman, C. & Greenawalt, J. W. (1968) J. Cell Biol. 38, 158-175. 9. Lemasters, J. J. & Hackenbrock, C. R. (1973) Fed. Proc. Fed. Am. Soc. Exp. Biol. 32,516. 10. Layne, E. (1957) Methods Enzymol. 3,447-454. 11. Hochli, M. & Hackenbrock, C. R. (1977) J. Cell Biol. 72,278291. 12. Hackenbrock, C. R. (1972) Ann. N. Y. Acad. Sci. 195, 492505. 13. Hackenbrock, C. R. (1973) in Mechanisms in Bioenergetics, eds. Azzone, G. F., Ernster, L., Papa, S., Quagliariello, E. & Siliprandi, N. (Academic, New York), pp. 77-88. 14. Hochli, M. & Hackenbrock, C. R. (1976) Proc. Natl. Acad. Sci. USA 73, 1636-1640. 15. Colbeau, A., Nachbaur, J. & Vignals, P. M. (1971) Biochim. Biophys. Acta 249,262-492. 16. Capaldi, R. A. & Tan, P. F. (1974) Fed. Proc. Fed Am. Soc. Exp. Biol. 33, 1515. 17. Harmon, H. J., Hall, J. D. & Crane, F. L. (1974) Biochim. Biophys. Acta 344, 119-155. 18. Keough, K. M., Oldfield, E., Chapman, D. & Beynon, P. (1973) Chem. Phys. Lipids 10, 37-50. 19. Keith, A., Bulfield, G. & Snipes, W. (1970) Biophys. J. 10, 618-629. 20. Feinstein, M. B., Fernandez, S. M. & Sha'afi, R. I. (1975) Biochim. Biophys. Acta 413,354-370. 21. Edidin, M. & Fambrough, D. (1973) J. Cell Biol. 57,27-37. 22. Liebman, P. A. & Entine, G. (1974) Science 185,457-459. 23. Poo, M-M. & Cone, R. A. (1974) Nature (London) 247, 438441. 24. Salemme, F. R. (1977) Annu. Rev. Biochem. 46,299-329.
Cell Biology
Lateral translational diffusion of cytochrome c oxidase in the mitochondrial energy-transducing membrane (membrane structure/protein diffusion/electron transfer/immunoglobulin probes/freeze-fracture)
MATTHIAS HOCHLI AND CHARLES R. HACKENBROCK Laboratories for Cell Biology, Department of Anatomy, School of Medicine, University of North Carolina at Chapel Hill, Chapel Hill, North Carolina 27514
Communicated by Albert L. Lehninger, November 17, 1978
ABSTRACT The degree of freedom for lateral translational diffusion by cytochrome c oxidase and other integral proteins in the energ-transducing membrane of the mitochondrion was determined by combining the use of an immunoglobulin probe monospecific for the oxidase with thermotropic lipid phase transitions. Lateral mobility of the oxidase was monitored by observing the distribution of the immunoglobulin probe on the membrane surface by deep-etch electron microscopy and by observing the distribution of intramembrane particles (integral proteins) in the hydrophobic interior of the membrane by freeze-fracture electron microscopy. Incubation of the membrane with the immunoglobulin resulted in a time-dependent clustering of predominantly large intramembrane particles. Low temperature-induced lipid phase transitions resulted in the close packing of all intramembrane particles and cytochrome c oxidase by lateral exclusion from domains of gel-state bilayer lipid and was completely reversible. However, when cytochrome c oxidase was crosslinked through an immunoglobulin lattice prior to returning the membrane to above the lipid phase transition temperature, small intramembrane particles rerandomized while the large oxidase-related particles remained clustered. These observations reveal that cytochrome c oxidase can diffuse laterally in the energy-transducing membrane, either independently of all other integral proteins or in physical union with one or more other integral proteins. In addition, many other as yet unidentified smaller integral proteins can diffuse independently of the oxidase.
