Use of an antibody to study the location of cardiolipin in mitochondrial membranes.

Abstract Rabbit antiserum to cardiolipin, which is reactive with the polar head but not the nonpolar fatty acid moieties of cardiolipin, was used to explore the location of the polar head of cardiolipin in mitochondrial membranes. Only a few per cent of the cardiolipin in intact mitochondria from rat liver, blowfly flight muscle, Saccharomyces cerevisiae, and Neurospora and none of the cardiolipin in intact beef heart mitochondria is available for binding of anticardiolipin antibody. Freezing and thawing, aging at 45°, or sonication, in the absence or presence of the antibody, increased only slightly the anticardiolipin antibody binding activity of various types of mitochondria. The only mitochondrial preparation showing complete ability to bind anticardiolipin antibody was a mitochondrial precursor fraction isolated from glucose-repressed, anaerobic yeast cells. The isolated outer and inner membrane fractions from rat liver mitochondria also showed very little capacity to bind the antibody; both the cytoplasmic side and the matrix side of the inner membrane, which contains most of the cardiolipin showed little antibody binding activity. Removal of the F1 ATPase molecules from inner membrane vesicles of beef heart mitochondria also failed to unmask antibody binding activity. Neither oxidative phosphorylation nor energy-linked Ca++ transport in intact rat liver mitochondria were influenced by addition of excess anticardiolipin antibody. It is concluded that the polar heads of most of the cardiolipin molecules in the mitochondrial membranes are buried within the structure of the membrane or shielded by the binding of other membrane components.


Use of an
SUMMARY Rabbit antiserum to cardiolipin, which is reactive with the polar head but not the nonpolar fatty acid moieties of cardiolipin, was used to explore the location of the polar head of cardiolipin in mitochondrial membranes.
Only a few per cent of the cardiolipin in intact mitochondria from rat liver, blowfly flight muscle, Saccharomyces cerevisiae, and Neurospora and none of the cardiolipin in intact beef heart mitochondria is available for binding of anticardiolipin antibody. Freezing and thawing, aging at 45", or sonication, in the absence or presence of the antibody, increased only slightly the anticardiolipin antibody binding activity of various types of mitochondria.
The only mitochondrial preparation showing complete ability to bind anticardiolipin antibody was a mitochondrial precursor fraction isolated from glucoserepressed, anaerobic yeast cells.
The isolated outer and inner membrane fractions from rat liver mitochondria also showed very little capacity to bind the antibody; both the cytoplasmic side and the matrix side of the inner membrane, which contains most of the cardiolipin showed little antibody binding activity. Removal of the F1 ATPase molecules from inner membrane vesicles of beef heart mitochondria also failed to unmask antibody binding activity. Neither oxidative phosphorylation nor energy-linked Ca++ transport in intact rat liver mitochondria were influenced by addition of excess anticardiolipin antibody. It is concluded that the polar heads of most of the cardiolipin molecules in the mitochondrial membranes are buried within the structure of the membrane or shielded by the binding of other membrane components.
The antigenic activity of cardiolipin (diphosphatidylglycerol) has been extensively studied in connection with its use in the cule; the specific antigenic activity is associated with the bridging phosphate group between the glycerol residues and the free 2hydroxyl group of the interior glycerol molecule (l-3). These findings suggested that the anticardiolipin antibody might be used as a specific probe to locate the antigenic "heads" of cardiolipin molecules in membranes containing this phospholipid and thus to explore aspects of the molecular topology of the membrane.
In this paper we esaminc the binding of this antibody to mitochondrial membranes, which characteristically contain large amounts of cardiolipin.

EXPERIMENTAL PROCEDURE
Membrane Preparations-Rat liver mitochondria and mitochondrial fractions enriched in outer membrane, inner membrane plus matrix, and inner membrane minus matrix were prepared according to the method of Schnaitman and Greenawalt (4), sonic particles of rat liver mitochondria according to the method of Gregg (5), and mitochondria from Saccharomyces cereuisiae strain D 261 by the method of Guarnieri et al. (6). Membrane fractions enriched in cardiolipin were isolated from X. cerevisiae grown under anaerobic conditions as described by Goffeau et al." Neurospora crassa cells and mitochondria were generously supplied by David Beck, and flight muscle mitochondria from the blowfly Phormia regina by Dr. 1~. Sacktor.
