Ribosome-independent Regulation of Translocon Composition and Sec61α Conformation*

In this study, the contributions of membrane-bound ribosomes to the regulation of endoplasmic reticulum translocon composition and Sec61α conformation were examined. Following solubilization of rough microsomes (RM) with digitonin, ribosomes co-sedimented in complexes containing the translocon proteins Sec61α, ribophorin I, and TRAPα, and endoplasmic reticulum phospholipids. Complexes of similar composition were identified in digitonin extracts of ribosome-free membranes, indicating that the ribosome does not define the composition of the digitonin-soluble translocon. Whereas in digitonin solution a highly electrostatic ribosome-translocon junction is observed, no stable interactions between ribosomes and Sec61α, ribophorin I, or TRAPα were observed following solubilization of RM with lipid-derived detergents at physiological salt concentrations. Sec61α was found to exist in at least two conformational states, as defined by mild proteolysis. A protease-resistant form was observed in RM and detergent-solubilized RM. Removal of peripheral proteins and ribosomes markedly enhanced the sensitivity of Sec61α to proteolysis, yet the readdition of inactive ribosomes to salt-washed membranes yielded only modest reductions in protease sensitivity. Addition of sublytic concentrations of detergents to salt-washed RM markedly decreased the protease sensitivity of Sec61α, indicating that a protease-resistant conformation of Sec61α can be conferred in a ribosome-independent manner.

Beginning with early morphological and biochemical studies defining the rough endoplasmic reticulum as the site of protein entry into the secretory pathway (1)(2)(3), an understanding of the mechanism of ribosome binding to the endoplasmic reticulum (ER) 1 membrane has been considered essential to the elucidation of the molecular mechanism of protein translocation. A theme consistent through the past decades of research into protein translocation has been the hypothesis that the membrane-bound ribosome performs regulatory functions that govern the structural and functional state of the translocon, the site of protein translocation in the ER (4 -9). In these models, the ribosome is thought to interact with protein components of the ER membrane and thereby initiate the assembly of a protein-conducting channel through which translocation proceeds (5-8, 10, 11). In addition to a role in the regulation of channel assembly, the ribosome is also thought to regulate the structural state of the protein-conducting channel during the translocation event itself and in this manner assist the assembly of proteins of complex topologies, such as apolipoprotein B-100, the prion protein, and polytopic membrane proteins (9,(12)(13)(14).
Diverse experimental approaches have provided evidence indicating a role for the ribosome in the regulation of translocon assembly. For example, following solubilization of rough microsomes (RM) with digitonin, ribosomes co-fractionate with Sec61␣, the core component of the protein-conducting channel (6,15). Furthermore, inactive ribosomes elicit Sec61p oligomerization in proteoliposomes comprising lipids and the purified Sec61p complex (7). In agreement with these observations, a direct physical interaction between inactive yeast ribosomes and yeast Sec61p was recently identified by cryo-electron microscopy (8). It remains uncertain, however, whether the Sec61p-ribosome interactions identified in these studies are identical to the ribosome-membrane junction as characterized in vitro (8, 16 -19).
Current experimental evidence supports the identification of Sec61␣ as the ribosome receptor, although substantial literature exists regarding the identification of ribosome receptor proteins other than Sec61␣ (20 -29). Surprisingly, although significant disagreement regarding the identification and characterization of ribosome receptor proteins persists, there is near unanimity in the choice of assay system used for their discovery. In this assay, radiolabeled biosynthetically inactive ribosomes are incubated with ribosome-stripped microsomal vesicles and the bound and free fractions are separated by flotation centrifugation (20). This assay identifies saturable and high affinity ribosome binding, although at physiological salt concentrations the observed ribosome binding stoichiometries are approximately 10% of those found in native membranes (20,28). For this reason, it is uncertain whether the ribosome binding activity observed in this assay is identical to that present during ribosome/nascent chain targeting and translocation. In fact, the hypothesis that ER membrane components other than Sec61␣ contribute to ribosome binding was presented in a recent study demonstrating the binding of ribosomes bearing nascent secretory chains to sites on the ER membrane other than Sec61␣ (30).
We report that solubilization of RM with digitonin yields macromolecular complexes comprising ribosomes, a diverse population of integral membrane proteins, and phospholipids. Such complexes were also observed following solubilization of ribosome-free membranes, indicating that the ribosome is not required for complex stability in detergent solution. In contrast to the results seen with digitonin, solubilization of RM with lysophosphatidylcholine or 1,2-diheptanoyl-sn-phosphatidylcholine (DHPC) yields monodisperse solutions of RM proteins that did not reside in stable association with ribosomes. In studies designed to assess the role of Sec61␣ in the regulation of ribosome interactions with the ER membrane, Sec61␣ was observed to exist in two conformational states. In native RM and detergent extracts of native RM, Sec61␣ assumed a protease-resistant conformation. Removal of bound ribosomes and peripheral proteins yielded a dramatic increase in the sensitivity of Sec61␣ to proteolysis; readdition of inactive ribosomes did not efficiently restore the protease-resistant conformation. However, the addition of sublytic concentrations of detergents to ribosome-depleted membranes markedly enhanced the resistance of Sec61␣ to proteolysis, indicating that Sec61␣ conformation can be regulated in a ribosome-independent manner.
