The binding of cytochrome b5 to phospholipid vesicles and biological membranes. Effect of orientation on intermembrane transfer and digestion by carboxypeptidase Y.

A method is described which allows the direct measurement of intermembrane protein transfer. We have used this method to examine the transfer of cytochrome bg from artificial phospholipid vesicles and biological membranes. !l!his method involves the incubation of small, sonicated phospholipid vesicles with either biological membranes or large unilamellar phospholipid vesicles (Enoch, H. G., and Strittmatter, P. (1979) Proc. Natl. Acad Sci. U. S. A. 76, 145-149) and subsequent separation by gel filtration. We have observed cytochrome ba transfer between large and small single bilayer vesicles when cytochrome b5 was bound to preformed egg phosphatidylcholine vesicles. The cytochrome bs of microsomes, however, did not transfer to small vesicles; neither did cytochrome bg reductase nor exogenous, bound cytochrome bs. In fact, no detectable protein was transferred when high salt-washed microsomes were mixed with small, sonicated vesicles. Similar results were obtained using mitochondria and nuclear membrane fragments. We conclude that integral membrane proteins in general do not readily undergo intermembrane transfer between biological membranes. The ability of cytochrome b5 to transfer from artificial membranes and not from biological membranes may reflect a difference in the nature of the protein binding. A nontransferable form of cytochrome bg, which may represent the microsomal type of binding, was obtained when cytochrome bg was bound to preformed vesicles of dimyristyl phosphatidylcholine or when cytochrome b5 was bound during the formation of phosphatidylcholine vesicles. A soluble, heme peptide fragment of cytochrome bs was released when vesicles containing cytochrome be in the transferable form were incubated with carboxypeptidase Y. In contrast, the nontransferable form of cytochrome b5 in microsomes and artificial vesicles was not released by carboxypeptidase Y treatment. We conclude that there are at least two possible orientations of cytochrome bs in phospholipid bilayers, and that these orientations result in either hindered or rapid intermembrane transfer.

A method is described which allows the direct measurement of intermembrane protein transfer.
We have used this method to examine the transfer of cytochrome bg from artificial phospholipid vesicles and biological membranes.
We have observed cytochrome ba transfer between large and small single bilayer vesicles when cytochrome b5 was bound to preformed egg phosphatidylcholine vesicles.  Bouma et al. (5) have shown that several proteins of erythrocyte membranes may be transferred to dimyristyl phosphatidylcholine vesicles in d-o, and one of these proteins was identified as acetylcholinesterase (6). In this report we describe a simple, rapid method for the detection and quantitation of intermembrane protein transfer. Using this method, we found that in a number of biological membranes there was no detectable transfer of any one of several membrane proteins. This suggests that these intracellular membrane proteins (integral) may have structural features which result in binding to membranes in essentially irreversible interactions.
Using the binding of cytochrome bs to artificial phospholipid vesicles as a model system, we found two types of protein binding: one was capable of intermembrane transfer, the other was not. Based on these properties, we have proposed a model for two different orientations of the protein in membranes.

MATERIALS AND METHODS
Cytochrome ba (7), the cytochrome bs heme peptide (8), and nonpolar peptide (9) and NADH-cytochrome 65 reductase from steer liver (IO), stearyl-CoA desaturase from rat liver (ll), carboxypeptidase Y from bakers' yeast (12, 13), and egg phosphatidylcholine from hen eggs (14) were prepared as described previously. Cell fractions were prepared by differential centrifugation and density gradient centrifugation: microsomes from rat, rabbit, and steer liver (15) RESUl.TS The experimental system designed to detect and quantitate intermembrane protein transfer is based upon the ability to mix and then to separate populations of large and small phospholipid vesicles. The small vesicles used were 200 to 250 A diameter, sonicated phospholipid vesicles, and the large vesicles were either 1000 A, unilamellar phospholipid vesicles or purified, cellular membrane fractions (microsomes, nuclear membrane fragments, or mitochondria).
