Biogenesis of a 35-Kilodalton Protein Associated with Outer Mitochondrial Membrane in Rat Liver*

Biochemical analyses following subcellular fraction- ation of rat liver indicated that the outer mitochondrial membrane contains a number of membrane-specific proteins of which a 35-kilodalton species (0"-35) is a prominent component. These results were confirmed and extended by electron microscope immunocyto- chemical tests based on the protein A-gold technique. deep

Biochemical analyses following subcellular fractionation of rat liver indicated that the outer mitochondrial membrane contains a number of membrane-specific proteins of which a 35-kilodalton species (0"-35) is a prominent component. These results were confirmed and extended by electron microscope immunocytochemical tests based on the protein A-gold technique. 0"-35 is tightly bound to the outer mitochondrial membrane, e.g. it was not released by sonication in the presence of 1.5 M KC1 and 0.1% sodium deoxycholate. However, it did not react with the photoaffinity probe azidopyrene, which indicates that OM"35 is located peripherally on the membrane rather than buried deep in the lipid bilayer as an intrinsic protein. Since low levels of detergent were required for 0"-35 in intact mitochondria to react with exogenous antibodies, 0"-35 is probably located on the side of the outer membrane which faces the interior of the organelle.
When rat liver mRNA was translated in a messengerdependent cell-free system derived from rabbit reticulocytes, antiserum against OM"35 precipitated a single polypeptide product which migrated on sodium dodecyl sulfate-polyacrylamide gels with molecular weight characteristics of a protein slightly larger (by M, = 500) than OM"35 obtained from isolated outer mitochondrial membrane. The mRNA coding for OMM-35 was recovered exclusively from membrane-free polysomes. Thus, the route followed for synthesis and subsequent insertion of 0"-35 into the outer membrane of mitochondria is the post-translational pathway which has been previously described for proteins destined for the interior compartments of this organelle.
The question of how content and membrane proteins of various organelles are channeled from their sites of synthesis in the cytoplasm to their final destination in the cell has come under considerable investigation in recent years. For organelles which make up the overd secretory pathway, for example, both their content proteins and many (but not all; see Ref. 16) of the intrinsic proteins buried deep in their delimiting membranes are initially assembled into the rough endoplasmic reticulum by a coupled translation-insertion mechanism prior I( Recipient of a predoctoral studentship from the Canadian Medical Research Council. to being transported to their final location (ct: Ref. 1). For proteins of a number of discrete organelles which do not obviously contribute to the exocytosis-endocytosis pathway, however, a very different mechanism appears to be operating. In the case of mitochondria (2-4), chloroplasts (5,6), and peroxisomes (7), transmembrane uptake of newly synthesized proteins destined for the interior compartments of these organelles involves an exclusively post-translational pathway, z.e. following release of newly completed polypeptide chains from membrane-free polysomes. Very little is yet known, however, concerning the mechanism(s) whereby proteins are assembled into the delimiting membranes of these three organelles. Even though these membranes may share certain compositional and structural features in common with other cellular membranes (most notably the endoplasmic reticulum), they also contain a number of unique proteins which render them functionally distinctive.
In the present study, we have investigated the biogenesis of a 35-kilodalton protein which is located in the outer mitochondrial membrane in rat liver. 0"-35' does not penetrate significantly into the lipid bilayer of the membrane but, nevertheless, is very tightly bound to the membrane at its inner surface. Here, we show that synthesis of 0"-35 occurs on membrane-free polysomes in the cytoplasm. Thus, assembly of this protein into the outer mitochondrial membrane must follow a post-translational insertion mechanism.

MATERIALS AND METHODS
General-For most of the routine procedures used in this study, the methods followed have been outlined elsewhere (see Refs. 4, 8, and 9 and articles cited therein). These include protein measurements, assays of trichloroacetic acid-insoluble radioactive products by the filter disc method, extraction of mRNA by either the chloroform/ phenol method or the guanidinium thiocyanate/CsCl method, protein synthesis in a messenger-dependent (10) cell-free protein-synthesizing system derived from rabbit reticulocytes, and SDS-polyacrylamide gel electrophoresis and radioautography of dried gels. Where individual protocols were employed only for a specific experiment, they are described in detail in the legends to the figures and tables. Young Sprague-Dawley rats (80-200 g) were used throughout this study and were deprived of food for 8-16 h prior to killing.