Beef heart mitochondria were prepared as described by Settlemire et al. (7) and beef heart mitochondrial S-particles and TU-particles according to Racker (8). Erythrocytes were obtained from rabbit or rat whole blood collected wit'11 1 part of 3.8% sodium citrate per 9 parts blood.
Analytical Methods-Lipids were extracted and determined according to methods described by Fleischer et al. (9). Xlonoamine oxidase and cytochrome oxidase were used as marker enzymes to estimate the purity of the inner and outer mitochondrial membrane fractions (4). Lipids for the preparation of serodiagnosis of syphilis. Anticardiolipin antibody has been found to combine with the polar head of the cardiolipin mole-the immunizing and standard antigen and for use as chromatographic markers were obtained from Supelco, Inc., Bellefonte, Pennsylvania.
Protein was determined by the method of Murphy and Kies (10).
Antiserum Preparation-The preparation of the immunizing antigen solution, the treatment of the rabbits, and tkre collection of antisera were conducted exactly as described by Inoue and Xojima (2). White male rabbits, 3 to 5 kg, were obtained from the Bar-F Rabbitry, Perry Hall, Maryland.
Eight rabbits were used, sis for the production of antisera and two for control sera.
Determination of Concentration of Anticardiolipin Antibody-The serum from each rabbit was assayed for antibody activity using a modification of the 'Quantitative Slide Test" described by the Venereal Disease Research Laboratory (referred to as VDRL) of the United States Public Health Service (11). The standard antigen solution used for this test contained 900 mg of cholesterol, 300 mg of phosphatidylcholinc, and 30 mg of cardiolipin in 100 ml of ethanol.
To carry out the assay, 20 ~1 of standard antigen suspension were mixed with 50 ~1 of antiserum. The amount of flocculation produced was graded visually on a 0, 1 +, 24, 3 +, 4+ basis. At the end of the S-week immunization period the average antiserum titer was 1:64, that is 50 ~1 of a 1 to 64 dilution of the antiserum was sufficient to induce 2f flocculation in 20 ~1 of the standard antigen suspension.
Bntiserum sterilized by passing through a 0.22 pm Millipore filter was stable at 5" for at least 2 mont'hs. Fresh, unfiltered antiserum was stable at -20" or -196" for at least 1 week. The activity of the antisera was only slightly decreased by lyophilization and storage of the dry powder at -20" for 1 week. Xeasurement oj Antibody Binding to l4itochondria and Other Membranes-Graded amounts of the membrane preparation were suspended in 200 ~1 of 0.15 ilz sodium chloride and mixed with 500 ~1 of various dilutions of the antiserum.
The mixture was shaken for 4 min at 25" and then centrifuged to sediment the membrane with its bound antibody.
To determine the amount of free antibody remaining in the clear supernatant medium, a 50-~1 aliquot of the latter was allowed to react with the standard antigen in the VDRL slide test, and the amount of flocculation recorded.
If no flocculation occurred (recorded as 0), it was assumed that 100% of the antibody originally mixed with the membrane preparation was bound by the latter; if maximum floccuIation (4+) was observed, no antibody was bound by the membrane.