Analysis of Protein and Lipid Association with Ribosomes-Detergent solubilization of microsomal vesicles was performed as follows: 10 equivalents (eq; defined in Ref. 31) of RM, or an equal quantity of ribosome-stripped membranes, were suspended in buffer containing 20 mM detergent (unless otherwise specified), 25 mM K-HEPES (pH 7.2), 150 mM KOAc, 5 mM Mg(OAc) 2 , 1 mM DTT, and 0.5 mM PMSF in a final volume of 200 l. Samples were incubated on ice 30 min, and ribosomes were pelleted by centrifugation for 30 min at 80,000 rpm in a TLA100 rotor at 4°C (247,000 ϫ g at r av ; k factor ϭ 10). Supernatant fractions were precipitated by addition of trichloroacetic acid to 10%, incubation on ice for 20 min, and centrifugation for 15 min at 15,000 rpm in a refrigerated microcentrifuge. Pellets and precipitated supernatants were then processed for SDS-PAGE. SDS-PAGE gels were transferred to nitrocellulose membranes using a semidry transfer apparatus (Bio-Rad) and a transfer buffer consisting of 50 mM CAPS (pH 11.0), 20% methanol, and 0.075% SDS. Immunoblots were visualized using Super-Signal detection reagents (Pierce), and images were quantified using NIH Image software.
Velocity sedimentation studies of ribosome-associated membrane proteins were performed by the following procedure; 30 eq of RM were suspended in buffer containing 20 mM detergent, 25 mM K-HEPES (pH 7.2), 150 mM KOAc, 5 mM Mg(OAc) 2 , 1 mM DTT, and 0.5 mM PMSF in a final volume of 600 l. Samples were incubated on ice for 30 min, and then loaded onto 10-ml 10 -30% sucrose gradients containing 25 mM K-HEPES (pH 7.2), 150 mM KOAc, 5 mM Mg(OAc) 2 , 1 mM DTT, and 0.1% detergent. Gradients were centrifuged for 2 h at 40,000 rpm in an SW40 rotor (Beckman) at 4°C. Gradients were manually fractionated by puncturing the tube bottoms, and fractions were precipitated by addition of trichloroacetic acid to 10%. Samples were then processed for immunoblot analysis as described above. To study interactions between detergent-solubilized membrane proteins, the following modifications were made; following solubilization of RM or ribosome-stripped membranes, samples were centrifuged at 80,000 rpm for 30 min in a TLA100.2 rotor at 4°C (228,000 ϫ g at r av ; k factor ϭ 18). Supernatants were then loaded onto 10-ml 5-20% sucrose cushions containing 25 mM K-HEPES (pH 7.2), 150 mM KOAc, 5 mM Mg(OAc) 2 , 1 mM DTT, and 0.1% detergent. Gradients were centrifuged for 24 h at 40,000 rpm in an SW40 rotor at 4°C and processed as described above.
To study the association of bulk ER lipids with ribosomes, samples were solubilized and fractionated through 10 -30% sucrose gradients as described above. The ribosome-enriched fractions were pooled and diluted 4-fold with 25 mM K-HEPES (pH 7.2), 150 mM KOAc, 5 mM Mg(OAc) 2 , and 1 mM DTT. Ribosomes were pelleted by centrifugation for 4.5 h at 45,000 rpm in a Ti50.2 rotor (Beckman) at 4°C. Ribosomes were then resuspended in ribosome buffer, and ribosome-associated lipids were extracted and analyzed as described below.
Reconstitution of Vesicles from DHPC-solubilized RM Extracts-1000 eq of RM were pelleted and resuspended in 3 ml of 25 mM K-HEPES (pH 7.2), 150 mM KOAc, 5 mM Mg(OAc) 2 , 1 mM DTT, 0.5 mM PMSF, 10% glycerol, and DHPC yielding a detergent:RM ratio of approximately 400:1 (nanomoles of detergent:eq of RM). Samples were incubated on ice for 30 min and then centrifuged for 30 min at 80,000 rpm in the TLA100.2 rotor at 4°C. Supernatant samples were adjusted to 600 mM KOAc and mixed with SM-2 Biobeads (Bio-Rad) that had been preequilibrated with 25 mM K-HEPES (pH 7.2), 600 mM KOAc, and 10% glycerol. Samples were incubated with shaking overnight at 4°C. The fluid phase was then removed from the beads, and the reconstituted vesicles were pelleted by centrifugation for 10 min at 60,000 rpm in a TLA100.2 rotor at 4°C. Vesicles were resuspended in 0.25 M sucrose, 25 mM K-HEPES (pH 7.2), 50 mM KOAc, and 1 mM DTT and stored at Ϫ80°C. The amounts of membrane proteins present in the vesicles, relative to RM, were determined by immunoblotting of serial dilutions of reconstituted vesicles and RM.
Ribosome Binding Assays-To analyze the binding of ribosomes to reconstituted vesicles, 2 pmol of purified ribosomes were combined with increasing quantities of vesicles in 50 l of buffer consisting of 25 mM K-HEPES (pH 7.2), 150 mM KOAc, 5 mM Mg(OAc) 2 , 1 mM DTT, and 0.5 mM PMSF. Samples were incubated on ice for 30 min, and Nycodenz was then added to a final concentration of 45%, to yield a final volume of 1 ml while maintaining constant ionic conditions. The 45% Nycodenzribosome binding reaction was overlaid with 600 l of 37.5% Nycodenz, 400 l of 30% Nycodenz and 100 l of 0% Nycodenz, all in 25 mM K-HEPES (pH 7.2), 150 mM KOAc, 5 mM Mg(OAc) 2 . Samples were centrifuged for 3 h at 55,000 rpm in an SW55 rotor (Beckman) at 4°C. The upper 600 l of the step gradients, which contained the floated vesicles, was collected and precipitated by addition of trichloroacetic acid to 10%. Samples were processed for SDS-PAGE and immunoblot analysis as described above.
