The Fate of Membrane-bound Ribosomes Following the Termination of Protein Synthesis*

Contemporary models for protein translocation in the mammalian endoplasmic reticulum (ER) identify the termination of protein synthesis as the signal for ribosome release from the ER membrane. We have utilized morphometric and biochemical methods to assess directly the fate of membrane-bound ribosomes following the termination of protein synthesis. In these studies, tissue culture cells were treated with cycloheximide to inhibit elongation, with pactamycin to inhibit initiation, or with puromycin to induce premature chain termination, and ribosome-membrane interactions were subsequently analyzed. It was found that following the termination of protein synthesis, the majority of ribosomal particles remained membrane-associated. Analysis of the subunit structure of the membrane-bound ribosomal particles remaining after termination was conducted by negative stain electron microscopy and sucrose gradient sedimentation. By both methods of analysis, the termination of protein synthesis on membrane-bound ribosomes was accompanied by the release of small ribosomal subunits from the ER membrane; the majority of the large subunits remained membrane-bound. On the basis of these results, we propose that large ribosomal subunit release from the ER membrane is regulated independently of protein translocation.

The endoplasmic reticulum (ER) 1 membrane, the site of nascent secretory and integral membrane protein translocation, contains an abundance of membrane-bound ribosomes. As is now well established, the association of biosynthetically active ribosomes with the ER membrane occurs through the activity of the signal recognition particle/signal recognition particle receptor targeting machinery (1)(2)(3). By this process, ribosomenascent chain complexes bearing secretory or membrane protein precursors are recognized early in synthesis and are trafficked from the cytosol to the ER membrane (4). Following the binding of the ribosome-nascent chain complexes to the resident translocon complex of the ER membrane, protein translation proceeds and nascent chains are translocated across or integrated into the ER membrane. Subsequently, the termination of protein synthesis is thought to elicit the release of the ribosomal subunits from the ER membrane to the free, cytoplasmic pool (5,6). In the cytoplasmic pool, the ribosomal subunits are free to participate in the protein synthesis initiation sequence. Should they engage in the synthesis of a secretory or membrane protein precursor, targeting to the ER membrane again occurs, thus describing a cycle of ribosome binding and release.
In the pioneering studies on ribosome-membrane interactions in the ER, it was observed that ribosomes bind asymmetrically, with the binding interaction being mediated entirely through the large subunits (7)(8)(9)(10). Thus, if isolated ER microsomes were extracted with increasing concentrations of EDTA, the small ribosomal subunit was preferentially released from the membrane, whereas the large ribosomal subunit remained membrane-bound (7,8). Confirmation of the nature of ribosome binding to the ER was later obtained by direct ultrastructural analysis of intact cells, where it was observed that the large ribosomal subunit contains the site of membrane attachment (10). On the basis of these observations, it may be predicted that the termination of protein synthesis by membrane-bound ribosomes would be accompanied by the regulated dissociation of large ribosomal subunits from the ER membrane.
Do ribosomes cycle between a membrane-bound and free state? This question was first addressed in biosynthetic labeling and exchange studies and yielded differing and somewhat contradictory conclusions (9,11). In the study of Baglioni et al. (9), it was proposed that large ribosomal subunits bind to the ER membrane prior to their assembly into polysomes. In contrast, Mechler and Vassalli (11), using a similar isotope incorporation protocol, concluded that small and large ribosomal subunits enter the membrane-bound polysome fraction from the cytosolic pool at similar rates. Although these studies are in disagreement regarding the entry path for large subunits into membrane-bound polysomes, two points are evident. 1) The biosynthetic labeling and exchange kinetics of membranebound large subunits differ from those of membrane-bound small subunits. 2) Membrane-bound and cytosolic ribosome pools constitute a common population. However, because ribosome synthesis and assembly are relatively slow and complex processes, the resolution offered by ribosome biosynthetic labeling studies is not sufficient to analyze directly the temporal coupling between protein synthesis on membrane-bound ribosomes and ribosome exchange between free and bound ribosome pools.
The purpose of this study was to investigate experimentally the compartmental fate of membrane-bound ribosomes following the termination of protein synthesis. Do ribosomes dissociate from the ER membrane coincident with the termination of protein synthesis? By using biochemical and morphometric methods in tissue culture cells, we report that in vivo termination yields the preferential release of ribosomal small subunits from the membrane, whereas the large ribosomal subunits remain predominantly in the bound state.