Quantitative comparisons of the antibody binding activity of different membrane preparations were made as follows. Each antiserum was first diluted to such a concentration that 50 ~1 would just produce maximum flocculation of 20 ~1 of the standard antigen suspension, which contained 0.4 nmole of cardiolipin. Each membrane preparation was then titrated with 500 ~1 of such a standardized dilution of antiserum (equivalent to 4.0 nmoles of cardiolipin) as well as with two additional, more concentrated solutions of known concentration prepared from the same antiserum. The amount of antibody bound by each membrane preparation was computed as the average value obtained from titrations with three different antiserum concentrations. For these calculations those flocculation reactions giving a score of either 0 or 4 + on the supernatant medium containing the remaining unbound antibody were not used. The complete details of a typical experiment, showing the procedures for scoring and cal-culating the amount of bound antibody, are given under "Results." RESULTS Amount of Antibody Bound by Intact Rat Liver kfitochondria-Typical experimental data and calculations used to construct a curve ( Fig. 1) describing the binding of anticardiolipin antibody to intact rat liver mitochondria are given in Table I. Graded amounts of mitochondria (from 0.50 to 3.0 mg of protein) suspended in 200 ~1 of 0.15 M potassium chloride were added to 500 ,!.d of three concentrations of standardized antiserum, representing 1: 16, 1:12, and 1:8 dilutions, equivalent to 4, 6, and 8 nmoles of cardiolipin, respectively.
After a 4-min incubation at 25", the mixture was centrifuged and the amount of unbound antibody remaining in the supernatant medium was estimated, using 50-~1 aliquots of the latter in the VDRL slide test. The  flocculation test score was then converted into the percentage of the added antibody that was bound by the mitochondria.
The amounts of antibody bound from the three antiserum dilutions were then averaged and the results expressed as antibody equivalents, defined as the amount of antibody required to titrate 1.0 nmole of cardiolipin in the form of the standard antigen.
As is seen in Table I, intact rat liver mitochondria bind 2.1 antibody equivalents per mg of mitochondrial protein. Fig. 1 shows a plot of the number of antibody equivalents bound per mg of protein as a function of mitochondrial concentration, from the data of Table I. It is seen that intact rat liver    At lower concentrations of mitochondria, below the linear zone, somewhat more antibody is bound per mg of protein than in the linear zone; presumably the enhanced swelling of mitochondria at low concentrations (13) causes exposure of more antibody binding sites.
The antibody binding capacity of fresh intact rat liver mitochondria varied only slightly among different preparations with high respiratory control ratios as is seen in Table II. However, mitochondria with low respiratory control ratios often exhibited much greater antibody binding capacity compared to tightly coupled mitochondria.
On the other hand, cycles of freezing and thawing had no effect on the ability of mitochondria to bind antibody.
From the number of antibody equivalents bound by the mitochondria it is possible to estimate the fraction of all cardiolipin molecules in the intact mitochondria which are available for reaction with the antibody.
Data in Table I show that intact rat liver mitochondria bind, per mg of total protein, an amount of antibody capable of reacting with 2.1 nmoles of cardiolipin. Because intact rat liver mitochondria contain 27 nmoles of cardiolipin per mg of protein, as determined following total lipid extraction and thin layer chromatography, it may be concluded that less than 9% of the cardiolipin in intact rat liver, yeast, and blowfly mitochondria and none of the cardiolipin molecules of heart mitochondria are accessible to the antibody. Neither intact erythrocytes nor Neurospora cells bound antibody (Table III).
However, the microsome fraction of rat liver bound a small amount of antibody; the significance of this effect will be considered below.

Reaction of Mitochondrial
Lipids with Antibody-The data in Table IV show that when the lipids are removed from rat liver mitochondria or microsomes by extraction with solvents, the remaining preparations no longer bind antibody. However, as shown in Table V, antibody-reactive material appears in the lipid extracts.
In these experiments total lipid extracts of yeast, beef heart, and rat liver mitochondria and of liver microsomes were mixed with the adjuvants phosphatidylcholine and cholesterol to yield antigen solutions in ethanol comparable in composition to the standard cardiolipin antigen solution.
The amount of flocculation produced on mixin g anticardiolipin antibody with antigen suspensions prepared with such lipid extracts of membranes showed that the cardiolipin present in total lipid extracts of various mitochondria (determined chromatographically) was as reactive, mole for mole, as the standard antigen prepared from purified cardiolipin.
Thus it is clear that the cardiolipin of mitochondria could be extracted in antibody-reactive form and that the other membrane lipids did not interfere in titrations of cardiolipin with antibody.