Proteolysis of Intact and Detergent-solubilized Membranes-Membranes were left untreated or solubilized with digitonin or DHPC as described above, and chymotrypsin was added to the concentrations indicated in the figures. Digestions were performed for 30 min on ice, after which time trichloroacetic acid was added to 10%. Samples were collected by centrifugation and processed for SDS-PAGE and immunoblotting. When SDS was used to solubilize membranes, the above procedures were carried out at 37°C to avoid precipitation of the SDS. In experiments involving ribosome binding to membranes prior to proteolysis, 10 pmol of ribosomes were incubated with 2 eq of PKRM for 30 min at 4°C in buffer containing 150 mM KOAc, 25 mM K-HEPES (pH 7.2), 5 mM Mg(OAc) 2 , and 1 mM DTT.
Protein Translation Activity of Detergent-isolated Ribosomes-2 pmol of ribosomes, derived from either reticulocyte lysate or RM solubilized with 20 mM DHPC, were added to 8 l of a rabbit reticulocyte lysate post-ribosomal supernatant supplemented with 500 ng preprolactin mRNA, 16 Ci of [ 35 S]Pro-Mix, 0.05 unit/ml RNasin, 1 mM DTT, and 20 M amino acid mix (minus methionine) in a total volume of 20 l. Following incubation for the indicated time periods at 25°C, 5 l of the translation reaction was spotted onto filter paper and protein synthesis levels were measured as described previously (33).
Lipid Extraction and Analysis-Samples were mixed with 4 volumes of a mixture of methanol/chloroform/HCl (100:50:1). After extraction at room temperature for 20 min, phase separation was induced by addition of 1.3 volumes each of chloroform and 2 M KCl. Samples were centrifuged for 2 min at 13,000 rpm in a microcentrifuge. The upper aqueous phase was removed, and the lower organic phase was washed twice with a mixture of methanol/water/HCl (50:50:1). The organic phase was reduced to dryness by vacuum centrifugation and resuspended in chloroform/methanol (2:1). Bulk ER lipids were resolved on silica thin layer chromatography (TLC) plates using a solvent system of chloroform/ methanol/acetic acid/water (100:30:35:3) and were visualized by staining with iodine vapor.
Antibodies-Anti-peptide rabbit polyclonal antisera against Sec61␣ and ␤, TRAP␣ and ␤, ribophorin I, and signal peptidase, as well as chicken antisera (IgY) directed against ribosomal proteins L3/L4, were used at 1:1500 dilution, as described previously (30). Following transfer to nitrocellulose, membranes were blocked in phosphate-buffered saline containing 5% nonfat milk and 0.1% Tween 20 for 1 h at room temperature. Primary incubations were performed for 1 h at room temperature in phosphate-buffered saline containing 1% nonfat milk and 0.1% Tween 20, and secondary incubations performed for 30 min in the identical buffer using the appropriate secondary antibodies at 1:3000 dilution.

Composition of Detergent-soluble Ribosome-Translocon
Complexes-To investigate the role of membrane-bound ribosomes in the regulation of translocon composition, native RM were solubilized with either digitonin or DHPC, and ribosome-associated membrane components were resolved by sedimentation of the detergent-soluble fraction on sucrose gradients. The selection of detergents was based on published studies identifying either ribosome-membrane protein complexes (digitonin; Ref. 6) or high recovery of native protein structure and activity (DHPC; Ref. 34). As depicted in Fig. 1 (A and C), when RM were solubilized with digitonin, the ribosome fraction sediments as a heterogeneous fraction containing Sec61␣, ribophorin I, TRAP␣, and ER phospholipids. Quantitation of the ribosomeassociated lipid fraction by chemical phosphate analysis indicated that phospholipids were present at approximately a 50fold molar excess to ribosomes (data not shown). When RM were solubilized with DHPC, none of the assayed membrane components (Sec61␣, ribophorin I, TRAP␣, TRAM, or signal peptidase complex) co-sedimented with ribosomes ( Fig. 1B). In addition, although phospholipids could be identified with the DHPC-solubilized ribosomes (Fig. 1C), the quantities of ribosome-associated phospholipids were approximately 2% of that seen with digitonin-soluble ribosomes.
Translocon Structure Persists in the Absence of Bound Ribosomes-The data presented in Fig. 1 indicated that, when RM were solubilized with digitonin, membrane-bound ribosomes sedimented in association with a large, heterogeneous complex of integral membrane proteins and phospholipids. When RM were solubilized with DHPC, however, stable ribosome-membrane protein interactions were not detected. To further investigate these differences, RM were stripped of bound ribosomes in puromycin/high salt buffers (6) and experiments similar to those depicted in Fig. 1 were performed (Fig. 2). In the absence of bound ribosomes, the digitonin-soluble membrane proteins would be predicted to sediment with a considerably smaller sedimentation coefficient than that depicted in Fig. 1. Thus, to adequately resolve the detergent-soluble protein fraction derived from ribosome-free membranes, the detergent extracts were centrifuged on sucrose gradients for extended time periods. As shown in Fig. 2A, when ribosome-free membranes were solubilized with digitonin, Sec61␣, ribophorin I, and TRAP␣ were again observed to migrate as a complex. Analysis of the total protein composition of these fractions indicated that the digitonin-soluble translocon complex was heterogeneous and consisted of at least 20 silver stain-reactive polypeptides ( Fig.   FIG. 1. Association of rough microsome components with ribosomes in detergent solution. 30 eq of RM were solubilized at 150 mM KOAc with 20 mM digitonin (A) or DHPC (B) and the soluble fraction subjected to velocity sedimentation in 10 -30% sucrose gradients. Gradient fractions were collected and processed for immunoblot analysis with antibodies against the indicated proteins. The immunoblot for L3/L4 indicates the position of ribosomes within the gradient. In panel C, the ribosome peak from each gradient was collected and the ribosomes recovered by centrifugation. Following organic extraction of the pelleted ribosomes, the phospholipid composition of the organic phase was determined by thin layer chromatography and detection by iodine staining.