EXPERIMENTAL PROCEDURES
Reagents-35 S-Labeled Pro-Mix ([ 35 S]methionine and -cysteine) and 5,6-[ 3 H]uridine were obtained from Amersham Pharmacia Biotech. Pactamycin was obtained from the Upjohn Co. Cell culture media and reagents were from Life Technologies, Inc. Digitonin was from Wako Biochemicals (Richmond, VA). Antibodies to human albumin were from Roche Molecular Biochemicals. Reagents for electron microscopy were obtained from Electron Microscopy Sciences (Fort Washington, PA). All other reagents were from Sigma.
Analysis of Protein Synthesis Inhibitor Activity-Cells were plated at 1 ϫ 10 6 cells/well on 6-well plates and used at 70% confluency. For pulse labeling studies, cells were first starved of methionine by incubation in serum-and methionine-free DMEM for 20 min at 37°C. Pulse labeling was subsequently initiated by replacement of the media with serumfree and methionine-free DMEM supplemented with 100 Ci/ml 35 Slabeled Pro-Mix. Where indicated, protein synthesis inhibitors were added directly to the pulse medium to final concentrations of 200 M cycloheximide, 200 M puromycin, or 0.2 M pactamycin. After the indicated chase periods, cells were lysed in phosphate-buffered saline containing 20 mM CHAPS and 10 mM EDTA and centrifuged (15 min, 15,000 ϫ g) to remove insoluble cellular debris. Albumin was immunoprecipitated from the cell lysate supernatant and media fractions with rabbit anti-human albumin antibody and protein A-conjugated Sepharose beads. Immunoprecipitates were processed for SDS-PAGE and separated on 12.5% gels. The incorporation of radiolabel into albumin, expressed in PSL units, was determined by phosphorimager analysis with a MacBAS-1000 PhosphorImager system (Fuji Medical Systems, Stamford, CT).
Electron Microscopy-HepG2 and BRL cells were plated on plastic coverslips in multiwell plates and grown to 70% confluency. Alterations in protein synthesis activity were elicited by addition of cycloheximide, puromycin, or pactamycin to the medium and incubation for 10 min at 37°C. The medium was then removed and replaced with ice-cold 0.1 M sodium cacodylate buffer containing 3% glutaraldehyde and 0.05% tannic acid. The samples were then fixed in 1% osmium tetroxide, stained in 2% uranyl acetate, dehydrated in a graded alcohol series, and embedded in Araldite 506 epoxy resin. 60 -80-nm ultrathin sections were cut with a diamond knife on an ultramicrotome, collected on coated copper grids, and stained with permanganate and lead citrate (12). Grids were viewed at 60 kV in a transmission electron microscope at magnifications of 15,000 -120,000 (JEOL, Peabody, MA).
For negative staining, BRL cells were grown in normal media and treated with cycloheximide, pactamycin, or puromycin as described above. After harvesting with trypsin/EDTA, the cells were washed and resuspended to 5 ϫ 10 6 cells/ml in hypotonic buffer containing 50 mM Tris-HCl, pH 7.4, 10 mM KCl, and 5 mM MgCl 2 (TKM). Swelled cells were homogenized in a Dounce-type homogenizer until ϳ95% rupture was achieved, as assayed by trypan blue staining. A post-mitochondrial supernatant was prepared by centrifugation of the homogenate for 15 min at 10,000 ϫ g. A microsome fraction was then prepared from the post-mitochondrial supernatant by centrifugation for 15 min at 100,000 ϫ g in a Beckman TLA 100.3 rotor. The microsome fraction was resuspended in TKM containing 110 mM KCl and 250 mM sucrose and solubilized by addition of Nikkol to 1% (v:v) and deoxycholic acid to 0.5% (w:v). The solubilized membrane fraction was placed over a 1.5 M sucrose cushion, and the ribosome fraction collected by centrifugation for 1 h at 80,000 rpm in a Beckman TLA 100.2 rotor. The ribosome pellet was resuspended in TKM containing 110 mM KCl. Rat liver microsomes prepared by the method of Gaetani et al. (13) were processed in the identical manner. Each ribosome preparation was diluted to 1 A 260 unit/ml in TKM and supplemented with 2 mM octyl glucoside. A 15-l drop of each sample was placed atop a carbon-coated copper grid for 1 min at room temperature. Grids were washed twice with TKM and stained with 2% uranyl acetate for 10 s. Dry grids were viewed immediately after drying or stored at room temperature.
