Inhibition of Protein Translocation across the Endoplasmic Reticulum Membrane by Sterols*

Cholesterol and related sterols are known to modulate the physical properties of biological membranes and can affect the activities of membrane-bound protein complexes. Here, we report that an early step in protein translocation across the endoplasmic reticulum (ER) membrane is reversibly inhibited by cholesterol levels significantly lower than those found in the plasma membrane. By UV-induced chemical cross-linking we further show that high cholesterol levels prevent cross-linking between ribosome-nascent chain complexes and components of the Sec61 translocon, but have no effect on cross-linking to the signal recognition particle. The inhibiting effect on translocation is different between different sterols. Our data suggest that the protein translocation machinery may be sensitive to changes in cholesterol levels in the ER membrane.

Sterols are known to modulate the physical properties of biological membranes (1)(2)(3)(4). In particular, cholesterol increases the orientational order and reduces the rate of motion of the phospholipid hydrocarbon chains (1)(2)(3) and decreases the effective free volume in the membrane (5). The induced changes in membrane properties can help to modulate the activity of membrane-resident proteins, in addition to more specific protein regulation mechanisms (2,6). Cholesterol can further affect the function of individual proteins by direct binding (7)(8)(9). Since cholesterol increases the lipid chain order, the insertion of cholesterol leads to a laterally more condensed membrane (1)(2)(3)(4). Cholesterol/sphingolipid rafts and caveolae in the plasma membranes are good examples of such tightly packed, laterally segregated cholesterol-rich domains (4,10,11). Raft-like structures have also been suggested to exist in the Golgi and in the endocytic pathway, while the endoplasmic reticulum (ER) 1 and other internal membranes are devoid of rafts (12).
Cholesterol levels differ markedly between different cellular membranes. The cholesterol concentration is low in the ER and increases throughout the secretory pathway. Most of the cellular cholesterol is found in the plasma membrane (13)(14)(15)(16). The fact that the ER membrane contains only small amounts of cholesterol (13,17,18) is quite amazing, since it is the site of cholesterol biosynthesis, esterification, and regulation (19,20). All of the regulatory proteins residing in the ER respond to changes in the local cholesterol level. Changes in plasma membrane cholesterol are reflected in the ER cholesterol levels, allowing feedback control of cholesterol synthesis and esterification in the ER (20 -22).
Given that membrane-bound ER resident enzymes normally function in a cholesterol-poor environment, they may be expected to be particularly sensitive to increases in cholesterol levels. Here we report that an early step in protein translocation across the ER membrane is reversibly inhibited by cholesterol levels significantly lower than those found in the plasma membrane. We have also examined whether a change in cholesterol double bond position or number affects protein translocation. 5␣-Cholestan-3␤-ol (dihydrocholesterol) and 4-cholesten-3␤-ol (allocholesterol) inhibit protein translocation, whereas 4-cholesten-3-one (cholestenone), 7(5␣)-cholesten-3␤-ol (lathosterol), and 8(14)-cholesten-3␤-ol (8-sterol) have no apparent effects. By UV-induced chemical cross-linking we further show that high cholesterol levels prevent cross-linking between ribosome-nascent chain complexes (RNCs) and components of the Sec61 translocon, but have no effect on crosslinking of RNCs to the signal recognition particle (SRP).

Enzymes and
(Madison, WI) or New England Biolabs (Boston, MA). T7 DNA polymerase, 14 C-methylated marker proteins, ribonucleotides, deoxyribonucleotides, dideoxyribonucleotides, the cap analogues m7G(5Ј)ppp(5Ј)G and G(5Ј)ppp(5Ј)G, and [ 14 C]oleoyl-coenzyme A (56 mCi/mmol) were from Amersham Pharmacia Biotech (Uppsala, Sweden and Piscataway, NJ). Protein A-Sepharose, puromycin, and 7-methylguanosine 5Ј-monophosphate were from Sigma. Affinity-purified rabbit antisera to the Cterminal 13 amino acids of TRAM were obtained from Research Genetics (Huntsville, AL). Most of the sterols used and methyl-␤-cyclodextrin were obtained from Sigma. Allocholesterol was purchased from Steraloids (Newport, RI) and 8-sterol from Research Plus Inc. (Bayonne, NJ). The purity of all sterols was examined by reversed phase high performance liquid chromatography, and all, except for dihydrocholesterol, were 99% pure and were used without further purification. Dihydrocholesterol was purified by crystallization in ethanol-water at Ϫ20°C overnight and was shown to be 99% pure by reversed phase high performance liquid chromatography. Dog pancreas microsomes were prepared as described in Ref. 24.