FIG. 2. Ribosome-independent macromolecular complexes of ER membrane proteins exist in digitonin solution, but not in DHPC solution.
In panel A, a digitonin-soluble PKRM extract was resolved on a 5-20% sucrose gradient. Fractions were collected and subjected to immunoblot analysis for the indicated proteins. In panel B, a digitonin-soluble PKRM extract was resolved as in A, gradient fractions were collected, the proteins were concentrated by trichloroacetic acid precipitation, and following resolution on a 12.5% SDS-PAGE gel, the protein components were identified by silver stain detection. A digital image of the silver-stained gel is shown. In panel C, a DHPCsoluble RM extract was processed as in panel A, except smaller fraction sizes were collected to obtain higher resolution. 2B). In contrast, when RM were solubilized with DHPC, Sec61␣, ribophorin I, and TRAP␣ exhibited overlapping, but non-identical sedimentation profiles (Fig. 2C). From these experiments, it can be concluded that the isolation of digitoninsoluble translocon complexes does not require bound ribosomes. Furthermore, when RM were solubilized with DHPC, the integral membrane protein components of the ER behaved in a monodisperse manner and no stable membrane proteinribosome interactions were observed.
In current views, ribosome binding to the ER membrane is maintained primarily through interactions with Sec61␣ (7,28). The data depicted in Figs. 1 and 2 support this view, although the molecular heterogeneity of the digitonin-soluble translocon complexes, as seen in both the presence and absence of bound ribosomes, prevents unequivocal identification of Sec61␣ as the ribosome receptor. Alternatively, the results obtained with DHPC can be interpreted to indicate that neither Sec61␣, ribophorin I, nor TRAP␣ alone constitute the primary site of ribosome-membrane interaction. To reconcile these conflicting interpretations, additional experiments were performed to critically assess the significance of the DHPC data and to further evaluate ribosome-Sec61␣ interactions.
Effects of DHPC on Oligomeric Protein Structure, Ribosome Binding, and Ribosome Function-DHPC is well documented to solubilize a diverse array of integral membrane proteins while maintaining both native structure and enzymatic activity (34,35). Nonetheless, the inability to detect stable translocon protein-ribosome interactions in DHPC solution could result from DHPC-induced disruptions in the structural integrity of ER oligomeric protein complexes, inactivation of ribosome binding activity, and/or DHPC-induced structural perturbations in the ribosome itself. To determine whether the inability to detect ribosome-membrane protein interactions in DHPC solution was a consequence of general disruptions in oligomeric protein integrity, the DHPC-soluble fraction from RM was fractionated by velocity sedimentation and the oligomeric structure of the Sec61, TRAP, and signal peptidase complexes determined. As shown in Fig. 3, the ␣ and ␤ subunits of Sec61 undergo partial dissociation in DHPC solution, whereas other membrane protein complexes, such as the TRAP and signal peptidase complexes, remain wholly intact. From these data, it can be concluded that DHPC does not elicit a general disrup-tion of oligomeric membrane protein complexes. Furthermore, because it has been demonstrated that the ␣ and ␤ subunits of Sec61 readily dissociate in detergent solution and that the ␤ subunit of Sec61 is not required for ribosome binding (36), the partial dissociation of the ␣ and ␤ subunits of Sec61 observed in DHPC solution does not provide a convincing explanation for the lack of stable membrane protein-ribosome interactions in DHPC-solubilized membranes.
The ribosome-Sec61␣ interaction, as assayed in digitonin solution, is highly electrostatic and is stable at salt concentrations approaching 1 M (6). Consistent with this observation, when native RM were solubilized in digitonin at 0.5 M salt and centrifuged, Sec61␣ was recovered in the ribosome-rich pellet fraction, whereas when ribosome-stripped membranes were solubilized with digitonin, Sec61␣ was recovered in the supernatant fraction (Fig. 4A, lanes 5 and 6). In contrast, solubilization of native RM with either DHPC or lysophosphatidylcholine at physiological salt concentrations yielded efficient recovery of Sec61␣ in the supernatant fraction (Fig. 4B, lanes 3-6) and, as illustrated in Fig. 4C, the DHPC-soluble Sec61␣ was not associated with ribosomes. This phenomenon is further illustrated in Fig. 4D, where the ability of the two detergents to release Sec61␣ from native RM was assayed as a function of detergent concentration. As is clear from the data in Fig. 4D, whereas solubilization of Sec61␣ with DHPC exhibits a clear maxima at a DHPC concentration of 16 -24 mM (detergent:RM of 400 -600 nmol detergent:eq RM, with 1 eq RM containing ϳ4 nmol of phospholipid), solubilization of Sec61␣ with increasing concentrations of digitonin does not alter the distribution of Sec61␣ into large (Ͼ100 S) complexes that are recovered in a ribosomerich pellet upon centrifugation. Thus, if ribosome association with the ER membrane is conferred via electrostatic interactions with Sec61␣, such interactions are disrupted in the presence of DHPC. Alternatively, ribosome binding to the ER membrane may require multiple weak interactions with ER membrane components and, under conditions in which the ER membrane is efficiently solubilized (addition of DHPC or lysophosphatidylcholine), weak bi-molecular interactions between ribosomes and individual membrane components are not captured.