Morphometry-High magnification TEM images were recorded on film with a microscope-mounted plate camera. The resulting negatives were digitized with a CCD camera and imported into Adobe Photoshop version 4.0. Digital images were adjusted for contrast and brightness and then cropped to remove identification marks. Each image was randomly assigned a code to reduce the possibility of operator bias during subsequent analysis. Morphometric analysis of the images was conducted with ObjectImage software (University of Amsterdam, Netherlands). For the analysis, a transparent digital overlay was made for each image. The overlay was then manually marked with segmented line traces along transversely sectioned ER membrane bilayers. Ribosomes intersecting the segmented lines were individually marked and counted as point particles, such that each line and its associated particles comprised one case. 697 cases and a total of 2170 ribosomes were measured from 37 micrographs representing the four experimental conditions. To compile the raw data, the micrographs were decoded, and the cases were grouped by micrograph and by experimental condition. A value representing the number of ribosomes per linear micron of membrane was calculated for all cases in each micrograph. An average value for the number of ribosomes per linear micron was then obtained for each condition. For statistical analysis of the number of membranebound ribosomes, analyses of variance and Student's t tests were performed on data sets of size n for each condition, where n was the number of micrographs. p values were assigned to each of the data sets from inhibitor-treated cells based on the T statistic, where p Յ 0.05 indicates a statistically significant difference between the data sets from inhibitor-treated and untreated cells.
Cell Fractionation-BRL cells were fractionated by sequential detergent extraction as follows. Cells at 70 -80% confluency were harvested by trypsin/EDTA treatment and centrifugation. The cells were washed and resuspended in phosphate-buffered saline at 4°C to a concentration of 2 ϫ 10 6 cells/ml. One-ml aliquots of cell suspension were centrifuged at 1000 ϫ g for 3 min and resuspended in a cytosol buffer containing 110 mM KOAc, 25 mM K-HEPES, pH 7.5, 2.5 mM Mg(OAc) 2 , 1 mM EGTA, and 2 mM dithiothreitol. Digitonin was added to final concentrations of 10 -60 g/ml. After 5 min of gentle agitation on ice, the cell suspension was centrifuged at 1000 ϫ g for 3 min. The supernatant was removed, and the cell pellet was washed twice in digitoninsupplemented buffer to yield a permeabilized cell fraction that was enriched in membrane-bound polysomes and depleted of free and cytoskeleton-associated polysomes. To release membrane-bound polysomes, the extracted cell pellet was then resuspended in a buffer containing 1% (v/v) Nikkol, 0.5% (w/v) deoxycholic acid, 110 mM KOAc, 25 mM K-HEPES, pH 7.5, and 5 mM Mg(OAc) 2 . After 2 min of gentle agitation the cell suspension was centrifuged at 3000 ϫ g for 5 min, and the supernatant fraction was collected.
Ribosome Detection-To radiolabel ribosomes, growth media for BRL cells at 35% confluency were replaced with fresh media supplemented with 10 Ci/ml of 5,6-[ 3 H]uridine, and the cells were cultured in the presence of [ 3 H]uridine for 18 h. Where indicated, cycloheximide, puromycin, or pactamycin was added 10 min prior to processing. For determination of total ribosome content in subcellular fractions, equivalent samples were placed atop a cushion of 1.5 M sucrose in buffer containing 110 mM KOAc, 25 mM K-HEPES, pH 7.5, and 2.5 mM Mg(OAc) 2 . The samples were centrifuged for 1 h at 80,000 rpm in a Beckman TLA 100.2 rotor, and the ribosome-enriched pellet fractions were resuspended in 5% SDS for determination of associated radioactivity. For the resolution of ribosomes by subunit distribution, 200 l of the digitonin soluble fraction containing free ribosomes or the Nikkol/DOC-soluble fraction containing the membrane-bound ribosomes were loaded onto continuous 0.5-1.5 or 0.3-0.9 M sucrose gradients containing 400 mM KOAc, 50 mM Mg(OAc) 2 , and 25 mM K-HEPES, pH 7.5. The gradients were centrifuged for 3 h at 35,000 rpm (4°C) in the Beckman SW41 rotor. 0.4-ml gradient fractions were collected manually by tube puncture and drop collection. The [ 3 H]uridine content in duplicates of each fraction was determined in a liquid scintillation spectrometer. Samples were counted twice within 24 h, and an average value for [ 3 H]uridine content was calculated for each sample.