Purification of OMM-35-Outer mitochondrial membrane was isolated on three separate occasions from a total of 90 g of rat liver by the digitonin technique, exactly as described by Greenawalt (11). Electrophoresis of the combined outer mitochondrial membrane fractions was performed using a total of 16 SDS-polyacrylamide (10%) slab gels containing 400 bg of protein/slot and 10 slots/gel. The gels were stained with Coomassie brilliant blue and destained, and the OMM-35 bands were excised. Bands were diced and embedded in a 1% agarose plug containing 1 ml of normal hamster serum, and protein ' The abbreviations used are: 0"-35, a M, = 35,000 protein in outer mitochondrial protein; SDS, sodium dodecyl sulfate; pOMM-35, the primary translation product of OMM-35 mRNA. 8761 was electroeluted into a dialysis bag. The protein was dialyzed against a total of 8 liters of Hz0 at 4 "C for 2 days and then lyophilized.
Antiserum against 0"-35"Lyophilized OMM-35 and hamster serum proteins were suspended in 2 ml of phosphate-buffered saline and mixed with 2 ml of Freund's complete adjuvant. Subcutaneous injections of this mixture were given to a hamster at 2-week intervals in aliquots containing % of the total preparation for the first two injections and % of the total preparation for the last three injections. Seven days after the final injection, serum was collected from the animal by cardiac puncture.
Immunocytochemical Localization of 0"-35 by the Protein Agold Technique-Small pieces of rat liver were fured for 2 h at room temperature in 1% glutaraldehyde and 0.1 M sodium phosphate buffer, pH 7.4, and embedded in glycol/methyacrylate (12). Thin sections (600-800 A) were obtained and mounted on 200-mesh nickel grids having a carbon-coated Parlodion film. 0"-35 was visualized by the protein A-gold technique (13, 14) as follows (14). Thin sections on grids were incubated for 5 min on a drop of phosphate-buffered saline containing 1% ovalbumin, followed by incubation for 1 h at room temperature on a drop of antiserum against 0"-35 (diluted 50:l digitonin to separate outer mitochondrial membranes from mitoplasts (see "Materials and Methods"). Digitonin-released material was centrifuged at 135,000 X g for 1 h to pellet outer mitochondrial membranes; the supernatant contained soluble proteins which had been located between the inner and outer mitochondrial membranes. To separate inner membranes from matrix proteins, mitoplasts were suspended in 10 ml of 10 m~ Tris-HC1, pH 8.5, and sonicated for eight I-min periods (8). The inner membrane was sedimented by centrifuging at 135,000 X g for 1 h; the supernatant contained soluble matrix proteins. Aliquots of the 4 submitochondrial fractions (-100 pg of protein) were co-electrophoresed in an 8-12% polyacrylamide gel containing SDS. The gel was stained with Coomassie brilliant blue.  (25,26). The suspension was sonicated at 4 "C for six 10-s periods (MSE microsonicator operating with a peak to peak amplitude of 12 pm) and incubated at 25 "C for 30 min. Membranes were sedimented at 105,000 X g for 1 h, suspended in 2 ml of H20, and recentrifuged. A portion of the pellet (125 pg of protein) was electrophoresed in an 8-12% polyacrylamide-SDS gel along with a sample of the same preparation of outer mitochondrial membrane which had not been subjected to sonication in the presence of KC1 and sodium deoxycholate. Closed arrow heads denote proteins associated with outer mitochondrial membrane which were no longer present following the sonication treatment (open arrow heads).
with phosphate-buffered saline). After a brief wash with phosphatebuffered saline, grids were placed on a drop of the protein A-gold complex (14) for 30 min. Finally, grids were washed thoroughly with phosphate-buffered saline, rinsed with distilled H20, and stained with uranyl acetate and lead citrate. They were examined under a Siemens 101 electron microscope.
Control reactions involved (a) preadsorbing antiserum with excess outer mitochondrial membrane prior to immunocytochemical tests, (6) performing immunocytochemical tests with nonimmune serum, and (c) performing tests in the absence of serum (protein A-gold alone). All three yielded negative results.
