In vitro binding of ribosomes to the beta subunit of the Sec61p protein translocation complex.

The Sec61p complex forms the core element of the protein translocation complex (translocon) in the rough endoplasmic reticulum (rough ER) membrane. Translating or nontranslating ribosomes bind with high affinity to ER membranes that have been stripped of ribosomes or to liposomes containing purified Sec61p. Here we present evidence that the beta subunit of the complex (Sec61beta) makes contact with nontranslating ribosomes. A fusion protein containing the Sec61beta cytoplasmic domain (Sec61beta(c)) prevents the binding of ribosomes to stripped ER-derived membranes and also binds to ribosomes directly with an affinity close to the affinity of ribosomes for stripped ER-derived membranes. The ribosome binding activity of Sec61beta(c), like that of native ER membranes, is sensitive to high salt concentrations and is not based on an unspecific charge-dependent interaction of the relatively basic Sec61beta(c) domain with ribosomal RNA. Like stripped ER membranes, the Sec61beta(c) sequence binds to large ribosomal subunits in preference over small subunits. Previous studies have shown that Sec61beta is inessential for ribosome binding and protein translocation, but translocation is impaired by the absence of Sec61beta, and it has been proposed that Sec61beta assists in the insertion of nascent proteins into the translocation pore. Our results suggest a physical interaction of the ribosome itself with Sec61beta; this may normally occur alongside interactions between the ribosome and other elements of Sec61p, or it may represent one stage in a temporal sequence of binding.

The cotranslational translocation of proteins across the endoplasmic reticulum (ER) 1 membrane of eukaryotes involves the binding of translating ribosomes to receptor sites on the ER membrane, where the nascent protein enters the ER lumen through a proteinaceous pore. Translocation is preceded by a targeting phase in which ribosomes bearing pre-secretory nascent chains are selectively recruited to the ER membrane through the reciprocal action of at least two cytosolic factors, the signal recognition particle (SRP) (1) and the nascent polypeptide-associated complex (2). An interaction between SRP and its membrane-bound receptor (3,4) is followed by transfer of the ribosome to the translocation site (translocon) in the ER membrane, where the nascent peptide passes directly from an aqueous tunnel or channel within the ribosome to an aqueous pore within the membrane (5)(6)(7)(8).
Tranlocation has been reconstituted in vitro (18) with membranes containing Sec61p in addition to only two other components as follows: SRP receptor and the translocating chain associated membrane protein (TRAM) (19), whose presence does not appear to be required for the transport of all presecretory proteins (19,20). Photocross-linking experiments have shown that the nascent chain is adjacent to Sec61␣ throughout its transit of the ER membrane (21). Sec61p thus appears to form the hydrophilic translocation pore in the ER membrane.
The ribosome-ER interaction depends on contacts between the body of the ribosome and protein components of the ER membrane, in addition to the interaction of the nascent chain and the translocation pore. Treatment of rough microsomes (RM) with puromycin causes the expulsion of nascent chains into the ER lumen (22), but the release of ribosomes from the membrane requires additional exposure to high salt concentrations (23); nontranslating ribosomes can rebind to puromycin/high salt-stripped RM (PKRM) at physiological salt concentrations (24). Nontranslating ribosomes also bind with high affinity to artificial liposomes reconstituted with the Sec61p complex alone (25). Therefore, Sec61p is believed to serve as both a translocation pore and a ribosome-binding site. The isolated Sec61p complex forms a roughly pentagonal doughnut-shaped structure (17,26), whose assembly in lipid membranes is induced by the presence of ribosomes (17). The yeast Sec61p complex has been visualized in the ribosomebound state in detergent solution, where the central pore of the complex appears to align with the presumed nascent chain exit site on the large ribosomal subunit (26). In this reconstruction, the connection between the ribosome and Sec61p is restricted to a thin stalk, suggesting that other proteins may contribute to the tight ER-ribosome seal in vivo or that the actively translocating ribosome-ER complex assumes a different conformation.