The presence of significant antibody binding activity in the microsomal lipids, together with other observations, led us to examine the specificity of the anticardiolipin antibody with purified phospholipids and other compounds. Although the results are to be described elsewhere2, the most pertinent  antibody. It yielded about 25 to 50% of the activity shown by cardiolipin. This observation suggests that the significant reactivity of the lipids from rat liver microsomes with the antibody is actually due to their high phosphatidylinositol content, about 100 nmoles per mg of protein (14), an amount capable of binding 5 to 10 times the antibody added in the flocculation tests. Thus it appears that the phosphatidylinositol of microsomal membranes is also relatively inaccessible to antibody.
Factors A$ectisg Accessibility of Mitochondrial Cardiolipin to Antibody-The data collected in Table VI indicates how accessibility to the antibody is affected by aging of freshly prepared, intact rat liver mitochondria.
For comparison, results from antibody binding studies with yeast and blowfly muscle mitochondria are shown.
When rat liver mitochondria were aged at 45" for 15 to 30 mm in the presence of antibody, the amount of antibody bound was approximately doubled. Similar results were obtained by brief (30 set to 2 min) sonication of rat liver mitochondria in the presence of antibody.
However, if antibody was added after liver mitochondria were aged or sonicated, no increase in antibody binding occurred, indicating that the cardiolipin is not exposed in the vesicles formed from the inner mitochondrial membrane (15), which contains 90% of the total cardiolipin.
Although beef heart mitochondria are very rich in cardiolipin they failed to bind antibody even after aging or sonication in the presence of antibody.
It is of significance, however, that the cardiolipin of a cytoplasmic membrane fraction of anaerobically grown yeast cells containing early precursors of mitochondria was almost totally accessible to the antibody.  Location of Cardiolipin-Because aging, a process which promotes swelling and rupture of the outer membrane, enhanced the accessibility of the cardiolipin of yeast, blowfly, and liver mitochondria to the antibody, mitochondrial subfractions were prepared to determine which mitochondrial membrane had the greatest capacity to bind antibody.
For reference, the phospholipid composition of the inner and outer mitochondrial membranes of rat liver mitochondria is shown in Table VII; it is seen that cardiolipin makes up about 7% of the outer membrane lipids and about 26yo of the inner membrane lipids.
Because the outer membrane contains about 10% and the inner membrane about 90% of the total mitochondrial membrane protein (4), there is about 10 times as much cardiolipin in the inner membrane as in the outer. The amounts of antibody bound by the outer membrane fraction and the inner membrane fraction are shown in Table VIII.
The outer membrane fraction binds on the average over twice as much antibody as an equiva-  This membrane contains almost all of the mitochondrial phosphatidylinositol, equivalent to 70 nmoles per mg of outer membrane protein, which may account for the enhanced antibody binding.
However, microsomes, which contain about 100 nmoles of phosphatidylinositol per mg of protein, have very low antibody binding capacity.
It is obvious that antibody binding by the outer membrane preparation cannot be solely accounted for by contamination with inner membrane material, whose magnitude is indicated by the finding that about 10% of the total cytochrome oxidase, an inner membrane marker enzyme, is present in the outer membrane fraction.
Frozen and thawed or sonicated inner membrane preparations bind no more antibody than untreated preparations.
When the intact inner membrane-matrix fraction, which contains about 35 nmoles of cardiolipin per mg of protein, is treated with Lubrol WX, the resulting inner membrane fraction showed a 4.fold increase in the amount of antibody bound per mg of l)rotein, Table IX. Recause this inner membrane fraction con-tains 73 nmoles of cardiolipin per mg of protein (cf . Table VII), the total antibody-accessible cardiolipin in inner membrane preparations is still only 16.4y0 compared to 9% for the inner membrane-matrix fraction.
Similarly, only 11% of the total cardiolipin of the sonic particles was accessible to antibody. Since the outer vesicular surface of Lubrol-treated or sonictreated inner membrane preparations correspond to the matrix side (RI side) of the inner membrane, it is clear that the mat'rix surface of the inner mitochondrial membrane is only slightly more reactive with anticardiolipin antibody than the cytoplasmic surface. When intact inner membrane preparations were mixed with antibody solutions and incubated at 45" for 15 min, the antibody-reactive sites increased about 3-to 5-fold, equivalent to about 307, accessibility of the total cardiolipin.