Isolation of mRNA from Free and Membrane-bound Polysomes-Free and membrane-bound polysomes were obtained from rat liver by the method of Ramsey and Steele (15) exactly as described (16).
RNA was extracted from the resulting pellets by the guanidinium thiocyanate/CsCl procedure (17) and the poly(A') fraction was isolated by oligo(dT)-cellulose chromatography.
Synthesis and Recovery of 0"-35 in VitroSynthesis in vitro was performed by incubating liver RNA in 100 p1 of a messengerdependent (10) rabbit reticulocyte cell-free system containing 0.4-0.6 mCi/ml of [%3]methionine (8,9). After 60 min at 29 "C, the reactionmixture was diluted with 0.9 ml of ice-cold medium containing phosphate-buffered saline, 1% Triton X-100, 20 mM methionine, 0.02% NaN3, and centrifuged at 45,000 rpm for 45 min in the Beckman type 75 rotor. The supernatant was collected and 0.4 ml of 4 M NaCl and 15 pl of antiserum against 0"-35 were added. Following incubation overnight at 4 "C, protein A-coated Staphylococcus cells were added and the mixture was rotated for 1.5 h at room temperature (8). The mixture was centrifuged and the pellet was washed three times at room temperature with medium containing phosphate-buffered saline, 20 mM methionine, 0.02% NaN3, 0.9% Triton X-100, and 0.9% SDS, washed once with saline, and finally boiled in SDS-polyacrylamide electrophoresis buffer to elute immunoreactants which were then loaded directly onto an SDS-polyacrylamide slab gel.

Association of 0"-35
with the Outer Mitochondrial Membrane-The four submitochondrial compartments, i.e. matrix, inner membrane, the space between inner and outer membranes, and outer membrane, were obtained from purified rat liver mitochondria (Fig. 1). An estimate of the quantitative distribution of protein between the 4 fractions is shown in Table I and was found to agree with other reported values (11, 18). The outer membrane fraction constituted roughly 9% of total mitochondrial protein (Table I) and was obtained in good yields (77% of the outer membrane marker monoamine oxidase was recovered here) and in a highly enriched form (as judged by the high specific activity of monoamine oxidase in this fraction). Cytochrome c oxidase measurements invariably showed that our outer mitochondrial membrane preparations were not significantly contaminated (~5 % ) with inner membrane. In order to identify proteins which are located predominantly, or exclusively, in the outer membrane, samples of the 4 submitochondrial fractions were compared by SDS-polyacrylamide gel electrophoresis (Fig. 1). One of the most prom- Submitochondrial fractions were obtained as described in Fig. 1 and were assayed for their relative protein content (23) and for monoamine oxidase activity as described in Ref. 24 inent of these was a polypeptide demonstrating a size of 35,000 daltons (designated 0"-35).

0"-35
is not associated with the outer membrane merely in an adventitious manner but represents a tightly bound structural component. For example, it was not liberated by sonication in 0.1% sodium deoxycholate and 1.5 M KC1 (Fig. 2).
Antiserum Against OMM-35"Antibodies to 0"-35 were obtained by eluting the protein from SDS-polyacrylamide gels and injecting into a hamster. Subsequent screening of the antiserum was performed by reacting it with SDS-polyacrylamide gel profiles of total protein of both outer and inner mitochondrial membranes blotted onto nitrocellulose paper; IgG-antigen complexes were then visualized by reacting the paper with '251-protein A (Fig. 3). Antiserum reacted only with OMM-35. Identity of 0"-35 was unequivocal since the autoradiogram of the nitrocellulose blot could be compared directly to the original SDS-polyacrylamide gel which had been subsequently stained with Coomassie brilliant blue (only about 30% of the 0"-35 band was transferred to the blot). The radioactive band on the blot (Fig. 3)  were subjected to electrophoresis in a 10% polyacrylamide gel containing SDS. A blot of the separated proteins was obtained (27) by electrophoretically transfemng the proteins in the gel to a sheet of nitrocellulose paper ( S and S BA85 paper, 0.45-1.1 pores) at 200 mA (33 V) for 19 h. The extent to which individual protein bands migrated to the nitrocellulose paper following 19 h of electrophoretic blotting varied from 15-100%; approximately 30% of the 0"-35 band was transferred. Blots were rinsed in 100 ml of saline for 2-3 min and incubated for 1 h at 37 "C in 10 ml of a medium containing saline, 3% bovine serum albumin (Sigma type V, fatty acid free), and 0.02% NaN3. Following a brief rinse with saline, the blot was incubated for 5 h at room temperature in a sealed plastic bag containing 100 pl of antiserum against 0"-35 and 10 ml of saline, 3% bovine serum albumin, 0.02% NaN3. The blot was again rinsed with saline and agitated over night in 100 ml of saline, 3% bovine serum albumin, 0.02% NaNa. It was then incubated in a sealed bag with 5 ml of saline, 3% bovine serum albumin, 0.02% NaN3, and 2 pCi of '"I-protein A (20 mCi/mg, Amersham). The mixture was shaken for 1 h at room temperature. The blot was briefly rinsed with saline containing 3% bovine serum albumin and 0.02% NaNl and, finally, was washed extensively over a period of 16 h with six changes of saline containing 0.02% NaN3. Excess liquid was removed from the blot and it was wrapped in Saran Wrap and exposed overnight at -70 "C to an x-ray film pressed onto a Cronex Lighting-plus (Dupont) intensifying screen. Lane A, outer mitochondrial membrane; lane B, inner mitochondrial membrane.