DNA Manipulations-Site-specific mutagenesis was performed according to the method of Kunkel (25,26). All mutants were confirmed by sequencing of plasmid DNA. All cloning steps were done according to standard procedures.
Lep Expression Plasmid-For cloning into and expression of Lep from the pGEM1 plasmid, the 5Ј end of the lep gene was modified, first, by the introduction of an XbaI site and, second, by changing the context 5Ј to the initiator ATG codon to a "Kozak consensus" sequence (27). Thus, the 5Ј region of the gene was modified to: . . Template for in vitro transcription of truncated Lep mRNA was prepared using PCR to amplify a fragment from the Lep-pGEM1 plasmid. The 5Ј primer was situated 210 bases upstream of the translation start, and the amplified fragment thus contained the SP6 transcriptional promoter from pGEM1. The 3Ј primer was chosen to produce a truncated fragment ending at codon 271 in Lep, and no stop codon was included. Translation/translocation reactions of truncated Lep mRNA for puromycin treatment (Fig. 3, panel B) were performed at 22°C in a total volume of 14 l. When relevant, 3.5 g of cholesterol (0.23 g/l) was added after 25-min translation. After an additional 10-min incubation, 1.9 l of potassium acetate (4 M), 1.5 l of magnesium acetate (20 mM), and 1.5 l of puromycin (30 mM) were added, and the incubation was continued at 22°C for another 10 min.
For alkali extraction, a 10-l translation mix was diluted with 90 l of ice-cold 0.1 M Na 2 CO 3 /NaOH, pH 11.5, and the samples were centrifuged for 10 min at 70,000 rpm in a Beckman tabletop 100.3 rotor. Before SDS-PAGE, the supernatants were precipitated with trichloroacetic acid, and all samples were heated at 95°C for 5 min.
Proteins were analyzed by SDS-PAGE, and gels were visualized on a Fuji FLA-3000 phosphoimager using the Fuji Image Reader 8.1j software.
UV-induced Cross-linking and Immunoprecipitation-mRNA coding for truncated preprolactin polypeptide was synthesized by in vitro transcription using linearized pSPBP4 DNA and SP6 RNA polymerase as described previously (29). The PvuII restriction site (after codon 86) was chosen to obtain a run-off transcript encoding the N-terminal 86 residues of preprolactin. In vitro translation for photolysis (25 l total volume) and immunoprecipitation (50 l total volume) were done in wheat germ cell-free extract (29,30)  Photoreactive N ⑀ -(5-azido-2-nitrobenzoyl)-Lys-tRNA (⑀ANB-Lys-tRNA) was prepared as detailed previously (31). Samples were photolysed on ice for 15 min using a 500-watt mercury arc lamp (29). After photolysis, membranes or ribosomes were sedimented through a sucrose cushion in a Beckman airfuge at 4°C for 5 min, 20 p.s.i. and 60 min, 30 p.s.i., respectively.
Microsome and ribosome pellets were resuspended in buffer (100 mM Tris-HCl, pH 7.6) containing detergent (0.25% w/v SDS for Sec61␣-and Sec61␤-specific antibodies and 1% w/v SDS for TRAM-and SRP54specific antibodies) and placed at 55°C for a minimum of 30 min. The volume was increased with buffer A (140 mM NaCl, 10 mM Tris-HCl, pH 7.6, 2% v/v Triton X-100, 0.2% w/v SDS) for Sec61␣ and Sec61␤ antibodies and buffer B (150 mM NaCl, 50 mM Tris-HCl, pH 7.6, 2% (v/v) Triton X-100, 0.2% SDS) for TRAM and SRP54 antibodies. Samples were then precleared by rocking with protein A-Sepharose at room temperature for 1 h before the Sepharose beads were removed by sedimentation.
Sec61␣-, Sec61␤-, TRAM-, or SRP54-specific antiserum was added to each supernatant, and the samples were rocked overnight at 4°C. Protein A-Sepharose was then added to each sample and incubated for a minimum of 2 h at 4°C. The immunoprecipitated proteins were analyzed by SDS-PAGE and visualized using a Bio-Rad GS-250 phosphorimager (29).