It would therefore appear from these experiments that the polar heads of the cardiolipin molecules are largely inaccessible from either the matrix or cytoplasmic surface of the inner membrane. is antibody-inaccessible, and that the membrane components responsible for this inaccessibility nre extremely stable. The only native mitochondrial fraction in which the cardiolipin was found to be accessible to anticardiolipin antibody is a mitochondrial precursor fraction isolated from anaerobically grown yeast cells.' This fraction, which lacks cytochromes, contains oligomycin-sensitive ATPase activity. This fact suggests that the ATPase enzyme complex is not one of the membrane components that masks or covers the polar head of cardiolipin. However, because of the importance of the BTPase enzyme complex to mitochondrial function, the relation between ATPase activity and anticardiolipin antibody binding was further studicd. The ability of inner membrane vesicles stripped of oligomycin-sensitive ATPase activity to bind antibody was examined.
The results of this experimeni. are shown in Table X. Beef heart S-particles prepared by sonicating beef heart mitochondria and TU-particles prepared by treating S-particles with trypsin and urea (8) were found l-o have negligible capacity to bind the antibody.
The very slightly increased ability of the TU-particles to bind the antibody is not in proportion to the 9570 loss in the ATPase activity of these particles.
Thus in beef heart mitochondria it does not appear likely that the polar heads of the membrane cardiolipin serve as binding sites for the oligomycinsensitive ATPase enzyme complex.

Reaction of Antibody with Delipidized LIIitochondrial
Enzymes and Other E+rote&s-In order to determine whether or not membrane proteins in general have the capacity to cover the antigenic head of the cardiolipin molecule, several lipid-free mitochondrial proteins were used to prcpure cardiolipin-protein complexes, which were then allowed to react with the antibody.
The delipidized proteins, suspended in 0.15 M NnCl, were mixed with cardiolipin in the same fashion used t'o prepare the immunizing antigen complex.
When delipidixed cytochrome c and a partially purified preparation of fi-hydrosybutyrate dehydrogenase (16) were mixed with cardiolipin, they inhibited antibody binding by the lipid to approximately 25 and 50yo,, respectively. When the total delipidized membrane proteins of rat liver mitochondria, liver microsomes, and rat erythrocgtes were mixed with pure cardiolipin more than 75% of the added lipid became antibody-inaccessible.
However, the degree of inhibition of antigenic activit,y varied somewhat with the technique used to Issue of Deccmbcr 29, 1971 extract the membrane lipids.
Delipidized yeast mitochondria inhibited only slightly (less than 12%) the binding of antibody to the cardiolipin antigen.
E&?d of CCP on ilntibody dccessibili2?/--8ddition of CaClz (80 nmolcs per mg of protein) significantly decreased the already small degree of binding of anticardiolipin antibody to intact rat liver mitochondria.
The effect was not observed wit,11 similar concentrations of illg zi-, ;\3n2+, Sr2f or with 8 nmoles of Las+ per mg of protein.
Dinit'rophenol partially reversed this inhibition. Control tests showed that Ca2+ did not interfere with the interaction between the standard antigen and the antibody.
Treatment of mitochondrial sonic particles or Lubrol membranes with EDT,\ had no effect on antibody binding.
Eflect of Anticardiolipin Antibody on Miloclzondrial E'unction-There was no effect on the Stat'e 4-State 3 respiratory transitions, nor in the stimulation of respiration by Ca++, as measured with the oxygen electrode, of rat liver mitochondria which had been suspended in 0.25 b~ sucrose solution containing antibody equivalent to 16 to 32 nmoles of cnrdiolipin per mg of mitochondrial protein for 3 to 4 hours at 0-F.
These observations indicate that neither oxidative phosphorylation nor energy-linked Cat'+ transport require or involve cardiolipin molecules in a role which their polar heads are exposed in such a manner that they are accessible to the antibody.