Whether or not proteins integral to the mitochondrial energy-transducing membrane can undergo two-dimensional lateral diffusion independently of one another relates directly to the molecular events implicit in the mechanisms of electron transfer and energy coupling in oxidative phosphorylation (1). Electron transfer between the respiratory proteins as well as energy coupling between these proteins and the oligomycinsensitive ATPase may occur through direct physical contact, either through a relatively permanent protein-protein complex or by alternating protein-protein collisions. Such direct physical contact is required in coupling mechanisms in which the energy derived from electron transfer is transduced through covalent high-energy bonds or through energized protein conformational states to generate the synthesis of ATP (2, 3). Alternatively, no physical contact, permanent or alternating, is required in coupling mechanisms in which the energy derived from electron transfer is transduced through a proton gradient across or in the mitochondrial membrane to generate the synthesis of ATP (4, 5). Cytochrome c oxidase is one of 30 or so proteins occurring in the energy-transducing membrane and participates in electron-transfer and energy-coupling processes. The oxidase is a completely transmembranous integral protein (6). In the present communication we report on the application of a
metabolic and visual immunoglobulin probe monospecific for cytochrome c oxidase combined with the use of thermotropic lipid phase transitions to examine the degree of freedom for lateral translational diffusion by cytochrome c oxidase and other integral proteins in the plane of the mitochondrial energytransducing membrane. MATERIALS AND METHODS Liver mitochondria were isolated from male Sprague-Dawley rats in an isolation medium containing 70 mM sucrose, 220 mM mannitol, 2 mM Hepes, 0.5 mg of bovine serum albumin per ml, and KOH to pH 7.4 (medium is designated H medium). Subsequent removal of the outer membrane and purification of the inner membrane/matrix (mitoplast) fraction was carried out by use of a controlled digitonin incubation (7, 8). The complex inner membrane was converted to a simple spherical, functional membrane by washing and resuspending in a 1:7.5 diluted (40 mosM) albumin-free H medium as described (7, 9) (medium is designated H40 medium). Preparation and analysis of affinity-purified rabbit immunoglobulin monospecific for cytochrome c oxidase, affinity-purified goat anti-rabbit immunoglobulin, and ferritin-conjugated immunoglobulin were as described (6). Inhibition of electron transfer by specific immunoglobulin was monitored with a Clark oxygen electrode (6). Protein determination was by a biuret method (10). Low temperature-induced lipid phase transitions were produced in the mitochondrial membrane as outlined (1, 11). For freeze-fracture, membranes in 30% (wt/vol) glycerol in H40 medium were centrifuged into micropellets and subsequently equilibrated at 25 or -10° C prior to rapid freezing in Freon 22 precooled by liquid nitrogen. For reversibility of the low temperature-induced lipid phase transition, membranes equilibrated at -100C were returned quickly to 250C and then rapidly frozen. For deep etching, after equilibration at 25 or -10°C, membranes were fixed overnight with 1.5% glutaraldehyde in 25% glycerol. The fixed membranes were then washed free of glycerol at 0C with 1:200 diluted albumin-free H medium. Deep etching, freeze-fracturing, and platinum/ carbon replication were carried out at -1000C at 2 X 10-6 torr in a Balzers BA360 freeze-etching apparatus equipped with electron guns. Electron micrographs were taken with a JOEL 100CX electron microscope operated at 80 kV. RESULTS Free Lateral Diffusion of Cytochrome c Oxidase in the
Membrane Lipid Bilayer. Mitochondrial energy-transducing membranes frozen rapidly from 250C revealed intramembrane particles in a dispersed, random distribution (Fig. la). The smallest intramembrane particles measured between 4 and 5 nm in diameter, and 70% of the largest particles measured between 10 and 18 nm in diameter (12, 13). When affinitypurified rabbit immunoglobulin monospecific for cytochrome
The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertisement" in accordance with 18 U. S. C. §1734 solely to indicate this fact.