The identity of the Sec61p components or domains involved in ribosome binding remains open to speculation. Previous results suggest that the ␤ subunit is the least essential for the overall function of the complex. Sec61␤ can be proteolytically degraded in rough microsomes without release of bound ribosomes, whereas Sec61␣ remains protected (25), and limited proteolysis of stripped RM has led to the identification of Sec61␣ domains that may be required for ribosome binding and translocation (48). Although yeast mutants lacking either the ␣ or the ␥ subunit are inviable (11)(12)(13), a deletion of both yeast Sec61␤ homologues (Sbh1p and Sbh2p) is not lethal and causes a reduction in growth rate only at elevated temperatures (27). Disruption of Sec61␤ in Drosophila, however, is lethal and causes embryonic defects consistent with impaired secretion (28). Sec61␤-depleted proteoliposomes bind ribosomes with an affinity comparable to that of proteoliposomes containing intact Sec61p (29). Sec61␤-depleted proteoliposomes showed impaired translocation in vitro, but the impairment could be overcome by allowing the SRP-dependent targeting step to occur at reduced temperature, in the absence of ongoing translation. It was proposed that Sec61␤ affects the kinetics of nascent chain insertion and that in the absence of Sec61␤ insertion proceeds at a reduced rate, allowing the competing process (elongation in the cytosol) to take precedence. Proteolytic digestion of Sec61␤ in stripped rough microsomes did not abolish their ability to translocate preprolactin in vitro (30). The present study addressed the physical interaction between ER membrane proteins and core components of the ribosome. We found that the purified Sec61␤ cytoplasmic domain (Sec61␤ c ) competes for ribosome binding to stripped ER membranes; Sec61␤ c also binds to ribosomes directly, with an affinity comparable to the affinity of ribosomes for stripped ER membranes. Based on these observations we propose that Sec61␤ may participate in ribosome binding and translocation by interacting directly with the body of the ribosome. This interaction may occur after, or concurrently with, an initial contact between the ribosome and other domains of Sec61p, and it does not preclude an interaction between Sec61␤ and the nascent chain.
Growth and Purification of GST Fusion Proteins-Plasmids pET28a-GST and pET28a-GST/Sec61␤ c were used to transform Escherichia coli strain BL-21(DE3), and the corresponding fusion proteins were expressed by induction of the liquid cultures with 1 mM isopropylthio-␤-D-galactoside for 1 h at 37°C. The fusion protein was purified using glutathione-agarose beads (Sigma) according to standard procedures (31) and stored at 4°C for up to 1 month. To elute GST fusion proteins, a volume of beads was transferred to a disposable column (Bio-Rad) and eluted with 5 mM glutathione, 50 mM Tris-HCl, pH 8.0. The protein concentration of each fraction was estimated by Bradford assay (32), using GST alone as a standard and using the value 1 A 280 ϭ 0.5 mg/ml to obtain the concentration of GST. Eluted fusion proteins were stored at Ϫ70°C.
Preparation of Ribosomes, Stripped Rough Microsomes, and Riboso-mal Subunits-Canine pancreatic RM were prepared as described (33). RM were stripped of ribosomes using puromycin/KCl (9), and the ribosomal and membranous components were collected separately and used in subsequent experiments. To prepare large and small ribosomal subunits, RM were incubated with puromycin/KCl as described (9), and 2-ml aliquots of the suspension were layered on continuous density gradients ( Pellets were resuspended in 0.25 M sucrose/RBB and stored at Ϫ70°C. Both the small subunit and the large subunit preparations had A 260 :A 280 ratios of 2.0. The purity of each preparation was assessed by Western blotting using antibodies against the ribosomal proteins S3 and L5. The small subunit preparation was virtually free of large subunits, whereas the large subunit preparation was slightly contaminated (20% or less) with small subunits.
Ribosomes were surface-labeled with t-butoxycarbonyl-L-[ 35 S]methionine N-hydroxysuccinimidyl ester ( 35 SLR; Amersham Pharmacia Biotech) as described previously (25). The labeled ribosomes were stored in 0.25 M sucrose/RBB and added directly to binding experiments. The specific activity of the ribosome preparation was typically about 5 ϫ 10 6 cpm per mg of ribosomes (34).
Assays for Ribosome Binding-Binding of ribosomes to puromycin/ KCl-stripped rough microsomes (PKRM) was measured according to an established method (24,35). Briefly, labeled ribosomes were mixed with 5-10 eq (33)  The top and bottom fractions were counted as bound and unbound, respectively, whereas the middle fraction, containing negligible radioactivity, was not included in subsequent calculations.