Moreover they also indicate that whatever membrane components are bound to t'he polar heads of the membrane cardiolipin, they are very stable and not displaced by the antibody. DISCUSSION The data reported in this paper indicate that intact mitochondria from rat liver, beef heart', yeast, and blowfly flight muscle show little or no capacity to bind rabbit antibody to cardiolipin, a characteristic component of mitochondrial lipids. Thus very little of the cardiolipin present in these membranes is oriented in such a way that the polar head of cardiolipin, which is the antigenic portion, is accessible to the antibody.
The small amount of antibody binding that t'akes place with intact rat liver, yeast, and bloIvily mitochondria can be increased only by heating or sonicating the mitochondria in the presence of the antibody. However, no more than 309$ of the toOa membrane cardiolipin becomes accessible to antibody following such treatments. Complete reactivity of the mitochondrial cardiolipin with antibody was observed only by extracting the cardiolipin from the mitochondria with chloroform-methanol mixtures. Beef heart mitochondria, which contain about four times as much cardiolipin as rat liver mitochondria, failed to bind any detectable amount of anticardiolipin antibody, even after they were aged at 45" for 30 min. Moreover, only a very small amount of antibody was bound by beef heart sonic particles. The complete failure of intact beef heart mitochondria to bind anticardiolipin antibody suggested the possibility that the small amounts of antibody bound by intact liver mitochondria may be due to the presence in the latter of reactive antigens other than cardiolipin.
We have found that phosyhatidylinositol but no other phospholipid reacts significantly with anticardiolipin antibody.2 Since beef heart mitochondrial lipids contain only 2 to 3% of phosphatidylinositol (14)) whereas those from rat liver mitochondria contain up to 10% phosphatidylinositol, all of which is concentrated in the outer membrane (Table VII), it appears possible that none of the cardiolipin in int.act rat liver mito-chondria is accessible to the antibody and that the limited antibody binding observed in intact rat liver mitochondria is due to outer membrane phosphatidylinositol.
Our observations also suggest that cardiolipin is probably not involved in the binding of Fr ATPasc molecules to the inner mitochondrial membrane. It may be noted, however, that acidic phospholipids appear to be essential components of the electron transfer process (17). Moreover, a portion of the cardiolipin of beef heart mitochondria is tightly bound to cytochrome oxidase (18) and can be isolated from a proteolipid fraction (19). Our results also suggest that phosphatidylinositol is located in the microsomal membrane such that its polar head is obscured by other membrane components.
Thus two types of acidic phospholipids seem to be "buried" in their membranes, but this architecture is not characteristic of all membrane lipids.
PllOSpholipase D readily hydrolyzes phosphatidylcholinc of intact mitochondria, indicating that the polar head of this phospholipid is largely exposed.3 Uoreover, the polar heads of some lipids are exposed in other membranes, as is shown by phospholipase C treatment (20) and antibody binding studies (21) on erythrocyte membranes and on myelin (22).
There are some indications that cardiolipin must be present in a specific three-dimensional micellar arrangement in order to be immunogenic a,nd to be reactive with the anticardiolipin antibody. For example, Inoue and Nojima (2) and Kataoka and Noima (23) have shown that phosphatidglcholine is a necessary auxiliary lipid for the immunogenicitg of cardiolipin. Moreover, Azzi et al. (24) have shown t'hat phosphatidylcholine influences the binding of cardiolipin to cytochrome c. Thus cardiolipin must be oriented in the membrane in such a way that cardiolipin-phosphatidylcholine interaction may occur. The experimental approach described in this paper appears to have some general applicability as a means of probing the molecular topology of other types of membranes.
Such studies recently have been reported for erythrocyte membranes by Nanni et al.
(21) and for myelin membranes by Rapport (22). Other antilipid antibodies have been reported, such as antiphosphatidylinositol (23), antiphosphatidylcholine (25), and antidihydrosphingosine (26). AnCbodics to several sphingolipids of neural tissues (27) have also been reported; they promise to be useful probes in topochemical studies of the mcmbranrs of the nervous system.