1236
Cell Biology: H6chli and Hackenbrock
Proc. Natl. Acad. Sci. USA 76 (1979)
1237
IMM
T
70 nmol °2
In
lgG
1 min
FIG. 2. Oxygen electrode traces of effects of rabbit immunoglobulin monospecific for cytochrome c oxidase and rabbit normal immunoglobulin on electron transfer in the energy transducing membrane. Reaction system: 9.3 mM sucrose, 29.3 mM mannitol, 0.26 mM Hepes, 0.33 mM potassium phosphate, 0.13 mM EDTA, 0.66 mM succinate, pH adjusted to 7.4 with KOH, 2 mg of inner membrane matrix protein (IMM). Temperature 250C; volume, 2 ml.
inhibition (Fig. lb) and progressed with increasing incubation time (Fig. lc). The smaller intramembrane particles appeared to remain randomly dispersed. The washing and preparation methods prior to freezing did not permit sampling at times shorter than 20 min. Control nonspecific rabbit immunoglobulin neither inhibited electron transfer (Fig. 2) nor aggregated intramembrane particles. These observations suggest that the divalent immunoglobulin not only binds to and inhibits the oxidase but also crosslinks those oxidases that randomly approach each other through free lateral translational diffusion in the plane of the membrane, resulting in a progressive increase in oxidase clustering (Fig. 3). Thermotropic Lateral Mobility of Cytochrome c Oxidase on the Membrane Surface. Membranes were cooled slowly to -10 C-i.e., to well below the onset temperature of the liquid crystalline-to-gel state lipid phase transition (1)-and then were rapidly frozen. The freeze-fractured membranes clearly revealed the development of large domains of highly ordered gel-state bilayer lipid that exclude integral proteins laterally (Fig. 4). That cytochrome c oxidase was excluded laterally could be clearly demonstrated. Rabbit affinity-purified immunoglobulin monospecific for the oxidase was added to membranes
FIG. 1. Freeze-fractured energy-transducing membrane rapidly frozen from 25°C (X67,500.) (a) Prior to the addition of immunoglobulin monospecific for cytochrome c oxidase, the distribution of large and small intramembrane particles is random. (b) After 20-min incubation with the immunoglobulin, small clusters of large intramembrane particles can be distinguished from regions free of large particles. (c) After 4 hr incubation with the immunoglobulin, large intramembrane particles are aggregated into large clusters resulting in distinct regions containing only small particles.
c oxidase was added to the mitochondrial membrane, an abrupt
and complete inhibition of electron transfer occurred (Fig. 2). After inhibition by specific immunoglobulin, freeze-fractured membranes revealed a nonrandom distribution of intramembrane particles characterized by a general clustering of the larger particles. Such clustering was apparent after 20 min of
FIG. 3. Result of immunoglobulin binding to cytochrome c oxidase. (A) Rabbit immunoglobulin monospecific for cytochrome c oxidase (R cyt ox IgG) binds to the oxidase (cyt ox) of the intact membrane at 250C. (B) During incubation at 25°C, divalent immunoglobulin crosslinks those oxidases that approach each other through free lateral diffusion, resulting in oxidase clusters. IP, unidentified integral proteins; LPL, liquid crystalline phospholipid.
1238
Cell Biology: Hochli and Hackenbrock
Proc. Natl. Acad. Sci. USA 76 (1979)
'K
. .
FIG. 4. Freeze-fractured energy-transducing membrane rapidly frozen from -100C. G, gel-state lipid domains. (X67,500.)
above the lipid phase transition temperature as well as to membranes below the lipid phase transition temperature. In both cases this was followed by latticing the rabbit immunoglobulin that was bound to the oxidase on the membrane surface by the further addition of a ferritin-conjugated affinity-purified goat anti-rabbit immunoglobulin. Visualization of this latticed immunoglobulin-ferritin probe was used to approximate the location of the oxidase on the membrane surface. Deep etching of the true surface of the membrane frozen from above the transition temperature revealed a random although slightly clustered distribution of the immunoglobulin-ferritin probe (Fig. 5a). Below the transition temperature the conjugate probe had a nonrandom distribution, appearing to be crowded between large, smooth membrane surface domains free of the oxidase (Fig. 5b). Freeze-fractured membranes revealed that the latticed immunoglobulin-ferritin probe was bound to the membrane surface over domains rich in intramembrane particles but not over domains of gel-state lipid free of intramembrane particles (Fig. 6). Lateral Mobility of Integral Proteins Independent of Cytochrome c Oxidase. The low temperature-induced lateral separations between domains of gel-state bilayer lipid and integral proteins (Fig. 4) were completely reversible. When membranes were returned to above their transition temperature
.-
FIG. 6. Freeze-fractured energy-transducing membrane, showing distribution of ferritin-conjugated immunoglobulin on the membrane surface (arrows) in register with intramembrane particles. Equilibration of membranes at -100C was followed by addition of rabbit immunoglobulin monospecific for cytochrome c oxidase and then ferritin-conjugated goat anti-rabbit immunoglobulin (as in Fig. 5b). G, gel-state lipid domains. (X82,500.)