Characterization of Experimental
System-To study the temporal coupling of ribosome exchange and protein translation on the ER membrane, tissue culture cells were treated with protein synthesis inhibitors selective for either the initiation or elongation cycle of protein translation, and ribosome-membrane interactions were assayed by morphometric and biochemical methods. Three protein synthesis inhibitors were investigated as follows: cycloheximide, puromycin, and pactamycin. The mode of action, site of action, and the predicted effects of each inhibitor on polysome structure are listed in Table I.
Because tissue culture cell lines often display differences in their sensitivity to protein synthesis inhibitors, each inhibitor was assayed to identify effective concentrations and to screen for secondary effects on protein secretion. In these experiments, methionine-starved HepG2 cells were pulsed with [ 35 S]methionine in the presence or absence of inhibitor, and the rate of isotope incorporation into albumin was determined by immunoprecipitation, SDS-PAGE, and PhosphorImager analysis. As shown in Fig. 1A, in the absence of inhibitor, [ 35 S]methionine incorporation increased linearly over the time course of the assay (7.5 min). At the concentrations used, cycloheximide and puromycin rapidly and effectively inhibited incorporation of [ 35 S]methionine into albumin. In the presence of pactamycin, [ 35 S]methionine incorporation into albumin proceeded linearly for a short time and then ceased. For pactamycin, the pattern of isotope incorporation was as predicted for an inhibitor of initiation; only those nascent chains undergoing elongation at the time of pactamycin addition were completed, and subsequent de novo protein synthesis was blocked. An additional series of controls was performed to identify any secondary effects of the protein synthesis inhibitors on protein secretion. In these experiments, pulse-chase studies of albumin synthesis and secretion were performed in the presence of each of the inhibitors. The kinetics of albumin secretion in the presence of either cycloheximide, puromycin, or pactamycin were essentially identical, with a half-time for secretion of approximately 20 -30 min (Fig. 1B). In summary, the three inhibitors used in this study disrupt the protein synthesis activity of HepG2 cells in the manner predicted from their established mode of action. Furthermore, these data demonstrate that the experimental manipulations necessary to evaluate the fate of ribosomal subunits arising from premature (puromycin) or natural (pactamycin) termination do not adversely alter ER function or the activity of the protein secretion machinery.
Ribosomes Remain Membrane-bound after Termination, Morphometric Analysis-To characterize any changes in ribosome-membrane interaction that might accompany the termination of protein synthesis, a series of ultrastructural and biochemical studies were conducted with cultured cells. In experiments on rough ER ultrastructure, HepG2 and BRL cell monolayers were treated with the described protein synthesis inhibitors for 10 min. This time period is sufficient to allow complete inhibition of protein synthesis and, in the case of puromycin and pactamycin, the discharge of nascent chains from the ribosome. Samples were then placed on ice and fixed with glutaraldehyde. Ultrathin sections were prepared, stained, and viewed by electron microscopy. A representative section from untreated HepG2 cells is illustrated in Fig. 2A. In this micrograph an abundant and extensive ER network can be readily distinguished from cytosol by the relatively dark staining of the membrane-bound ribosomes and the granular, lumenal reticuloplasm. The ER cisternae are seen in both trans-verse transversely sectioned and obliquely sectioned membrane sheets. Two high magnification views of polyribosomes on the rough ER (Fig. 2, B and C) illustrate the resolution observed in oblique and transverse sections. In the oblique view, a polysome is visible in the commonly observed rosette pattern (14). The spatial orientation of ribosomal subunits can also be discerned in these panels, where large subunits are oriented toward the outside of the polysome. Small subunits appear atop the large subunits and face the polysome interior.