Cytochrome br, was mixed with either the large or small vesicles and allowed to bind; after binding was complete, vesicles of the alternate size were added and incubated for appropriate times to allow transfer to occur. The preparations were then subjected to gel fiitration (or centrifugation in the case of mitochondria) to separate the large and small vesicle populations.
Cytochrome bs and phospholipid concentrations were measured in each population and the amount of intermembrane transfer was determined.
The first experiments were carried out in a completely artificial system using egg phosphatidylcholine to form both the large and small vesicles. An example of cytochrome bg transfer from large to small vesicles is illustrated in Fig. 1. As shown in Fig. LA, the large and small vesicles were well separated by gel filtration on Sepharose 4B. We have previously reported that there was no fusion between large and small vesicles, even after 24 h of mixing (24). This was tested and verified in each experiment in the present study by measuring the recovery of lipid in the two fractions; in no case was vesicle fusion detected. Fig. 1B shows that the binding of Gel filtration was performed on a column (60 x 0.9 cm) of Sepharose 4B at 4"C, 0.5-ml fractions were collected and analyzed for cytochrome bs and phospholipid. A, 0.4 ml of large vesicles (5.2 mu) was mixed with 0.07 ml of small vesicles (30 mu) and incubated 90 min at 30°C; B, 0.5 ml of large vesicles (4.8 mM) was mixed with 0.05 ml of cytochrome bs (0.45 mu) and incubated 2 h at 30°C; C, 0.5 ml of large vesicles (5.8 mu) was mixed with 0.06 ml of cytochrome bg (0.45 mM) and incubated 2 h at 30°C; 0.075 ml of small vesicles (40 111~) was then added and incubation continued for 90 min at 30°C. 0, phospholipid; A, cytochrome b5.
cytochrome bs to large vesicles was complete after 2 h at 30°C. After complete binding of cytochrome b5 to large vesicles, small vesicles were added, allowed to incubate for 90 min, and then the preparation was subjected to gel filtration ( Fig. 10. In this case, cytochrome bs was distributed between the large (24%) and the small (76%) vesicles indicating that intermembrane transfer of the protein must have occurred. In similar studies the time of mixing was varied in order to obtain an estimate of the rate of transfer. It was found that transfer was approximately 50% of maximum after 5 min of mixing and essentially complete after 1 h (no further transfer was observed after an additional 3 to 4 h of mixing). Subsequent experiments were all performed using 90 min of mixing. A number of the other conditions of this experiment were varied to test their effect on the extent of cytochrome bs transfer ( Table I). The extent of transfer was little altered when the time of binding of cytochrome bg to large vesicles was varied from % to 72 h or the ratio of cytochrome bg to phospholipid was varied from 1:lOO to 1:250. However, when the ratio of the donor to acceptor phospholipid concentration was varied from 1:l to 1:4 a small increase in the amount of transfer was observed. This result with a large excess of acceptor membrane (1:4) suggests that a small amount of the cytochrome bg bound to the large vesicles (-10%) may not be transferable. The data also indicate that the transferable form of the protein may bind preferentially to the small vesicles. By essentially reversing the order of vesicle addition, we were able to show that cytochrome bg bound to small vesicles (formed from either egg or dioleyl phosphatidylcholine) can undergo transfer to large vesicles (Table I, Experiment 2). The amount of cytochrome b5 capable of undergoing transfer in this case is difficult to estimate because the affinity of the protein for the small vesicles is apparently relatively higher (see above). Several transfer experiments were also carried out using either the nonpolar peptide segment of cytochrome b5 or cytochrome b5 reductase. Both proteins were transferred from small vesicles to large vesicles and vice uersa using egg phosphatidylcholine vesicles throughout (data not shown). We next asked the question: can cytochrome b5 and other proteins of natural membranes undergo intermembrane transfer? We found that no detectable cytocrome bs, cytochrome bs reductase, or protein appeared to undergo transfer to small vesicles from high salt-washed microsomes obtained from three different species of animals (Table II, Experiment 1). The catalytic segments of both the reductase and cytochrome bs are known to be exposed to the outside face of microsomal vesicles (15) and this was verified in our preparations by showing that cytochrome bs was fully reduced by the addition of NADH. No transfer could be detected for the cytochrome b5 or protein of nuclear membrane fragments or for the protein of mitochondria. This indicates that the proteins on the outer face of these membranes also do not readily undergo intermembrane transfer. When added to these membranes and allowed to bind, exogenous cytochrome bs did not transfer to small vesicles (Table II, Experiment 2) in contrast to the result obtained with cytochrome b5 bound to large vesicles of egg phosphatidylcholine.