for a few seconds, the intramembrane particles were observed to be completely and randomly dispersed (Fig. 7). This result demonstrates that the integral proteins, including the cytochrome c oxidase-related large particle, can diffuse and rerandomize rapidly in the plane of the membrane during melting-i.e., during the gel state-to-liquid crystalline lipid phase transition. Such reversible thermotropic transitions are not disruptive to the structure or normal metabolic functions of the mitochondrial membrane (14). Thermotropic lipid phase transitions were used to induce lateral diffusion of integral proteins in the plane of the membrane, and immunoglobulin latticing was used to prevent the free lateral diffusion of cytochrome c oxidase. Rabbit immunoglobulin monospecific for the -oxidase and ferritin-conjugated goat anti-rabbit immunoglobulin were added in sequence to membranes below the lipid phase transition temperature (Figs. 5b and 6). After such immunoglobulin latticing of the oxidase on the membrane surface, freeze-fracture revealed that lateral diffusion and rerandomization of the large intramembrane particles did not occur when the membranes were returned to well above their lipid phase transition temperature (Fig. 8). However, under such latticing conditions specific for cytochrome c oxidase, small intramembrane particles (4.4-7.4 nm in diameter) were not inhib-
.TAeX * .iS-Ar..................b
FIG. 5. Deep-etched surface of energy-transducing membrane, showing general distribution of cytochrome c oxidase revealed by ferritin-
conjugated immunoglobulin. Rabbit immunoglobulin monospecific for the oxidase was added to membranes followed by ferritin-conjugated goat anti-rabbit immunoglobulin. (X132,000.) (a) Membranes equilibrated at 250C followed by addition of immunoglobulins. (b) Membranes equilibrated at -10C followed by addition of immunoglobulins. G, gel-state lipid domains.
Cell Biology: H6chli and Hackenbrock
Proc. Natl. Acad. Sci. USA 76 (1979)
1239
"I V
I Cool
to
%-WII"to
r7Q
FIG. 7. Freeze-fractured energy-transducing membrane equilibrated at -100C, returned to 250C, and then rapidly frozen. Intramembrane particles are random. (X67,500.)
At
-nInor IL
ox IgG 10Add RF-Gcytanti-R
A1C4Add
ited from rerandomizing when membranes were returned to above the transition temperature. In control experiments, treatment with nonspecific rabbit immunoglobulin followed by ferritin-conjugated goat anti-rabbit immunoglobulin had no effect on the lateral diffusion of intramembrane particles of any size. These results demonsf rate that lateral diffusion of cytochrome c oxidase in the membrane bilayer lipid can be inhibited by immunoglobulin latticing of the oxidase on the membrane surface and that many smaller, as yet unidentified, integral proteins can diffuse laterally in the hydrophobic interior of the mitochondrial membrane independently of the oxidase (Fig. 9).