The morphological analysis of ribosome-membrane interactions following translation inhibition was restricted to transverse views of the ER membrane. Extensive sets of micrographs were prepared for cells that had been treated with cycloheximide, pactamycin, and puromycin, as well as untreated cells. Representative micrographs of the kind used in the analysis are depicted in Fig. 3A. A total magnification of  approximately 100,000 afforded clear views of individual ribosomes along the length of transversely sectioned ER in untreated cells (Fig. 3A). The ribosomes are clustered in groups of 3-5 along some regions, which indicates the association of these ribosomes in polysomes at the time of fixation. Although it was not possible to consistently resolve individual subunits at this magnification, the asymmetric positive staining of the membrane-bound particles suggests that both large and small subunits are bound in untreated cells. A similar distribution of ribosomes along the ER membrane was observed in cells after treatment with cycloheximide (not shown). Visual inspection of transverse ER sections from pactamycin-and puromycin-treated BRL cells revealed an extensive number of membrane-bound ribosomes, as shown in Fig. 3, B and C. This finding was surprising in light of the activity of the inhibitors and the predicted behavior of ribosomes upon termination. Following run-off translation or premature termination, the ribosomes would be expected to dissociate into subunits and discharge from the membrane. However, we observed a distribution of membrane-bound ribosomes which approximated that seen in untreated and cycloheximide-treated cells, where the majority of membrane-bound ribosomes contained translocating nascent chains. Along the plane of the membrane, the ribosomes of pactamycin-and puromycin-treated cells did not have the same degree of lateral organization found in control cells (Fig. 3A). A more random distribution of ribosomes in pactamycin or puromycin-treated cells may be attributed to the loss of polysome structure. Thus, following pactamycin or puromycin treatment, no rosette or linear polysome patterns in the cytoplasm or on the rough ER membrane were observed (data not shown).
To evaluate quantitatively the distribution of bound ribosomes, micrographs of cells from all experimental conditions were digitized and adjusted to uniform contrast and brightness levels (see "Experimental Procedures"). Because of the small diameter of ribosomes relative to resin sections and the propensity of polysome-associated ribosomes to extend above and below the plane of section, it is difficult to quantify the number of ribosomes present per unit area of membrane in an oblique section (15,16). However, transverse sections in which the membrane-bound ribosomes are of similar dimension can be quantitated and the relative ribosome density expressed as ribosomes per linear micron. At least 430 individual data points were collected for each treatment from 7 or more unique micrographs. Differences between results from untreated cells and inhibited cells were assessed using the Student's t test. The results of the analysis are depicted in Table II. Following elongation arrest with cycloheximide, ribosomes were present at a density of 12.17 ribosomes per linear micron of ER membrane. This number was not statistically different from the density of ribosomes measured in untreated cells, indicating that no net release or addition of ribosomes occurred upon incubation with cycloheximide. Following treatment with puromycin, the density of bound ribosomes present in cross-section was 8.24 ribosomes/m, a decrease of 35% compared with control. A similar decrease was observed for cells treated with pactamycin, where 8.03 ribosomes/m were present. Statistical analysis of the data indicate that these differences from control cells are significant, with p values of 0.0001. It is apparent from these results that following release of the nascent chain from the ribosome upon either premature termination or run-off translation, approximately one-third of the ribosomes completely detach from the ER, whereas the majority remain in stable association.
Biochemical Fractionation of Free and Membrane-Bound Ribosomes-Morphometric analysis of the ribosome-membrane association in intact cells indicated that ribosomes and/or ribosomal subunits are found in association with the ER membrane following inhibitor-elicited termination. However, in traditional thin sections, it is difficult to distinguish unequivocally large and small ribosomal subunits. To gain insight into the structural composition of the bound ribosomal particles re- maining upon termination, inhibitor-treated cells were fractionated and ribosomal subunit identity determined by velocity sedimentation on sucrose gradients. A method was developed to allow rapid separation of membrane-bound and free ribosomes in BRL cells, based on earlier fractionation procedures with this cell line (17,18). Cell monolayers were cultured for 18 h with [ 3 H]uridine to radiolabel RNA. Semi-intact cells devoid of cytoplasm were subsequently generated by digitoninbased permeabilization in isotonic buffer. The digitonin-extracted cells were then solubilized with nonionic detergent, to release membrane-bound ribosomes, and centrifuged to remove insoluble components. Membrane-derived ribosomes were resolved by centrifugation through continuous sucrose gradients or discontinuous sucrose cushions.