To test the effect of GST or Sec61␤ c /GST as competitors in the ribosome binding assay, the purified protein (after elution from glutathione-agarose; see above) was added to the binding assay in the amounts indicated in the figure legends. The fusion protein was mixed with labeled ribosomes before the addition of membranes. Compensating volumes of elution buffer (5 mM glutathione/50 mM Tris-HCl, pH 8.0) were added so that each 20-l reaction contained 2.5 l of fusion protein and/or elution buffer. The presence of elution buffer alone in this amount did not affect ribosome binding.
Direct binding of ribosomes to GST or Sec61␤ c /GST immobilized on glutathione-agarose beads was measured as follows. 20 l of beads with bound fusion protein were equilibrated in RBB and mixed with SLRlabeled ribosomes in a final volume of 0.1 ml. In some experiments additional agents were added as described in the figure legends. Each reaction was incubated on ice for 10 min with intermittent agitation to resuspend the beads; the samples were then incubated for 10 min on a rotating wheel at 4°C and stored on ice again for 10 min with intermittent agitation as before. 0.4 ml of RBB was added to each sample to improve the separation of bound and unbound material in the following step. The tubes were mixed once by inversion and immediately spun for 30 s in a microcentrifuge at room temperature. The supernatant was transferred to a new tube, and residual liquid was removed with a Hamilton syringe. Equivalent volumes of the bead fraction and the supernatant were measured by scintillation counting. For calculations, 0.1 ml was taken as the reaction volume. (In control experiments, addition of 0.4 ml of buffer at the end of the initial binding reaction caused a relatively slow re-equilibration of ribosomes that did not affect the calculated values substantially.)

RESULTS
The results presented here developed out of a screen for the ability of potential ribosome receptor domains to interfere with the in vitro binding of ribosomes to stripped rough microsomes when expressed as glutathione S-transferase fusion proteins. The Sec61␤ cytoplasmic domain ((Sec61␤ c ) was placed upstream of GST, with an intervening thrombin cleavage site (see "Experimental Procedures"). In the resulting fusion protein, the C-terminus of Sec61␤ c is fused to GST, whereas the N terminus of Sec61␤ c is free; this corresponds to the predicted disposition of the native Sec61␤, whose C terminus is thought to be anchored in the ER membrane (10). The fusion protein was expressed in bacteria and purified using glutathione-agarose beads as described under "Experimental Procedures." A construct encoding GST without the added Sec61␤ c sequence was used in parallel as a negative control. Fig. 1A shows the purified GST and Sec61␤ c/ GST with or without thrombin protease cleavage. The intact Sec61␤ c/ GST fusion protein without thrombin cleavage was used in all subsequent experiments presented here. After elution from glutathione-agarose, the GST and Sec61␤ c /GST fusion proteins, respectively, were mixed with labeled nontranslating ribosomes (ϳ100-fold molar excess of fusion protein over ribosomes) before addition of stripped rough microsomes. Ribosomes were chemically surface-labeled with sulfur labeling reagent (SLR); this did not measurably affect the membrane-binding properties of the ribosomes. The ribosome binding assay was carried out as described under "Experimental Procedures." The presence of GST alone in the reaction did not affect the extent of ribosome binding appreciably, compared with a sample in which no competitor was present (not shown). By contrast, the fraction of membrane-bound ribosomes in the Sec61␤ c /GST-containing sample was reduced to ϳ50% of the value obtained when GST alone was added as a competitor (Fig. 1B). To get an estimate of the amount of the fusion protein required to inhibit binding, we performed competition assays with increasing amounts of Sec61␤ c /GST. As shown in Fig. 1C, the inhibition of binding did not follow a linear relation to the amount of fusion protein added; up to a 16-fold excess of Sec61␤ c /GST over binding sites (25-fold excess over ribosomes) resulted in only a minor reduction in binding, whereas a 64-fold excess over binding sites caused a more substantial reduction (fraction bound ϭ 52% of the control value). GST normally dimerizes in solution (36), and under the conditions used in our ribosome binding experiments a significant portion of free GST formed aggregates large enough to be recovered by centrifugation at 100,000 ϫ g for 1 h (not shown). This behavior may account partially for the relatively large excess of free Sec61␤ c /GST needed to affect the binding of ribosomes to stripped rough microsomes.