To determine if the lack of stable ribosome-membrane protein interactions seen in DHPC solution were due to disruption of the presumed electrostatic ribosome-membrane interaction by the choline headgroup of DHPC, RM were treated with 0.5 M choline and ribosome release was assayed. Under these conditions no ribosome release was observed (data not shown). In additional experiments, RM were solubilized with digitonin and choline was subsequently added to 0.5 M. Again, ribosomemembrane component interactions remained intact (data not shown). These data indicate that the inability to detect stable ribosome-Sec61␣ interactions following solubilization with DHPC was not solely due to the choline headgroup, and, because lysophosphatidylcholine treatment of RM also yielded soluble, ribosome-free Sec61␣ at moderate salt concentrations, not a unique consequence of the relatively short (7 carbon) acyl chains present on DHPC.
To directly evaluate the effect of DHPC treatment on the ribosome binding activity of microsomal membranes, the following experiments were performed. In the first series of experiments, proteoliposomes were reconstituted from DHPCsolubilized RM components and their ribosome binding activity determined. Reconstituted vesicles were incubated with ribosomes at physiological salt concentrations, and the membranebound ribosomes separated by gradient flotation on Nycodenz step gradients. Bound ribosomes were then quantified by immunoblotting for the large ribosomal subunit proteins L3/L4. At maximal binding stoichiometries, 2 pmol of ribosomes were bound to reconstituted proteoliposomes containing the amount of Sec61␣ found in 3-4 eq of RM (Fig. 5A). Native RM contain approximately 1 pmol of bound ribosomes/eq (data not shown), and so proteoliposomes prepared from DHPC extracts of RM display approximately 50% of the ribosome binding activity observed in native membranes. A 50% relative binding stoichiometry is consistent with a stochastic reconstitution of Sec61␣ in the proper and inverse topologies. Thus, solubilization of RM in DHPC and subsequent reconstitution of the detergent ex-tract into proteoliposomes does not substantially alter the binding of inactive ribosomes.
To assess whether DHPC elicited an alteration in ribosome structure, as reflected in translation activity, the effects of DHPC on protein synthesis were examined. In these experiments, a post-ribosomal supernatant of reticulocyte lysate was supplemented with identical quantities of ribosomes obtained either by centrifugation of DHPC-solubilized RM, or resuspension of the reticulocyte ribosomes remaining from preparation of the lysate post-ribosomal supernatant. The capacity of the ribosome-supplemented post-ribosomal supernatant to translate preprolactin was then assayed. As shown in Fig. 5B, the translation activity of the DHPC-treated ribosomes was virtually identical to that seen with the reticulocyte ribosomes, and thus DHPC does not alter ribosome structure in a manner that alters ribosome function. Furthermore, ribosomes recovered from DHPC-solubilized RM are capable of binding to PKRM in a manner similar to that depicted in Fig. 5A (data not shown). Fig. 5 support the conclusion that solubilization of RM with DHPC preserves inactive ribosome binding activity upon reconstitution, it was necessary to address the hypothesis that Sec61␣, in DHPC solution but not in digitonin solution, assumed a conformation unsuitable for ribosome binding. To test this hypothesis, experiments were performed to examine the conformation of Sec61␣ in the detergent-solubilized state, using the proteol- In panel A, 10 eq of RM or EKRM were solubilized with 20 mM digitonin at the indicated KOAc concentrations. Ribosomes were pelleted by centrifugation at 80,000 rpm for 30 min in the TLA100 rotor. The pellets and supernatants were then subjected to immunoblot analysis with an anti-Sec61␣ antibody. In panel B, 10 eq of RM were solubilized with 20 mM lysophosphatidylcholine or DHPC, and were processed as in panel A. In panel C, 10 eq of RM were solubilized with 20 mM DHPC, and were processed as in panel A, with the exception that antibodies recognizing ribosomal proteins L3/L4 and S9 were used in immunoblot analysis. A similar distribution of ribosomal components was observed when RM were solubilized with digitonin or lysophosphatidylcholine. In panel D, the efficiency of Sec61␣ solubilization from RM was assayed as a function of detergent concentration at physiological salt concentrations. Note that RM contain ϳ4 nmol of phospholipid/eq.

FIG. 5. Effects of DHPC on ribosome binding and protein translation activity.
In panel A, vesicles containing the indicated amounts (eq) of Sec61␣ were incubated with 2 pmol of ribosomes and were then subjected to flotation through Nycodenz cushions as described under "Experimental Procedures." Vesicles were harvested and subjected to immunoblot analysis with antibodies recognizing the large ribosomal subunit proteins L3/L4 (lanes 3-11). As controls, unfloated vesicles were immunoblotted with the anti-L3/L4 antibody to confirm the absence of ribosomal contamination (lane 1), and 2 pmol of ribosomes were immunoblotted with the anti-L3/L4 antibody to indicate the total amount of ribosomes present in the assay (lane 2). In panel B, reticulocyte ribosomes (q), ribosomes obtained after solubilization of RM with 20 mM DHPC at 150 mM KOAc (E), or a control sample lacking ribosomes (ϫ) were used in cell-free translation assays, as described under "Experimental Procedures." Following translation of full-length preprolactin mRNA for the indicated time periods, samples were processed to determine the amount of radioactive methionine incorporated into trichloroacetic acid-insoluble protein.