the polyacrylamide gel. In this particular experiment, however, 0"-35 was detected in both outer and inner membrane fractions, perhaps because release of outer membrane from p d l e d mitochondria by digitonin treatment had not been complete (but see below). That the antiserum was monospecific, i.e. reacted with a single antigen in the outer mitochondrial membrane, was demonstrated by two-dimensional rocket immunoelectrophoresis (Fig. 4). Subcellular Distribution of OMM-.3--Biochemical analyses indicated that 0"-35 is enriched in the outer mitochondrial membrane (Fig. 1); it was not present in detectable levels in either total microsomal preparations (Fig. 5) or in a postmicrosomal cytosolic fraction (data not shown). These findings were c o n f i i e d a n d extended by immunocytochemical tests using the protein A-gold technique (Fig. 6). Of a total of 225 grains analyzed, the majority (82%) were located over mitochondria, of which 50% was present on the outer membrane and 30% was present on the inner membrane. The remainder (20%) could not be ascribed to a particular mitochondrial locus. Since there is three times more inner membrane protein compared to outer membrane protein (Table I)

:hondrial Membrane Protein
Topographical Orientation of 0"-35 in the Outer Mitochondrial Membrane-Although 0"-35 was shown to be tightly bound to outer mitochondrial membrane, e.g. it was not released by sonication in the presence of 0.1% sodium deoxycholate and 1.5 M KC1 (Fig. 2), such experimentation does not indicate if this tight association is mediated by peripheral interactions between 0"-35 and the surface of the membrane or results because the protein is buried, either wholly or in part, as an intrinsic component of the lipid bilayer. Fig. 7 provides strong evidence that 0"-35 is associated with outer mitochondrial membrane in a peripheral fashion; it does not enter the lipid bilayer to any significant degree. When isolated outer mitochondrial membrane was reacted with the membrane-penetrating photoaffinity reagent azidopyrene (19), 0"-35 was not among the proteins which became fluorescently labeled following subsequent photolysis. The major intrinsic proteins reacting with azidopyrene were located in the M, = 30,000-33,000 range (Fig. 7 ) .
The next step was to determine on which side of the outer membrane 0"-35 faces, i.e. toward the cytoplasm, toward the inside of the mitochondrion, or both. This was tested by incubating intact mitochondria with antiserum against OMM-35 and measuring the accessibility of 0"-35 to the exogenous antiserum in the presence and absence of low levels of detergent. The results showed ( Table 11) that detergent was required to facilitate antibody-antigen interaction, presumably because the detergent rendered the outer mitochondrial membrane porous to the antibody. Since 0"-35 was not noticeably available for interaction with antibody in the ab-  Purified mitochondria (350 pg protein) were suspended in 100 pl of medium A (phosphate-buffered saline, 0.3 M sucrose, 0.5% bovine serum albumin, and 0.02% NaNs), either with or without 0.01% sodium deoxycholate. Then, 20 pl of either normal serum or antiserum against 0"-35 were added and the mixtures were briefly sonicated and incubated at 25 "C for 2.5 h. The reaction mixtures were diluted with 5 ml of medium A and centrifuged in a type 40 rotor (Beckman) for 1 h at 40,000 rpm. Pellets were resuspended in 200 pl of medium A f 0.01% sodium deoxycholate and briefly sonicated. '"1-protein A (1.0 pCi, 20 mCi/mg, Amersham) was added and, following incubation at 25 "C for 30 min, 4 ml of medium A were added and the mixtures were centrifuged as above. The pellets were rinsed briefly with phosphate-buffered saline, dissolved in NCS (Amersham), and assayed for radioactivity.  (Table 11), it is concluded that the protein is located on the side of the outer membrane which faces the interior of the mitochondrion.