The digitonin concentration was optimized to allow efficient removal of cytosolic ribosomes without affecting ER membrane integrity. As shown in Fig. 4, 40 g/ml digitonin was the minimum detergent concentration necessary to elicit selective plasma membrane permeabilization. Under these conditions, cytoplasmic radiolabeled RNA species were removed with two washes in digitonin-supplemented isotonic buffer. The radiolabeled RNA remaining in the cell was not susceptible to further extraction at 4°C and thus was operationally defined as membrane-associated. Integral ER membrane proteins were observed to remain in quantitative association with the cell pellet (data not shown).
Large Ribosomal Subunits Remain Bound after Termination-The digitonin-based cell fractionation procedure was used to determine biochemically the state of ribosome assembly following the termination of protein synthesis. In these experiments, [ 3 H]uridine-labeled cells were incubated with inhibitor-supplemented media for 10 min at 37°C. For each analysis, 2 ϫ 10 6 cells were then processed to yield the cytosol and membrane-bound ribosome fractions. Ribosomes present in these extracts were separated by velocity sedimentation on sucrose gradients, and fractions from the gradients were analyzed for [ 3 H]uridine content by liquid scintillation spectrometry. A representative set of experiments is shown in Fig. 5A, which depicts ribosome migration profiles for the cytosol fraction of inhibitor-treated cells. To identify the peaks in the upper portion of the gradient, the positions of purified rat liver 60 S ribosomal subunits and 80 S ribosomes were determined in parallel gradients (not shown). By comparison with these standards, the three major [ 3 H]uridine-labeled peaks represent, respectively, 60 S, 80 S, and polysomal fractions. As predicted for the inhibitors used in this study, cycloheximidetreated ribosomes maintain polysome structure, whereas pactamycin-and puromycin-treated polysomes dissociate into monomers and subunits. Monomer dissociation was accompanied by a compensatory increase in the size of the 80 S and 60 S peaks (Fig. 5A). Small subunits were not well resolved under these gradient conditions.
The migration profiles of the membrane-associated ribosome fraction (Fig. 5, B and C) differ markedly from those of the cytosolic fraction. Consistent with the findings from the morphometric analysis, we observed that ribosomal subunits remained in association with the ER membrane in the absence of nascent polypeptide chains and polysomal structure. In the presence of pactamycin or puromycin, the dominant radiolabeled, membrane-bound species is the large ribosomal subunit. 40 S subunits were present only at sub-stoichiometric levels (Fig. 5C). 80 S ribosomes are nearly absent on the membrane, indicating that termination and subsequent dissociation of the large and small subunits had occurred. In contrast, the membrane-bound ribosomes from cycloheximide-treated cells have not undergone termination and thus remain predominantly in polysomes. These results indicate that after nascent chain discharge, small ribosomal subunits are preferentially released from the ER membrane, whereas large subunits remain bound. Ultrastructural Analysis of Membrane-bound Ribosomal Subunits-Biochemical studies with permeabilized cells showed that large subunits remain bound to the ER membrane after termination, whereas small subunits were preferentially released. To evaluate further these findings, a series of experiments was conducted to image the membrane-associated particles after inhibitor treatment. BRL cells were treated with cycloheximide, pactamycin, or puromycin, and a microsomal fraction was prepared. After solubilization of the microsome fraction and centrifugation, the soluble microsomal contents were adsorbed onto copper grids, negatively stained with ura-

FIG. 4. Isolation of membrane-bound ribosomes from tissue culture cells. BRL cells were labeled with [ 3 H]uridine and harvested
at 75% confluency. Cells were washed and resuspended to 2 ϫ 10 6 cells/ml in isotonic buffer and subsequently treated with digitonin at the indicated concentration for 5 min at 4°C. Soluble cellular material was then removed following brief centrifugation, and the cell pellet was subjected to a series of washes in detergent-supplemented isotonic buffer. To recover the membrane-bound ribosomes, cell pellets remaining after the final wash were solubilized in a buffer containing 1% Nikkol and 0.5% deoxycholic acid, and remaining nuclei and/or nuclear fragments were removed by centrifugation. Cytosolic ribosomes released upon digitonin extraction, and membrane-bound ribosomes released upon detergent solubilization were recovered by sedimentation through 1.5 M sucrose cushions, and ribosomal RNA was quantitated by scintillation counting.