ysis pattern of Sec61␣ as an index of conformation. To generate a conformation-sensitive proteolysis profile for Sec61␣, RM were treated with increasing concentrations of SDS at 37°C and subjected to chymotrypsin digestion. As shown in Fig. 6A, in the absence of SDS, Sec61␣ is relatively insensitive to proteolysis by chymotrypsin, with the majority of the protein being recovered as the full-length form and a minor fraction being recovered as a faster migrating (ϳ28 kDa) form, indicated by the double asterisk (Fig. 6A, lane 1). At low concentrations of SDS, the conformation of Sec61␣ was significantly altered. Under these conditions, the N-terminal epitope recognized by the Sec61␣ antisera was degraded and thus no limit digestion products were detected (Fig. 6A, lanes 2-5). However, at higher concentrations of SDS, where the chymotrypsin itself becomes inactivated, the faster migrating limit digestion product was readily detected (Fig. 6A, lanes 6 -8). At the highest SDS concentration tested, in which the protease was completely inactivated, Sec61␣ was recovered largely in the intact form (Fig. 6A, lane 9).
Similar experiments were then conducted to compare the Sec61␣ digestion pattern obtained in native and ribosomestripped RM, prior and subsequent to solubilization with digitonin or DHPC (Fig. 6B). As demonstrated previously, in native RM, Sec61␣ was nearly completely insensitive to digestion by exogenous proteases (Fig. 6, A, lane 1; B, lanes 1 and 2). In light of the observation that ribosomes remain bound to Sec61␣ following solubilization of RM with digitonin, the digitoninsoluble Sec61␣ fraction was predicted to be insensitive to protease digestion, and this was observed (Fig. 6B, lanes 3 and 4). On the basis of the data presented in Fig. 4, we postulated that in a DHPC extract of RM, Sec61␣, which is not in association with ribosomes, would be protease-sensitive. This prediction was not met. Rather, the proteolysis pattern of Sec61␣ in DHPC solution was identical to that observed in digitonin solution (Fig. 6B, lanes 5 and 6 versus lanes 3 and 4). Thus, when solubilized from RM, Sec61␣ exists in a protease-resistant conformation regardless of whether it is (digitonin solution), or is not (DHPC solution), associated with ribosomes.
These data indicate that the protease resistance of Sec61␣ in native membranes cannot be attributed solely to bound ribosomes and furthermore in native RM Sec61␣ exists in at least two conformations, which differ in their relative sensitivity to chymotrypsin digestion.
The protease sensitivities of Sec61␣ in detergent extracts of ribosome-stripped membranes (PKRM) were identical (Fig. 6B,  lanes 9 -12), yet distinct from those observed in intact PKRM (Fig. 6B, lanes 7 and 8). In PKRM, virtually the entire population of Sec61␣ had been converted to the chymotrypsin sensitive form, and was recovered as the 28-kDa form (Fig. 6B, lanes  7 and 8). In contrast, when PKRM were solubilized with either digitonin or DHPC, a third limit digestion product was identified. In this conformational state, Sec61␣ was cleaved to yield a form of intermediate mobility to the full-length and 28-kDa forms, and is indicated by the single asterisk (Fig. 6B). Because the antisera used in these experiments was raised against an N-terminal epitope, the mobility of this form on SDS-PAGE is consistent with proteolysis occurring in a domain of Sec61␣ C-terminal to that generating the 28-kDa limit digestion product. As will be demonstrated in Fig. 7, this digestion product is likely an intermediate in the formation of the 28-kDa form. In summary, the data presented in Fig. 6 indicate that: 1) the conformation of Sec61␣ in digitonin solution is similar to that observed in DHPC solution; 2) removal of bound ribosomes and peripheral proteins elicits a conformational change in Sec61␣; 3) the inability of proteases to cleave Sec61␣ in native membranes cannot be attributed solely to bound ribosomes, and 4) Sec61␣ can exist in at least two conformations, both of which are present in native RM.
Previous studies demonstrating that the addition of inactive ribosomes to PKRM elicits a protease-insensitive conformation of Sec61␣ support the conclusion that ribosome binding to Sec61␣ is the primary determinant governing protease accessibility (28). As the data presented in Fig. 6 indicate that the conformational status of Sec61␣ can be altered in a ribosomeindependent manner, the effects of ribosome addition on the conformation of Sec61␣ were investigated. In these experiments, protease titrations were performed on RM, PKRM, and PKRM to which excess inactive ribosomes were added (Fig. 7A). Comparison of the proteolysis patterns indicates that, as demonstrated previously, Sec61␣ in RM exists in two conformational states, with the predominant form being resistant to chymotrypsin cleavage (Fig. 7A, lanes 1-5). Addition of increasing concentrations of chymotrypsin to PKRM results in the appearance of two forms: one of intermediate mobility, which is identical to that seen upon digestion of detergent-solubilized PKRM (Fig. 6B, lanes 10 and 12, single asterisk), and one of higher mobility (Fig. 6, A and B, double asterisk), which is of identical mobility to that seen in low abundance in native RM (Fig. 7A, lanes 6 -10). When chymotrypsin titrations were performed on PKRM supplemented with excess ribosomes, the overall pattern of digestion was similar to that seen in the absence of ribosomes (Fig. 7A, lanes 7-10 versus lanes 12-15). Comparison of the proteolysis patterns at intermediate concentrations of chymotrypsin did indicate, however, a modest influence of the inactive ribosomes on the conformation of Sec61␣, primarily with regard to the intermediate digestion product seen most readily in detergent solubilized PKRM (Fig. 7A,  compare lanes 8 and 9 with lanes 13 and 14). Clearly, however, the conformation of Sec61␣ in PKRM supplemented with a 5-fold molar excess of free ribosomes is markedly different from that present in RM. Additional experiments provided evidence for a ribosome-independent regulation of Sec61␣ conformation. As shown in Fig. 7B, following treatment of PKRM with a sublytic concentration of DHPC, Sec61␣ was markedly resist- In panel A, 10 eq of RM were solubilized with the indicated concentrations of SDS for 30 min at 37°C. Chymotrypsin was then added to a final concentration of 25 g/ml, and digestion occurred for 30 min at 37°C. Digestion was stopped by addition of trichloroacetic acid to 10%. Samples were then processed for immunoblot analysis with an anti-Sec61␣ antibody. In panel B, for lanes 1 and 2 and lanes 7 and 8, 10 eq of RM or PKRM were either untreated or subjected to proteolysis at 4°C for 30 min with chymotrypsin at a final concentration of 25 g/ml. For lanes 3-6 and 9 -12, RM or PKRM were first solubilized with 20 mM digitonin or DHPC at 150 mM KOAc, and were then subjected to proteolysis as described above. Relative molecular weights are indicated to the left of the panels. Sec61␣ limit digestion products are indicated by single and double asterisks.