12511-Protein
Biogenesis of OMM-35-When antiserum against 0"-35 was reacted against total polypeptide products synthesized in vitro in a messenger-dependent rabbit reticulocyte system programmed with rat liver mRNA, a single polypeptide was precipitated (Fig. 8). On SDS-polyacrylamide gels, this putative translation product of 0"-35 mRNA was found to demonstrate size characteristics just slightly larger (by about M , = 500) than the form of 0"-35 as it exists in isolated outer mitochondrial membrane (Fig. 8). Although this size difference is small, it is reproducible and has been observed recurrently using a number of different preparations of both outer mitochondrial membrane and messenger. In addition to the immunological evidence, however, it was impossible to demonstrate further equivalence between pOMM-35 synthesized in vitro and 0"-35 synthesized in vivo (e.g. by peptide map comparisons) because of very low levels of incorporation of radioprecursor into 0"-35 which occurred in vivo both in the intact animal and in liver explant cultures. Nevertheless, since only a single radioactive product was precipitated following immunoreaction of reticulocyte lysates with antiserum against 0"-35 (Fig. 8) and since the antiserum was shown to react only with 0"-35 (Figs. 3 and 4), it seems unlikely that the immunoprecipitated product of mRNA translation is anything other than the primary biosynthetic form of OMM-35.
In Fig. 9, translational assays were used to determine the subcellular location of poly(A') RNA coding for 0"-35. Free and membrane-bound polysomes were separated essentially according to the rapid procedure of Ramsey and Steele (15) and mRNA was then extracted from polysomal pellets by the guanidine thiocyanate/CsCl procedure (17). Following incubation of mRNA from free and bound polysomes in the reticulocyte protein-synthesizing system, pOMM-35 was precipitated only from lysates containing mRNA from free polysomes (Fig. 8); pOMM-35 was virtually undetectable in lysates programmed with membrane-bound mRNA.
Conversely, when these same two messenger preparations were tested for their ability to direct synthesis of the secretory protein serum albumin, this polypeptide was detected only in reticulocyte  8 (left). Synthesis of OM"35 in vitro. Rat liver mRNA was incubated in a messenger-dependent (10) rabbit reticulocyte cellfree protein-synthesizing system containing [?3]rnethionine as described (8,9) and, after 60 min, was reacted with antiserum against 0"-35 (see "Materials and Methods"). The resulting immunoprecipitate was electrophoresed beside 150 pg of outer mitochondrial membrane protein in an 8-12% polyacrylamide slab gel containing SDS. The gel was stained, destained, and dried. The gel was then marked in various locations with radioactive ink and radioautographed for 2 days. The radioactive ink spots (small arrow heads) enabled the dried gel and the developed radioautogram to be accurately realigned. The position of stained 0"-35 marker is indicated by large black dots (lane A ). pOMM-35 refers to the in vitro product precipitated by antiserum against OMM-35 (lane B ) . A photograph of the radioautogram is shown. FIG. 9 (right ). OM"35 is synthesized by free polyribosomes. Poly(A') RNA from membrane-bound and free polysomes (16) was incubated in the messenger-dependent rabbit reticulocyte system as in Fig. 8, and total polypeptide products (6.0 and 10.2 X 10" cpm, respectively) were reacted with antiserum against 0"-35.