TABLE II
Morphometric analysis of ribosome-membrane interactions following premature or natural termination Micrographs of inhibitor-treated or control BRL cells were digitized and measured as described under "Experimental Procedures." Membranebound ribosomes along each segment of transversely sectioned rough ER were counted individually. Data for number of bound ribosomes per linear micron of membrane were compiled by treatment. n, number of micrographs in set; c.i., 95% confidence interval for variation from mean. p Յ0.05 indicates a statistically significant difference between data sets from untreated and experimental populations. nyl acetate, and viewed by transmission electron microscopy. Images of membrane-derived ribosomal subunits after experimental treatments were compared with purified subunits, and polysomes from rat liver microsomes were prepared in the same manner. Determination of ribosome structure after negative staining was based on previously published results (19,20). Images of negatively stained small subunits, large subunits, and 80 S ribosomes from rat liver are shown in Fig. 6. 40 S subunits (Fig. 6A) are characterized by an elongated profile with one or two stain-accessible regions that traverse the subunit and are sometimes accompanied by indentations along the sides (see Ref. 19, plates I and II). 60 S subunits (Fig. 6B) are generally round and may also appear "skiff-shaped." Their most striking characteristic is the appearance of one or more small protrusions on one side of the subunit in the skiff orientation. In this view, the opposite side is demarcated by an even deposit of stain around the periphery of the subunit (see Ref. 19, plates III and IV). Monomeric ribosomes and polysomes are noticeably larger than either 40 S or 60 S subunits (Fig. 6C).
They are identifiable by their distinct bilobal appearance, wherein a deposit of stain is visible at the large subunit-small subunit interface. The interface is visible along the width of the structure or as a dense spot on one side (see Ref. 19, plates IX and XI).
The composition of membrane-derived subunits from inhibitor-treated cells was identified on the basis of these standards. Composites of these subunits are shown in Fig. 7. As expected, treatment with cycloheximide yields polysomes and 80 S monosomes (Fig. 7A). Negatively stained intact polysomes were not as numerous as we predicted from sucrose gradient centrifugation; this is likely due to degradation during the homogenization and isolation procedure. Ribosomes from cycloheximidetreated BRL membranes appear very similar to those from untreated rat liver microsomes. The monomers are asymmetric and contain a stain-filled groove between large and small subunits in the frontal orientation.
In contrast, negatively stained particles from pactamycinand puromycin-treated BRL cell membranes (Fig. 7, B and C) strongly resemble purified rat liver large subunits. They share common features with large subunits from both our preparation and that described in Ref. 19, including a round appearance and small protuberances on one side marked by heavier stain deposits in the skiff orientation. The particles are also similar in diameter to purified 60 S subunits and noticeably larger than 40 S subunits (Fig. 6A), while lacking the large spot of stain that demarcates intact monomers. We conclude that the distribution and structure of these membrane-bound ribosomal particles after premature or natural termination in vivo is consistent with our biochemical observations of large subunit retention and small subunit dissociation. DISCUSSION We report that the termination of protein synthesis on membrane-bound ribosomes in vivo results in the enhanced release Rat liver rough microsomes were prepared by homogenization and centrifugation according to standard protocols. Microsomes were solubilized in 1% Nikkol, and the soluble ribosomes were isolated by centrifugation through sucrose cushions. Ribosomal subunits were separated by treatment with a puromycin-high salt treatment followed by sucrose gradient centrifugation. Preparations containing 40 S subunits, 60 S subunits, and 80 S ribosomes were adsorbed onto carbon-coated copper grids without fixation. The grids were stained with uranyl acetate before viewing in a transmission electron microscope at 100,000 -150,000-fold magnification. Representative particles were selected from micrographs of each fraction. A, purified 40 S subunits; B, purified 60 S subunits; C, microsomederived 80 S monomers and polysomes (lower left).
of small ribosomal subunits from the membrane-bound pool; large ribosomal subunits remain in stable association with the ER membrane. Although these data do not support the proposal that the termination of protein synthesis on membranebound ribosomes yields the release of large and small ribosomal subunits into the free, cytoplasmic pool, they are consistent with prior data demonstrating that, in vitro, the release of nascent chains from membrane ribosomes results in the enhanced exchange of small but not large ribosomal subunits (21).