IgG
G anti-R IgG
GPL
LPL
-10 C
0B
1
ox
I
Warm to 25 C
DISCUSSION
The combined use of an immunoglobulin probe monospecific for cytochrome c oxidase and thermotropic lipid phase transitions reported here reveals that cytochrome c oxidase can be observed, by freeze-fracture electron microscopy, to be related to large intramembrane particles in the hydrophobic interior of the highly complex energy-transducing membrane of the mitochondrion. From monitoring the distribution of the monospecific immunoglobulin on the membrane surface by deep-etch electron microscopy, and the distribution of large
anti-R IgG
F
\ \ LPL
R
250C
FIG. 9. Result of combining thermotropic lipid phase transitions with immunoglobulin latticing of cytochrome c oxidase. (A) Membrane above the lipid phase transition at 250C. (B) Membrane below ;.S .... Ha .... * t --+W @-> monospecific for cytochrome c oxidase (R cyt ox IgG) binds to the n sat~-. Ir f-i.ra \ oxidase (cyt ox) at -100C. Ferritin-conjugated goat anti-rabbit imI' ... *-Ai l (F-G anti-RIgG) binds to the immunoglobulin momunoglobulin ~. -i-. -.1~4~ .~" ospecific for the oxidase, resulting in a cross-atcnofm bre zoidases. (D) Membrane temperature raised above the lipid phase -. transition to 250C. Lateral diffusion of crosslinked oxidases does not toccur but other unidentified, small, integral proteins (IP) diffuse POW!,"-laterally, independently of the oxidase. LPL, liquid crystalline .~~.~N ~ phospholipid; GPL, gel-state phospholipid.
-
a
FIG. 8. Freeze-fractured energy-transducing membrane equilibrated at -100C, followed by addition of rabbit immunoglobulin monospecific for cytochrome c oxidase followed by addition of goat antirabbit immunoglobulin (as in Fig. 6) and finally warmed to 25C before rapid freezing. Large intramembrane particles are clustered but small intramembrane particles are random (cf. Fig. 7).
(X67,500-)
intramembrane particles in the hydrophobic interior of the membrane by freeze-fracture electron microscopy, it is apparent that cytochrome c oxidase can diffuse laterally in the
offteenergy-transducing membrane. Whether the idase diffuses entirely independently of all other integral proteins in the membrane or diffuses in physical union with one or more integral proteins is not known at present. In either case, ane
ox-
1240
Cell Biology: Hochli and Hackenbrock
our use of immunoglobulin latticing to prevent lateral diffusion of cytochrome c oxidase clearly demonstrates that smaller, as yet unidentified, integral proteins can diffuse laterally independently of the oxidase. The high potential for free lateral translational diffusion by cytochrome c oxidase and other unidentified, smaller integral proteins in the mitochondrial membrane reported here is consistent with compositional and metabolic characteristics indicative of a highly fluid mitochondrial membrane. Although the mass of the energy-transducing membrane is approximately 75% protein (15), only 50% of the protein is integral to the membrane (16, 17). Thus, although an unusually large number of specific metabolically active proteins are associated with the mitochondrial membrane compared to most other membranes of eukaryotic cells, the total protein integral to this membrane is not significantly different from that of most other cell membranes. Our results indicate that the lipid bilayer of the mitochondrial membrane is considerably less crowded with protein than recognized heretofore and that the lateral space requirement for protein diffusion in this membrane is unrestrictive. The unique composition of bilayer lipid in the mitochondrial membrane offers potentially less resistance to the lateral mobility of integral proteins compared to any other membrane of eukaryotic cells. The high percentage of unsaturated phospholipids together with the virtual lack of cholesterol (15) most likely accounts for the relatively low onset temperature of the lipid crystalline-to-gel state lipid phase transition (1), the exceptionally high mobility