Because the Sec61␤ c /GST fusion protein was able to prevent the binding of ribosomes to stripped rough microsomes, we suspected that the fusion protein might act by binding directly to ribosomes at the site where the native Sec61␤ normally binds. We therefore investigated the direct binding of labeled ribosomes to the GST fusion proteins. The Sec61␤ c /GST fusion protein was again expressed in bacteria and purified using glutathione-agarose beads, but rather than eluting the fusion protein from the beads (as in the competition experiment shown in Fig. 1), we tested the Sec61␤ c /GST-coated glutathione-agarose beads as a solid substrate for ribosome binding. By using different inputs of SLR-labeled ribosomes, we obtained a Scatchard plot for binding to Sec61␤ c /GST immobilized on glutathione-agarose, as shown in Fig. 2 (open squares). The affinity value (K D ) obtained by this method was 1.2 ϫ 10 Ϫ8 M, a value identical to that obtained by Borgese et al. (24) for the binding of tritium-labeled nontranslating ribosomes to stripped rough microsomes at low salt concentrations (25 mM KCl) and within the range of values reported by Kalies et al. (25) for the binding of SLR-labeled nontranslating ribosomes to stripped rough microsomes at 25 and 150 mM KCl, respectively. We emphasize that the very close correspondence between our affinity value for ribosome binding to Sec61␤ c /GST and measured values for ribosome binding to stripped rough microsomes may be fortuitous, given the inherent differences in the methods of measurement; however, we conclude that the affinity of ribosomes for the fusion protein is roughly on the same order as their affinity for stripped rough microsomes. Glutathione-agarose beads coated with GST alone (Fig. 2, closed squares) possessed a much lower ribosome binding activity; this binding was unspecific, as indicated by the flat profile of the Scatchard plot. Based on the X intercept of the Scatchard plot, we calculated that ϳ875 molecules of Sec61␤ c /GST are required per ribosome bound. This high figure may be accounted for by the additive effect of at least three considerations as follows: 1) the ribosome may have multiple binding sites for Sec61␤, as evidenced by the estimate of 3-4 Sec61p heterotrimers per translocon (17); 2) the large size of the ribosome (M r ϳ4.5 ϫ 10 6 ) relative to Sec61␤ c /GST (M r ϳ3.2 ϫ 10 4 ) means that a single bound ribosome may shield access to many Sec61␤ c /GST molecules on the beads that are not actively involved in binding; and 3) due to the porosity of the glutathione-agarose, a large fraction of the Sec61␤ c /GST may be bound in interior spaces that are not accessible to ribosomes.
The cytoplasmic domain of Sec61␤ has a predicted isoelectric point of 12.0, with ϳ10% basic residues and a predicted net charge of ϩ8 at neutral pH. This raised the possibility that the ribosome binding activity of Sec61␤ c /GST might result from a relatively unspecific charge-based interaction with ribosomal RNA. If this were the case, yeast transfer RNA would be expected to compete efficiently with labeled ribosomes for binding to Sec61␤ c /GST-coated glutathione-agarose beads. In Fig.  3, yeast transfer RNA was added to the direct binding assay at a molar excess of up to 2500-fold relative to inactive labeled ribosomes. Given that the ratio of the masses of total ribosomal RNA to tRNA (per mol) is ϳ125:1, an unspecific binding of ribosomal RNA should have been reduced 40 -50% by a 100fold molar excess of tRNA, even if all the ribosomal RNA was available for binding. In actuality a substantial amount of the ribosomal RNA is likely to be buried within the particle (37), so an even greater reduction of ribosome binding might be expected if the binding were simply charge-dependent. Instead, the reduction in the binding of labeled ribosomes to Sec61␤ c / GST does not approach 50% until a 2500-fold molar excess of tRNA is added (Fig. 3, left panel). Unlabeled ribosomes, however, compete efficiently (Fig. 3, right panel).