Addition of Sterols to Microsomes-To enrich the microsomes with cholesterol or sterol analogues, complexes between methyl-␤-cyclodextrin (CyD) and sterols were prepared by a slight modification of the procedure described in Ref. 32. 32 mg of CyD was dissolved in 1 ml of MS buffer (0.25 M sucrose, 20 mM Hepes-KOH, pH 7.5), and this solution was added to a thin film of sterol (1 mg) in a tube. The molar ratio of sterol to CyD was 1/10. The mixture was sonicated in a bath sonicator for 20 min. Translation mixes including microsomes were incubated with various amounts of sterol in the form of sterol-CyD complexes during the translation/translocation assay. Sterol-CyD complexes were added to the translation mix together with mRNA and microsomes at the beginning of the incubation, unless otherwise stated.
Analysis of Microsomal Lipids-Microsomes were extracted with hexane/2-propanol (3/2, v/v) and water, and again with hexane, to obtain the total lipids. The organic phase was then collected and evaporated to dryness. The extracted lipids were applied on a high performance thin layer chromatography plate and eluted with chloroform: methanol:water (25:10:1.1 by volume) to separate the phospholipid classes. The phospholipids were visualized by staining with cupric acetate (3%, w/v and 8% H 3 PO 4 , w/v) and heating the plates for 15 min at 150°C to develop the color and identified from standards run in parallel. The absorbance of the lipid spots was determined with a scanning densitometer (Camag TLC Scanner 3). The amount of sphingomyelin in the microsomal preparation was calculated from the absorbance data compared with a sphingomyelin standard series.
The amount of cholesterol in the microsomal membranes was analyzed from the total lipid extract by gas-liquid chromatography (Shimadzu GC-14A). Epicoprostanol was added as an internal standard to the samples. The cholesterol amount in the sample was calculated compared with a standard sample with known cholesterol concentration. The samples were injected on a Sac TM -5 column (0.25-m film, 0.25 mm ϫ 30 m, Supelco) with the inlet block at 260°C and separated on the column with a temperature program of 6°C⅐min Ϫ1 from 260 to 280°C. The samples were detected with a flame ionization detector at 300°C.
The amount of sphingomyelin and cholesterol in the microsomal preparation was calculated relative to the protein concentration in the microsomes. The protein concentration in the sample was determined by the method of Lowry (33), using bovine serum albumin as standard.
Measurement of Cholesterol in Microsomes Using ACAT-The cholesterol measurements were done by an ACAT assay as described in Ref. 18. In vitro translations were done in reticulocyte lysate in the presence of rough microsomes and cholesterol-CyD complexes or unloaded CyD. After translation the microsomes were sedimented in a Beckman tabletop centrifuge (4°C, 70 K, 10 min, 100.3 rotor). The pellet was resuspended in MS buffer containing dithiothreitol (1 mM), bovine serum albumin (1 mg/ml), and [ 14 C]oleoyl-coenzyme A (30 M) and incubated at 37°C for 60 min to allow ACAT to esterify cholesterol residing in the ER fraction. After incubation, microsomes were sedimented, the enzyme was inactivated, and cholesterol esters were measured. Briefly, microsomal lipids were extracted with hexane/2-propanol (3/2 v/v), the solvent was evaporated, and the lipids were re-dissolved in hexane/2-propanol. The lipid samples were applied on normal phase thin layer chromatography plates and separated with hexane/diethyl ether/acetic acid (130/30/2 v/v/v). Lipid spots were visualized by stain-ing with iodine and identified from standards run in parallel. The cholesterol ester spots were cut into scintillation vials and counted for radioactivity with a LKB Rack-Beta liquid scintillation counter.

Cholesterol Reversibly Inhibits Protein Translocation across
the Microsomal Membrane-To study the effect of cholesterol on protein translocation across the ER membrane, rough dog pancreas microsomes were incubated with different amounts of methyl-␤-cyclodextrin (CyD)-bound cholesterol in an in vitro protein translocation assay. As shown in Fig. 1, N- To ascertain that the inhibition of glycosylation was caused by an inhibition of translocation rather than an effect on the oligosaccharyltransferase, microsomes were subjected to alkaline extraction to assay for membrane integration of Lep (Fig.  1, panel C). In the absence of cholesterol, essentially all glycosylated molecules remained in the alkali-resistant membrane pellet (P), whereas most of the molecules were found in the supernatant (S) when translation was carried out in the presence of cholesterol.
Finally, translocation of Lep was also assayed by signal peptide cleavage. For this, a construct with a signal peptidase cleavage site at the C-terminal end of an engineered H2 segment was used (the H2 sequence was . . . KKKKL 14 VPSAQAϩA . . . where the "ϩ" sign indicates the signal peptidase cleavage site; see Ref. 34) (Fig. 2, panel A). Again, translocation was completely inhibited in microsomes that were exposed to 0.10 g/l cholesterol (panel B).