From a historical perspective, the question of whether membrane-bound ribosomes participate in a translation-dependent exchange with free ribosomes closely followed the observation that free, cytosolic ribosomal subunits associate and dissociate coincident with the initiation and termination stages of translation (22,23). Since it had been established that ribosomes were segregated between free and membrane-bound pools, studies were thus performed to determine whether such ribosome pools were kinetically exchangeable (6,9,11,21,24). Two experimental approaches were taken. In one approach, radioisotope incorporation studies were performed to determine the rate of entry of ribosomal subunits into the free and membranebound pools of intact cells. In the study of Baglioni et al. (9), the primary conclusion was that large ribosomal subunits bind to the ER membrane, and polysome assembly occurs as small subunit initiation complexes joined with membrane-bound large subunits to yield translationally active 80 S ribosomes. This conclusion was disputed by Mechler and Vassalli (6,11,24) in a detailed and rigorous series of investigations. These authors concluded that the kinetics of isotope incorporation into large ribosomal subunits of the free and bound ribosomal pools was such that the formation of 80 S ribosomes must occur in the free ribosome pool. Mechler and Vassalli (11) did report, however, that the kinetics of membrane dissociation of large and small ribosomal subunits was different, with large ribosomal subunit release lagging that of small subunit release. In contrast to the above referenced studies, which were performed in intact cells, Borgese et al. (21) examined ribosomal subunit exchange in vitro and reported that following the release of nascent chains, ribosomal subunit exchange was strictly limited to small subunits; large subunits remained in stable association with the microsomal membranes. Because it could not be unequivocally concluded that the in vitro system faithfully executed the termination reaction as seen in vivo, and it was known that free and membrane-bound ribosomal subunits were structurally and metabolically similar, conclusions regarding the physiological significance of an inexchangeable large ribosomal subunit pool were appropriately conservative (21).
Although the isotope incorporation studies provided valuable data demonstrating that free and membrane-bound ribosomes existed in metabolic equilibrium, the experimental approach is hampered by a significant kinetic constraint. That is, the rates of ribosomal subunit biosynthesis and nuclear export are such that isotope incorporation rates must be followed for periods of hours (6,9,11). Protein synthesis, however, occurs in the time frame of seconds to minutes, and thus the relevant exchange kinetics differ by 1-2 orders of magnitude. With this experimental limitation in mind, we re-examined the fate of membrane-bound ribosomes following the termination of protein synthesis through a combined morphometric and biochemical analysis of ribosome-membrane interactions in intact cells. The results of our studies best support the hypothesis that the termination of protein synthesis on membrane-bound ribosomes results in the free exchange of small ribosomal subunits, with the large ribosomal subunits remaining in stable association with the ER membrane. These data thus confirm and extend the conclusions obtained by Borgese et al. (21) and demonstrate that in the intact cell, large ribosomal subunit release from the ER membrane does not occur coincident with the termination of protein synthesis.
What is the fate of membrane-bound large ribosomal subunits after termination? Should such subunits be competent for protein translation, it is likely that protein synthesis could be initiated on the endoplasmic reticulum membrane. Assuming this to be true, it then becomes necessary to determine whether membrane-bound ribosomes can select mRNA substrates and thus whether membrane-bound ribosomes can catalyze the synthesis of free, cytosolic proteins. Furthermore, as membrane-bound ribosomes are thought to reside in intimate association with the protein conducting channel component of the ER translocon, it is equally important that the compartmental fate of such translation products be determined. As it is known that membrane-bound large ribosomal subunits exchange with the cytoplasmic pool, it is essential that the mechanism of large subunit release be determined and the factors governing this release process be identified. Insights into these questions are presented in the accompanying manuscript (31). for 15 min before harvesting. Cells were washed, resuspended in hypotonic buffer, and ruptured with a Dounce-type homogenizer. A rough microsome fraction was prepared by centrifugation on a discontinuous sucrose gradient and solubilized in nonionic detergent. The membranederived ribosomal particles were adsorbed onto EM grids and processed as described in the legend to Fig. 6. Representative particles were selected from micrographs of each fraction.