We also examined the salt dependence of ribosome binding to Sec61␤ c /GST. Stripped rough microsomes have a large ribosome binding capacity at low salt concentrations (25 mM KCl); the fraction of ribosomes bound in vitro falls off steeply as the salt concentration is increased to 0.5 M (24). However, the affinity of binding is higher at a more physiological salt concentration (150 mM KCl) than at 25 mM KCl (25). To examine the salt sensitivity of ribosome binding to the Sec61␤ c /GST fusion protein, we performed the direct binding assay described above, at a range of salt concentrations. As shown in Fig. 4, the ribosome binding activity of Sec61␤ c /GST (open squares) was stabilized by moderate salt concentrations and did not fall off appreciably unless the KOAc concentration was raised from 250 to 500 mM, a concentration at which the dissociation of nontranslating ribosomes from native rough microsomes is essentially complete (24). The base-line level of binding obtained with GST alone (Fig. 4, closed squares) was insensitive to the salt concentration. The fact that the binding of ribosomes to Sec61␤ c /GST, unlike the binding of ribosomes to stripped rough microsomes, is actually favored at moderate versus low salt concentrations, may be due to the absence, in the Sec61␤ c /GST agarose beads, of low affinity interactions that account for much of the binding to stripped rough microsomes at low salt concentrations (25).
Ribosome binding to the ER membrane is believed to occur largely through the large ribosomal subunit; in some reconstructions, however, the subunits lie side-by-side on the membrane, and an interaction between the small subunit and the membrane remains possible (38 -40). In the studies of Borgese et al. (24), isolated large ribosomal subunits are capable of re-binding in vitro to stripped rough microsomes; small subunits also bind but to a lesser extent. We wished to test whether the Sec61␤ c /GST fusion protein binds preferentially to the large ribosomal subunit, as is the case with native rough FIG. 1. A soluble GST fusion protein containing the Sec61␤ cytoplasmic domain competes for attachment of ribosomes to stripped rough microsomes. A, expression of Sec61␤ c /GST. The GST and Sec61␤ c /GST fusion proteins were expressed in bacteria and purified using glutathione-agarose as described under "Experimental Procedures." After elution with free glutathione, the fusion protein was left untreated or else subjected to thrombin digestion to cleave between the Sec61␤ c and GST sequences (50 g of fusion protein were incubated with 1 g of thrombin protease overnight at room temperature). To cleave between the Sec61␣ and GST sequences, equal aliquots of each sample (ϳ5 g per lane), with or without prior thrombin treatment as indicated, were subjected to SDS-PAGE on a 12% acrylamide gel, followed by Coomassie Blue staining. The free Sec61␤ c sequence (ϳ8 kDa) was detected transiently in control experiments but was degraded by the extended thrombin treatment needed for complete cleavage of the fusion protein and is not seen in the figure. The uncleaved forms of both GST and Sec61␤ c /GST were used in all subsequent experiments. B, activity of 1 Sec61␤ c /GST in the ribosome binding/competition assay using labeled nontranslating ribosomes. The ribosome binding/competition assay was done as described in the text, using an input of 2 g of SLR-labeled ribosomes and 2 g of the indicated fusion protein. The excess of fusion proteins over ribosomes was ϳ140:1, assuming a M r of 4.5 ϫ 10 6 for the ribosome and 3.2 ϫ 10 4 for each fusion protein. After 10 min on ice, 10 eq of PKRM was added, and the reaction was incubated for an additional 30 min on ice, followed by flotation to separate the bound and unbound fractions and scintillation counting. The fraction bound (y axis) was calculated by dividing the bound value by the total (bound plus unbound); this value was set equal to 1 in the case of the control sample (GST alone), and the fraction bound in the presence of Sec61␤ c /GST was expressed relative to the control value. C, titration of the Sec61␤ c /GST competitive activity. The binding experiment was performed as in B. 2 g of SLR-labeled ribosomes were mixed with the indicated molar excess of the Sec61␤ c /GST fusion over binding sites. (The number of binding sites was calculated from the X intercept of a Scatchard plot generated in the absence of competitor (not shown).) A 64-fold excess of fusion protein over binding sites corresponds to approximately a 100-fold excess over ribosomes at the constant input of ribosomes used in this experiment. The fusion protein was serially diluted with elution buffer (5 mM glutathione, 50 mM Tris-HCl, pH 8.0), and 2.5-l was added to each 20-l reaction. The control sample received 2.5 l of elution buffer. Reactions were incubated on ice for 10 min before addition of 10 eq of PKRM. The binding and flotation steps were performed as in B. The fraction of ribosomes bound was calculated as in B.