in the phosphilipid hydrocarbon chains (18), and the intrinsically low microviscosities (in the range of 0.1-0.9 P) reported for the mitochondrial lipid component (19, 20). Because the integral proteins of the energy-transducing membrane do not appear to be latticed or anchored through strong molecular interactions (11), the potential for lateral diffusion by such proteins should be approximately proportional to the degree of fluidity of the lipid bilayer, which in this membrane is quite high. Indeed, in the cholesterol-containing, higher-viscosity, lower-fluidity plasma membranes, lateral diffusion coefficients for integral proteins are as high as 1-5 X 10-9 cm2_sec-1 at 20'C (21-23). These coefficients are equal to an approximate lateral linear displacement of 40 nm in 1 msec. Thus, the potential rate for free lateral translational diffusion of integral proteins in the cholesterol-free, low-viscosity, high-fluidity lipid bilayer of the mitochondrial energy-transducing membrane is well within the time structure of the known rates of electron transfer and energy transduction at physiological temperature. Whether or not cytochrome c oxidase undergoes independent diffusion or diffuses in physical union with the oligomycinsensitive ATPase, its reductant cytochrome c, or indeed, with the remainder of the respiratory chain through cytochrome cl
Proc. Natl. Acad. Sci. USA 76 (1979)
has not been determined. Available data suggest that cytochrome c reacts alternately through its heme group with the hemes of cytochrome cl and cytochrome c oxidase through a weak complementary charge-mediated interaction (24). Such alternate interaction may require two-dimensional lateral diffusion of cytochrome c on the surface of the membrane or diffusion of cytochrome cl and cytochrome c oxidase in the lipid bilayer of the membrane or both. This investigation was supported by Research Grants BMS75-02372 and PCM77-20689 from The National Science Foundation to C.R.H. and research fellowships to M.H. from the Swiss National Foundation and The Muscular Distrophy Association of America. 1. Hackenbrock, C. R., Hochli, M. & Chau, R. M. (1976) Biochim. Biophys. Acta 455,466-484. 2. Slater, E. C. (1953) Nature (London) 172, 975-978. 3. Boyer, P. D. (1965) in Oxidases and Related Redox Systems, eds. King, T. E., Mason, S. & Morrison, M. (Wiley, New York), Vol. 2, pp. 994-1008. 4. Mitchell, P. (1961) Nature (London) 191, 144-148. 5. Williams, R. J. P. (1961) J. Theor. Biol. 1, 1-17. 6. Hackenbrock, C. R. & Miller Hammon, K. (1975) J. Biol. Chem. 250,9185-9197. 7. Hackenbrock, C. R. (1973) J. Cell. Biol. 53, 450-465. 8. Schnaitman, C. & Greenawalt, J. W. (1968) J. Cell Biol. 38, 158-175. 9. Lemasters, J. J. & Hackenbrock, C. R. (1973) Fed. Proc. Fed. Am. Soc. Exp. Biol. 32,516. 10. Layne, E. (1957) Methods Enzymol. 3,447-454. 11. Hochli, M. & Hackenbrock, C. R. (1977) J. Cell Biol. 72,278291. 12. Hackenbrock, C. R. (1972) Ann. N. Y. Acad. Sci. 195, 492505. 13. Hackenbrock, C. R. (1973) in Mechanisms in Bioenergetics, eds. Azzone, G. F., Ernster, L., Papa, S., Quagliariello, E. & Siliprandi, N. (Academic, New York), pp. 77-88. 14. Hochli, M. & Hackenbrock, C. R. (1976) Proc. Natl. Acad. Sci. USA 73, 1636-1640. 15. Colbeau, A., Nachbaur, J. & Vignals, P. M. (1971) Biochim. Biophys. Acta 249,262-492. 16. Capaldi, R. A. & Tan, P. F. (1974) Fed. Proc. Fed Am. Soc. Exp. Biol. 33, 1515. 17. Harmon, H. J., Hall, J. D. & Crane, F. L. (1974) Biochim. Biophys. Acta 344, 119-155. 18. Keough, K. M., Oldfield, E., Chapman, D. & Beynon, P. (1973) Chem. Phys. Lipids 10, 37-50. 19. Keith, A., Bulfield, G. & Snipes, W. (1970) Biophys. J. 10, 618-629. 20. Feinstein, M. B., Fernandez, S. M. & Sha'afi, R. I. (1975) Biochim. Biophys. Acta 413,354-370. 21. Edidin, M. & Fambrough, D. (1973) J. Cell Biol. 57,27-37. 22. Liebman, P. A. & Entine, G. (1974) Science 185,457-459. 23. Poo, M-M. & Cone, R. A. (1974) Nature (London) 247, 438441. 24. Salemme, F. R. (1977) Annu. Rev. Biochem. 46,299-329.