In order to test whether the nontransferable nature of the cytochrome bs found in microsomes was due to the difference in the phospholipid composition of the microsomal membrane, we bound cytochrome bg to large vesicles formed from a lipid extract of microsomes. The cytochrome b5 on these vesicles was bound in a form which was transferable to small vesicles (Table II, Experiment 3) suggesting that the tight binding of cytochrome b5 to microsomes is not determined solely by the lipid composition. Two types of cytochrome bg binding are thus distinguishable, one type found in microsomes ("tightly bound") and  another observed in some artificial membrane systems ("loosely bound"). We next attempted to find a model system which would mimic the tight binding of cytochrome bg in microsomes. When cytochrome bs was included during the detergent was then removed by gel filtration aa described (< 1 mol of deoxycholate remained/700 mol of phospholipid). Following the binding of cytochrome bs the pH was adjusted to 6.5 with 0.1 N acetic acid and carboxypeptidase Y was added (1 mo1/20 mol of cytochrome bb and final concentrations of cytochrome b, of 40 to 50 PM). After digestion for 24 h (2 h in Experiment 5) at 3O"C, the preparations were subjected to gel filtration on Sephadex G-75 at 25°C to separate the vesicle-bound cytochrome bs (excluded fraction) from the released, soluble cytochrome bs (included fraction). formation of large diameter phosphlipid vesicles in the presence of sodium deoxycholate as described for Experiment 3, Table III, the protein appeared to bind in a form that was not transferable (data not shown). It was shown using NADH and the soluble catalytic fragment of cytochrome bg reductase (25, 26) that 95% of the heme portion of the cytochrome bg was exposed to the outside of this vesicle preparation.' The vesicles were indistinguishable by electron microscopy (24)

648'6
Cytochrome bg Membrane Binding during protein binding, either detergent interaction with the protein prior to binding or detergent interaction with, the phospholipid during binding could lead to the insertion of cytochrome bg into the bilayer in a different orientation thereby resulting in a nontransferable form of the protein. The difference was not due to the continued presence of deoxycholate which was removed by gel filtration following binding; less than 1 mol of detergent/50 mol of cytochrome b5 was present in these preparations.
Tight binding of cytochrome b5 to artificial vesicles can be produced by several other techniques. For example, incubation of cytochrome bs with small, sonicated vesicles of dimyristyl phosphatidylcholine alone resulted in complete "tight" binding of the protein (data not shown). Thus, in previous experiments in this laboratory using cytochrome b5 bound to dimyristyl phosphatidylcholine vesicles (27) the protein was in the tightly bound form similar to microsomal cytochrome bs in terms of its ability to transfer and its susceptibility to proteolytic cleavage (see below). Incubation of cytochrome bs with large vesicles of egg phosphatidylcholine containing stearyl-CoA desaturase also led to complete tight binding of the heme protein (data not shown). The cytochrome b5, thus bound was functional with cytochrome b5 reductase (shown by cytochrome bs reduction) and stearyl-CoA desaturase (shown by formation of oleyl-CoA). This is a further indication that the tightly bound form of cytochrome bs in vesicles has an orientation similar to that of the protein in the microsomal membrane. It has not yet been possible to determine whether the loosely bound form of cytochrome b5 is also functional with desaturase. In these same experiments no transfer of desaturase from large vesicles to small vesicles could be detected either by enzymatic activity or protein measurements.