FIG. 2. The Sec61␤ c /GST fusion protein immobilized on glutathione-agarose beads binds directly to ribosomes. The binding experiment was performed as described under "Experimental Procedures" for the direct binding of ribosomes to glutathione-agarose beads with the bound fusion protein. 20 g of Sec61␤ c /GST or GST alone, bound to glutathione-agarose (ϳ25 l of beads), was used for each data point. A Scatchard plot was obtained from measurements of the bound and unbound counts, and the dissociation constant (K D ) for binding of ribosomes to Sec61␤ c /GST was obtained from the slope of the plot. The total number of binding sites was estimated from the X intercept, using the value given in the legend to Fig. 1 for the molecular weight of the fusion protein.
microsomes. A fixed amount of labeled ribosomal monomers (80 S) was added to a binding reaction containing Sec61␤ c / GST-agarose beads (Sec61␤ c /GST, Fig. 5, left panel) or stripped rough microsomes (PKRM, Fig. 5, right panel). An equimolar amount of unlabeled small (40 S) or large (60 S) subunits or monomers (80 S) was added as a competitor, and the extent of binding of the unlabeled component was inferred from the extent to which the binding of labeled monomers was reduced. In both stripped microsomes and Sec61␤ c /GST-agarose beads, unlabeled monomers (80 S) could compete efficiently for binding. Small subunits (40 S) had a more modest effect on binding. The effect of large subunits (60 S) was greater in stripped rough microsomes than in ␤ c /GST-agarose beads, but this difference may not be significant in comparison to the effect of the other unlabeled competitors. In the case of Sec61␤ c /GST, the competitive activity of large subunits was consistently greater than that of small subunits, implying that Sec␤ c /GST binds preferentially to large subunits. DISCUSSION By using two complementary methods, we have obtained results suggesting physical contact between the Sec61 ␤ sub-unit and the ribosome. First, the Sec61␤ cytoplasmic domain (Sec61␤ c ) prevents ribosome attachment to ER membranes when it is added to an in vitro ribosome binding assay in the form of a GST fusion protein (Fig. 1). Second, the Sec61␤ c /GST fusion protein binds directly to ribosomes (Fig. 2), and the ribosome binding affinity of Sec61␤ c /GST is comparable to that of stripped rough microsomes.
Our results are surprising in view of the work of Kalies et al. (29) showing that Sec61␤-depleted proteoliposomes bind to nontranslating ribosomes with an affinity comparable to that of proteoliposomes containing intact Sec61p. Based on this it appears possible that the ribosome binding activity of Sec61␤, as detected in our experiments, is merely redundant in vivo. However, the reported ribosome-binding affinities of FIG. 3. Yeast transfer RNA does not compete efficiently for the direct binding of nontranslating ribosomes to Sec61bc/GST. 3 g of SLR-labeled ribosomes was incubated with the indicated molar excess of yeast tRNA (suspended in RBB) or unlabeled ribosomes before the addition of 20 g of Sec61␤ c /GST bound to glutathione-agarose. Volumes were compensated with RBB. The binding reaction and subsequent analysis were carried out as described in Fig. 2. RNA concentration was estimated by A 260 . The M r of RNA in the ribosome was assumed to be 2.5 ϫ 10 6 , and the average M r of yeast tRNA was assumed to be 2 ϫ 10 4 . The fraction bound (y axis) represents the number of bound counts relative to the sample with no competitor, which was set equal to 1 in each panel. FIG. 5. Large ribosomal subunits compete more efficiently than small subunits in an assay for ribosome binding to Sec61␤ c / GST. SLR-labeled 80 S ribosomes were mixed with an equimolar amount of unlabeled 80 S monomers (80 S), small subunits (40 S), or large subunits (60 S). The amount added was determined by the A 260 of each preparation, following the relative values given by Borgese et al. (24). To this mixture was added a limiting amount of Sec61␤ c /GST beads (left panel) or PKRM (right panel). The binding reactions and subsequent measurements were carried out as described under "Experimental Procedures" for ribosome binding to Sec61␤c/GST glutathioneagarose beads (as in Figs. 2-4) and for ribosome binding to stripped membranes (as in Fig. 1), respectively. Values in the graph are expressed as a fraction of bound:unbound counts, relative to a sample where no competitor was added (no comp.). 