Immunoprecipitates were electrophoresed in an 8-12% polyacrylamide-SDS gel. The gel was dried and radioautographed. Lane C, control, no mRNA added to reticulocyte lysates; lane MB, mRNA from membrane-bound polysomes; lane F, mRNA from free polysomes. The free and membrane-bound mRNA preparations were also tested for translation of serum albumin; it was synthesized only in lysates incubated with mRNA from membrane-bound polysomes and could not be detected in lysates containing mRNA from free polysomes (data not shown).
lysates incubated with membrane-bound mRNA (data not shown).

DISCUSSION
Two approaches were used in this study to ascertain the subcellular distribution of 0"-35: a biochemical analysis involving separation and isolation of various membrane fractions ( Figs. 1 and 5) and an immunocytochemical analysis where antibody-antigen complexes were allowed to form on thin sections of liver tissue and then visualized following reaction with protein A-gold (Fig. 6). The immunological approach, of course, requires that the antiserum is monospecific and interacts only with 0"-35.
That this was the case was demonstrated by incubating the antiserum with replicas of SDS-polyacrylamide gel patterns of outer and inner mitochondrial membrane proteins blotted onto nitrocellulose paper (Fig. 3). Only 0"-35.IgG complexes were detected following reaction with "'I-protein A there was no evidence for the presence of even minor levels of contaminating antibodies. Moreover, two-dimensional rocket immunoelectrophoresis analyses showed that anti-0"-35 reacted only with a single antigen in the outer membrane (Fig. 4), i.e. the 0"-35 band did not contain additional immunoreactive polypeptides which might have arisen by contamination from the inner membrane. At least with respect to mitochondrial membranes, therefore, there is little doubt that the immunocytochemical technique is detecting anything but 0"-35 (Fig. 6).
When SDS-polyacrylamide gel patterns of outer and inner membranes were compared (Fig. l), 0"-35 was found to be enriched in the outer membrane, even taking into account the fact that there is three times more inner than outer membrane protein (Table I). From experiment to experiment, however, there was a certain amount of variability in that 0"-35 was sometimes detected in rather significant levels in the inner membrane fraction as well (e.g. see Fig. 3). It was usually assumed that such variations arose because of variabilities encountered with the efficient release of outer membrane from intact mitochondria by the digitonin procedure. Immunocytochemical tests (Fig. 6), however, showed a positive reaction over both inner and outer membranes, albeit to a considerably lower extent (&fold) in the inner membrane when calculated per unit membrane surface area. Whether the residual immunoreactivity over the inner membrane arose artifactually or is indicative of an additional location for 0"-35 in the inner membrane remains to be determined. Evidence to elucidate the topographical orientation of 0"-35 in the outer mitochondrial membrane was obtained from a combination of approaches. First, the protein was shown to be tightly bound to the outer membrane (Fig. 2). However, it is not buried in the lipid bilayer as an intrinsic protein; 0"-35 did not react with azidopyrene when this membrane-permeating photoaffinity label was reacted with outer membrane (Fig. 7). Since treatment of intact mitochondria with low levels of detergent was required in order to render 0"-35 accessible to exogenous antibody (Table II), the protein is presumably located on the side of the outer membrane which faces toward the interior of the mitochondrion.
Thus, in terms of functional compartmentalization, OMM-35 must be considered an intramitochondrial protein and, therefore, might be expected to conform to what is already known concerning assembly of proteins which are synthesized in the cytoplasm and transported to their fiial destination inside the mitochondrion. The main feature of this uptake process is that it occurs post-translationally (2-4); newly made polypeptides are transported across either one or both mitochondrial membranes only after release of completed polypeptide chains from cytoplasmic free polysomes has occurred. In the present study, we showed that 0"-35 is made on membrane-free polysomes (Fig. 9) and, therefore, its uptake into mitochondria must likewise follow the post-translational pathway for insertion into a membrane.
It is curious, however, that the primary translation product of OMM-35 mRNA demonstrates a slightly retarded mobility on SDS-polyacrylamide gels relative to the in vivo form of 0"-35, at least as it exists in isolated outer membrane (Fig.  8). The difference in size between the two is only on the order of about M , = 500. Higher molecular weight biosynthetic precursors have been identified for a variety of mitochondrial proteins (3,4,20-22), but in all cases they exist with an extra Mr = 2,000-6,000 piece present in the precursor. A possibility which must be considered to explain the present results, therefore, is that 0"-35 is synthesized in vitro in its final form and the differential which is observed between 0"-35