We have also established that the transferable and nontransferable forms of cytochrome bs differ in their susceptibility to digestion by carboxypeptidase Y. When large or small, single bilayer egg phosphatidylcholine vesicles containing the loosely bound form of cytochrome bs were incubated with carboxypeptidase Y, a soluble, intermediate size fragment of cytochrome bs was initially released from the vesicles (2 to 4 h) and this fragment was then further digested (20 to 24 h) to a lower molecular weight form which may correspond approximately to the soluble, heme peptide of cytochrome bg (Fig.  2). The amount of cytochrome b5 releasable by this treatment was variable with up to 95% released from large vesicles and 93% from small vesicles (Table III). We tested whether the carboxypeptidase Y-releasable fraction corresponded to the transferable fraction of cytochrome bs by measuring both fractions in a single preparation of large vesicles. Excellent correlation was obtained (72% of the cytochrome bg transferable, 71% of the cytochrome bs released by carboxypeptidase Y) suggesting that both of these functions are properties of the "loose binding" form. In contrast, no release of soluble heme peptide fragments could be detected after treatment of the tightly bound form of cytochrome bs (inserted in egg phosphatidylcholine with deoxycholate or in dimyristyl phosphatidylcholine vesicles) under the same conditions for up to 24 h ( Fig. 2 and Table III). Furthermore, there was no release of the exogenously bound cytochrome b5 of microsomes by carboxypeptidase Y (Table III), thus extending the correlation between proteolytic release and the ability of cytochrome bs to undergo intermembrane transfer. A number of experiments have been carried out in order to determine whether the loose binding form of cytochrome bg can be easily and completely converted to the tight binding form. Two techniques have thus far been successful. The fist (  cles with deoxycholate (1 mo1/2 mol of lipid). This level of detergent causes the vesicles to become highly permeable without disrupting the bilayer structure (24, 28). Following this treatment the cytochrome b5 was no longer releasable by carboxypeptidase Y and this property was retained even after removal of the detergent by gel filtration. Lower levels of deoxycholate added to vesicles (50.1 mol/mol of lipid), which presumably cause considerably less perturbation of the bilayer, did not result in any change in the amount of cytochrome b5 releasable by carboxypeptidase Y. Cytochrome bg bound to small egg phosphatidylcholine vesicles was converted to the tightly bound form by sonication (Table IV,  Experiment 2). Similarly, only the tightly bound form was obtained when cytochrome bg was present during initial vesicle formation by sonication. In preliminary experiments, other methods were tested for their effect on the conversion to tight binding: prolonged incubation (-24 h) of cytochrome bs-containing vesicles at elevated temperature (37"C), inclusion of other membrane-binding proteins in the vesicles (e.g. gramacidin S), and inclusion of lysophosphatidylcholine in the vesicles (up to 20% of the lipid). Thus far these methods have been only partially successful and have not led to the complete conversion of cytochrome b5 from the loose binding to the tight binding form. DISCUSSION We have described an experimental system which allows one to detect and quantitate intermembrane protein transfer. The system should be directly applicable to the study of the transfer of other membrane components including other membrane proteins, phospholipids, glycolipids, steroids, proteolipids, etc. We have reported here that cytochrome b5 and cytochrome b5 reductase bound to artificial membranes formed from egg phosphatidylcholine are capable of undergoing intermembrane transfer, thus, confiiming previous observations (3,4), and these results were extended to include the nonpolar peptide segment of cytochrome bg. The major and contrasting new observation is that intrinsic (or integral) membrane proteins on the endoplasmic reticulum, mitochondria, and nuclear membranes do not undergo transfer from their natural membrane loci in U&O. Since no exceptions were found, it appears that this may be the dominant situation in Go. Although intermembrane protein transfer in uiuo may not be a general phenomenon, it may occur under specific circumstances with certain proteins (5).