ant to proteolytic digestion (Fig. 7B, lanes 1-7 versus lanes  8 -14). At the concentration of DHPC used in these experiments (0.75 mM), integral membrane proteins, such as Sec61␣, TRAM, and the signal peptidase complex were not solubilized (data not shown), and lumenal proteins such as BiP were not released from the membrane vesicles (Fig. 7C). Because the protease accessibility characteristics of Sec61␣ as seen in intact RM were displayed upon addition of sublytic concentrations of detergent to ribosome stripped membranes, it appears that low concentrations of detergent either alter membrane structure in a manner that effects a change in Sec61␣ conformation and/or alter protein-protein interactions that contribute to the regulation of Sec61␣ conformation. The observed DHPC-dependent changes in Sec61␣ protease accessibility could also be obtained by addition of a structurally unrelated detergent (BigCHAP) or a compound that preferentially distributes in the outer membrane leaflet (dinitrophenol), indicating that this phenomenon is not unique to DHPC (data not shown). These data clearly demonstrate that the conformation of Sec61␣, as displayed by limited proteolysis, can be modulated in a ribosome-independent manner. DISCUSSION In this study, the role of membrane-bound ribosomes in the regulation of the macromolecular organization of the endoplasmic reticulum translocon and the conformation of Sec61␣ were investigated. Six conclusions can be drawn from these studies. 1) In digitonin solution, the translocon is a heterogeneous complex consisting of multiple integral membrane proteins, phospholipids and ribosomes. 2) The molecular composition of such complexes is ribosome-independent. 3) When solubilized under conditions in which ER membrane proteins are recovered in monodisperse solution, membrane protein-ribosome interactions are not maintained. 4) Sec61␣ can be found in at least two conformational states. 5) The proteolytic sensitivity of Sec61␣ is enhanced upon removal of peripheral proteins and bound ribosomes. 6) Conversion of Sec61␣ to a protease-resistant conformation, as is predominantly found in native RM, can be elicited by sublytic concentrations of detergent and with very modest efficiency by inactive ribosomes.
With regard to the role of the ribosome in regulating translocon structure, it was observed that following solubilization of RM with digitonin, ribosomes remained in association with large macromolecular complexes containing Sec61␣, ribophorin I, TRAP␣, and approximately a 50-fold molar excess of phospholipid to ribosomes. Related studies by Chevet et al. (37) indicate that similar detergent-derived complexes also contain calnexin and the ER-to-Golgi recycling protein ␣2p24. The isolation of such macromolecular complexes is in agreement with previous studies (48), and was not dependent on bound ribosomes, as complexes of similar composition were obtained following solubilization of ribosome-free microsomes (PKRM) with digitonin. Because the digitonin-soluble translocon complexes are heterogeneous and apparently not dependent upon the ribosome for their higher order structure, it is difficult to determine whether the ribosome binding activity displayed by such complexes reflects the activity of any single component of the complex, as would be expected from published data (6,8,15,28,38).
Previous studies have postulated that the interaction between the Sec61 complex and the ribosome occurs through stable electrostatic interactions (6). In support of this hypothesis, the release of Sec61␣ from digitonin-solubilized ribosome pellets required salt concentrations in excess of 1 M (6). However, when RM were solubilized with DHPC or lysophosphatidylcholine, no stable interactions between ribosomes and Sec61␣ were observed. Rather, following solubilization in DHPC at physiological salt concentrations, resident integral membrane proteins behaved in a monodisperse manner and were readily resolved from membrane-derived ribosomes. This observation was unexpected and thus necessitated a series of control experiments to evaluate the effects of DHPC treatment on the structure and activity of the two assumed binding partners, ribosomes and Sec61␣. In these experiments, it was demonstrated that DHPC did not elicit a general disruption of membrane protein quaternary structure, that ribosome binding activity could be efficiently recovered in proteoliposomes reconstituted from DHPC extracts of RM, that the protease accessibility of Sec61␣ in either digitonin or DHPC extracts was similar, and that DHPC treatment did not alter the protein translation activity or membrane binding activity of ribosomes. Solubilization of RM with DHPC does not, therefore, appear to alter the structural or functional characteristics of the ER components that are presumed responsible for ribosome binding. It should be noted, however, that although extensive control experiments indicated that solubilization of RM with DHPC did not irreversibly affect ribosome binding, we cannot eliminate the possibility that DHPC elicits a heretofore uncharacterized conformational change in Sec61␣ that disrupts Sec61␣-ribosome interactions. Equally as important, the substantial molecular heterogeneity of the digitonin-soluble translocon complexes and the observation that translocon composition in digitonin is ribosome-independent makes the identification of the ribosome receptor(s) in digitonin extracts of RM difficult.