3-5 replicates were used to obtain each data point shown in the graph. the Sec61␤-depleted and the mock-depleted preparations (50 and 45 nM, respectively) in the latter study, although comparable to each other, are both considerably weaker than the affinities measured for ribosome binding to native stripped microsomes in the work of the same authors (25,41) which agree with the values from our own experiments (not shown; see also Ref. 24). As we did not attempt to reproduce the depletion and reconstitution experiments in the present study, we cannot interpret this discrepancy with any confidence. It nonetheless leaves room for speculation that the effect of Sec61␤ depletion in the report by Kalies et al. (29) is masked by a general reduction in the affinity of both Sec61␤depleted and mock-depleted proteoliposomes and that the contribution of Sec61␤ to in vitro ribosome binding is more substantial than the latter results suggest. The difference in ribosome binding affinity between ER-derived reconstituted proteoliposomes and purified Sec61p-reconstituted membranes also leaves open the possibility that, in the former case, ribosome binding involves ER membrane proteins other than Sec61p. At least two ER membrane proteins with independent ribosome binding activity, p34 (42,43) and p180 (44), have been identified previously, although neither has been shown to associate with the translocon. Binding of ribosomes to membrane proteins other than Sec61p in reconstituted proteoliposomes might mask an effect of Sec61␤ depletion on the ribosome binding affinity of Sec61p itself.
The study of Kalies et al. (29) also shows that the depletion of Sec61␤ impairs the in vitro transport of preprolactin into Sec61-reconstituted membranes, although this impairment can be overcome by allowing sufficient time for targeting of the ribosome-nascent-chain complex before resumption of translation. In the absence of a requirement for Sec61␤ in ribosome binding, the authors (29) suggest that Sec61␤ might be involved in a post-targeting step, possibly assisting in the insertion of the nascent chain into the translocation pore. This agrees with observations by Laird and High (45) that the nascent chain can be chemically cross-linked to Sec61␤ at a point near its site of synthesis on the ribosome, presumably before it enters the translocation pore. In our study, the Sec61␤ c /GST fusion protein with high affinity to nontranslating ribosomes suggests that the physiological role of Sec61␤ is not limited to an interaction with the nascent chain. However, our results are not incompatible with the idea that Sec61␤ normally acts only at a post-targeting stage. If the ribosome normally binds first to another membrane component and then to Sec61␤, depletion of Sec61␤ from membranes should not affect their apparent ribosome binding activity; presentation of Sec61␤ out of its normal context (as in our study) would reveal its latent ribosome binding activity. This would reconcile our findings with those of Kalies et al. (29), where depletion of Sec61␤ from the Sec61p complex did not affect its affinity for ribosomes. However, given that the cytoplasmic domain of Sec61␤ is large compared with the other cytoplasmic portions of the Sec61p complex, it seems unlikely that this domain is normally blocked or buried within the complex. Consistent with the idea that Sec61␤ is not essential for translocation, preliminary studies in our laboratory did not reveal an effect on the translocation of preprolactin when the Sec61␤ c /GST fusion protein was added to a reticulocyte lysate translation system in excess over ribosomes (not shown).
The question of whether Sec61␤ c /GST binds primarily to ribosomal protein or to ribosomal RNA was not addressed in this study. The identity of the ribosomal components involved in ER membrane binding has not been firmly established. Although previous studies identified groups of ribosomal proteins that are relatively inaccessible to chemical modification in the membrane-bound versus the free state (46,47), the recent work of Prinz et al. (41) indicates that the ribosome binds to the ER membrane through the 28 S ribosomal RNA and that extensive proteolysis of ribosomes does not affect their ability to bind ER membranes. However, the type of ribosome binding measured in these experiments was insensitive to the presence of Sec61␤ (29,41), and it remains possible that the binding of ribosomes to Sec61␤ reported here involves a different cognate ribosomal component. The in vitro binding of ribosomes to Sec61␤ c /GST is the most reductive system available at present for the study of the ER-ribosome junction, and further biochemical studies using this system may avoid some difficulties inherent in studies using native rough microsomes.