The orientation of endogenous and exogenously bound cytochrome ba on microsomes appears to be similar not only on the basis of the observed interactions with the cytochrome b5 reductase and stearyl-CoA desaturase (15) but also in binding in a nontransferable form which is resistant to carboxypeptidase Y digestion (this report). Apparently, proper insertion of cytochrome b5 is achieved in the isolated microsomes in vitro (also mitochondrial and nuclear membranes) suggesting that insertion of the protein into the membrane in uiuo may occur subsequent to and independently of protein synthesis. This would be in contrast to the insertion of a class of proteins that bind to the membrane at their NH2 terminus which is then extruded through the membrane during protein synthesis (29). Finally, we have described a model membrane system in which cytochrome b5 may bind in two different orientations as recognized by differences in the abiltiy of the protein to undergo intermembrane transfer and digestion by carboxypeptidase Y. Since only one of these forms is observed in biological membranes, it is important to know the structural basis of these two types of binding. When pure cytochrome bg is incubated with single bilayer egg phosphatidylcholine vesicles the protein binds to the membrane in such a way that (a) the COOH terminus is available for digestion by carboxypeptidase Y and (b) the protein can undergo intermembrane transfer and is referred to here as loose binding. Both of these properties are lost when cytochrome b5 is bound during the formation of large, single bilayer egg phosphatidylcholine FIG. 3. Models for the orientation of the nonpolar, membranebinding segment of cytochrome 65 in the membrane. Heme peptide, residues 1 to 87; linkage peptide, residues 88 to 97; nonpolar peptide, residues 98 to 133 (9, 31). I, ZZ, loose binding forms; ZZZ, ZV, V, tight binding forms.
vesicles in the presence of deoxycholate (referred to as tight binding).
We propose that these two operationally defined forms of cytochrome bs actually represent two different modes of orienting the nonpolar, membrane-binding segment of the protein in the bilayer. When cytochrome b5 binds to pure, unperturbed bilayers, the loose binding form is predominately obtained. However, if the bilayer is in a perturbed state due to presence of deoxycholate or another integral membrane protein, i.e. desaturase, in the bilayer then cytochrome bs is inserted in the tight binding form. Only the tight binding form of cytochrome bs is observed with dimyristyl phosphatidylcholine which forms less stable, highly permeable vesicles (30). These results suggest that the insertion of cytochrome b5 into the membrane in the tight binding form may require some perturbation of the phospholipid bilayer.' Consistent with this, we have found that the loose binding form could be converted to the tight binding form by sonication or addition of deoxycholate, both of which may cause local disruption or perturbation within the bilayer. Barring any perturbation of the bilayer, the two orientations of cytochrome 65 appear to be stable and do not undergo interconversion.
We are now attempting to define the molecular basis for the difference in the properties of the two types of cytochrome bs binding and also to determine the difference in the orientation of the polypeptide chain with respect to the bilayer, or in the tertiary structure of the nonpolar peptide per se, or both. Fig. 3 diagrams some of the orientations which may account for the observed differences in cytochrome b5 intermembrane transfer and in the accessibility of the COOH terminus to digestion by carboxypeptidase Y. The most direct interpretation of the data is that the loss of accessibility of the COOH terminus to carboxypeptidase Y arises from the placement of the COOH terminus at the aqueous interface in the interior of the vesicle, 111, which is not accessible to the protease. Wickner (32) came to a similar conclusion concerning the orientations of the Ml3 virus coat protein after a study of the proteolytic digestion of the protein in artificial vesicles. Such a model for cytochrome by, binding would also explain the observed absence of intermembrane transfer. Placing the COOH terminus on the inside of the vesicle results in a thermodynamic barrier to transfer due to the energy required to move the charged residues (9) through the nonpolar interior of the bilayer as predicted by the hydrophobic effect (33,34). We have presented several other models (Fig. 3) that are * In contrast, the insertion of cytochrome bs reductase in a tight binding form appears to be more complex since rapid transfer of the reductase between dimyristyl phosphatidylcholine vesicles has been observed (3). consistent with our data. but the thermodvnamic basis for the 15. Strittmatter, P., Rogers, M. J., and Spatz, L. (1972)

membranes. Effect of orientation on intermembrane transfer and digestion by
The binding of cytochrome b5 to phospholipid vesicles and biological