To reconcile our experimental findings with previous studies, we propose that the differences in the mechanism of membrane solubilization displayed by the two detergents determine whether ribosome association with the translocon can be captured. It is clear that digitonin and DHPC solubilize biological membranes by different mechanisms (39 -43). Digitonin displays numerous structural similarities with the bile acid class of detergents. Principally, digitonin is a planar, heterocyclic nonionic detergent containing a polyglycidic side chain. Previous studies have shown that detergents of this structural class act by releasing membrane fragments enclosed at their periphery with detergent molecules (44,45). These studies also noted that solubilization of membranes with such detergents preserves the structural organization of the bilayer in the mixed micelle (44,45). In contrast to digitonin, DHPC is a short-chain phosphatidylcholine derivative, which, because of its pronounced structural similarities to native membrane lipids and solubility in aqueous media, efficiently disperses biological membranes (34,46). In so doing, DHPC is remarkable in its ability to preserve native membrane protein structure and function in the solubilized state (34,35). With these differences in solubilization mechanism in mind, the data presented herein can be interpreted to indicate that digitonin treatment of native or stripped RM yields the release of large macromolecular complexes constituting a defined mixed detergent/protein/lipid domain. DHPC, because it is not subject to the solubilization constraints imposed by the structural characteristics of heterocyclic planar detergents such as digitonin, efficiently disperses the protein and lipid components of the membrane bilayer (34). Thus, whereas in digitonin the ribosome can maintain multiple contacts with the membrane components that comprise the digitonin-soluble membrane domain, in DHPC such membrane domains are not captured and thus stable ribosome-membrane protein interactions are not maintained. On the basis of these data, we postulate that membrane-bound ribosomes maintain their association with the ER membrane through multiple, weak interactions, rather than via a high affinity bimolecular interaction. In other words, the kinetic basis for ribosome binding to protein components of the ER membrane is an avidity rather than an affinity-based process. The hypothesis that ribosome binding to the ER membrane is regulated through interactions involving multiple membrane components is supported by reconstitution studies, where it was observed that proteoliposomes containing the minimal translocation machinery display approximately 10% of the ribosome binding observed in native membranes, and by studies of the binding of biosynthetically active ribosome nascent chain complexes, where ribosome binding to sites other than Sec61␣ was reported (28,30). Proteoliposomes containing the SRP receptor and Sec61␣ function in translocation, and so Sec61␣ likely figures prominently in the regulation of ribosome binding (15).
As an experimental approach, detergent-based co-fractionation studies did not provide unequivocal data identifying a stable, stoichiometric association between an ER translocon component(s) and membrane-bound ribosomes, and so additional criteria were examined. As reported by Kalies et al. (28), evidence for the identification of Sec61␣ as the ribosome receptor was obtained in studies demonstrating that Sec61␣ is protected from proteolysis by membrane-bound ribosomes, and that the addition of inactive ribosomes to ribosome-stripped microsomes restores the protease resistance of Sec61␣. To further examine this conclusion, the protease sensitivity of Sec61␣ was compared in native membranes, native membranes solubilized with digitonin (where a stable ribosome-membrane junction is maintained), and native membranes solubilized with DHPC (where the ribosome-membrane junction is lost). Contrary to prediction, the protease sensitivity of Sec61␣ did not increase following solubilization with DHPC. If native membranes were stripped of bound ribosomes with high salt and puromycin prior to solubilization and proteolysis, however, Sec61␣ was readily accessible to chymotrypsin digestion. These data indicate that the conditions used to extract membranebound ribosomes from ER microsomes alter Sec61␣ protease accessibility, but as the data obtained with DHPC extracts depicts, the protease-resistant conformation seen in native membranes need not be conferred by the bound ribosome. Furthermore, when ribosome-stripped membranes are supplemented with free ribosomes, only a very modest restoration of protease resistance is observed. From these data, it appears that Sec61␣ can assume at least two conformations and that the removal of peripheral proteins and bound ribosomes alters Sec61␣ conformation in a manner that is only inefficiently reversed by the addition of excess inactive ribosomes. It follows that the binding of inactive ribosomes to ribosome-stripped membranes is mechanistically different from that occurring during the physiological targeting event and so inactive ribosomes, should they bind to Sec61␣, do not protect it from proteolysis.
Because the protease-resistant conformation of Sec61␣ could be efficiently elicited by addition of sublytic concentrations of detergent (Fig. 7) or the amphiphile dinitrophenol (data not shown), we speculate that the differences in Sec61␣ protease accessibility seen following the extraction of peripheral proteins represent, at least in part, a conformational change in the protein. The observation that Sec61␣ conformation is sensitive to the addition of sublytic concentrations of amphiphiles suggests that the physical state of the membrane can influence Sec61␣ conformation. When viewed with respect to previous studies demonstrating that membrane-bound ribosomes increase the structural order parameters of microsomal membranes (47), these data identify a mechanism whereby membrane-bound ribosomes, by influencing the physical properties of the membrane, could elicit changes in Sec61␣ conformation in the absence of direct, stable interactions.