The reversibility of the cholesterol-dependent inhibition of translocation was tested by first preincubating microsomes at 30°C for 20 min with an inhibiting concentration of cholesterol (0.10 g/l) followed by a second preincubation for 20 min with an excess of uncomplexed cyclodextrin to extract cholesterol back from the microsomal membranes. The treated microsomes (including the added CyD and cholesterol) were then used in a normal in vitro protein translocation assay. As shown in Fig. 3, a concentration of 5.2 g/l uncomplexed cyclodextrin in the second preincubation was sufficient to restore the translocation activity, showing that the inhibiting effect of cholesterol is reversible.
To better define the step at which cholesterol exerts its inhibiting effect, we used Lep mRNA truncated at codon 271 (56 residues downstream of the glycosylation site) to produce nonglycosylated, translocon-bound RNCs, and then checked whether addition of cholesterol could still block full translocation (as assayed by glycosylation) after release of the nascent chain by treatment with puromycin and high salt (35). As shown in Fig. 4, released nascent chains became glycosylated both in the absence (lane 5) and presence (lane 7) of cholesterol, albeit somewhat less efficiently in the latter case. It is unlikely that the cholesterol did not have time to equilibrate between the CyD-bound form and the microsomal membranes during the 10-min incubation preceding the addition of puromycin/ high salt, since translocation of full-length Lep is efficiently blocked even when cholesterol is added together with Lep mRNA to the translation/translocation mix (Fig. 1). Since translation was carried out at 22°C in this experiment rather than at 30°C as is the experiments using full-length Lep mRNA, we also ascertained that translocation of full-length Lep was completely blocked when the same concentration of CyD-cholesterol as used with the truncated chains (0.23 g/l cholesterol) was included in the standard in vitro translation reaction at 22°C (data not shown).
We conclude that cholesterol has at best a marginal effect on translocation through the Sec61 translocon per se and thus that the main inhibition must be at an earlier step in the translocation pathway.
Cholesterol Levels in the Microsomal Membranes-Since the microsomal preparations used also contain varying levels of contaminating membranes (the sphingomyelin and unesterified cholesterol levels in the preparation were, respectively, 1.4 Ϯ 0.05 nmol and 1.8 Ϯ 0.05 nmol per mg of protein, suggesting some contamination by plasma membranes, endosomal membranes, and possibly by some Golgi membranes), the cholesterol levels in the ER-derived microsomes cannot be assessed by simply measuring overall cholesterol in the membrane fraction. Instead, an assay originally developed by Lange and Steck was used (18), in which ER cholesterol is specifically esterified by the ER enzyme ACAT. Cholesterol levels were measured both in nontreated microsomes and in microsomes incubated with 2 g of CyD-cholesterol and were found to be increased 3.7-fold in the treated sample (mean of three experiments; data not shown). Since the typical cholesterol level in the ER is about 0.2% of total cellular unesterified cholesterol (18), this suggests that the ER membranes in the treated microsomes contained maximally 0.8% of cellular free cholesterol equivalent. Thus, the amount of cholesterol in the ER membranes in the cholesterol-loaded microsomes is much less than the amount of cholesterol found in the plasma membrane. Depending on the cell type and assay method used, cellular plasma membranes have been reported to contain between 40 and 90% of the total cellular unesterified cholesterol (13)(14)(15)(16). Golgi membranes contain an intermediate level of cholesterol as compared with ER and the plasma membrane (36,37). Cholesterol is asymmetrically distributed within the Golgi, with greater amounts of cholesterol found in the portion of the Golgi located near the plasma membrane than in the Golgi located near the endoplasmic reticulum (37). Compilation of published estimates of lipid content in various organelles suggest that whereas the cholesterol/phospholipid molar ratio in ER membranes is 0.08 or less, it is about 0.16 in Golgi membranes (38). However, these values differ in cells from different tissues (see e.g. Ref. 39), so a direct comparison is difficult to make. Nevertheless, considering that the cholesterol-loaded microsomes in our preparations contain the equivalent of only Յ1% of the cellular free cholesterol, the cholesterol concentration of ER membranes in the loaded microsomes is most likely lower than the amount of cholesterol found in Golgi membranes.
We conclude that cholesterol blocks protein translocation across microsomal membranes at a concentration that is similar to its concentration in the Golgi and significantly lower than in the plasma membrane. Cholesterol Inhibits Nascent Chain Cross-linking to Components of the Sec Translocon but Not to SRP-To further characterize the cholesterol-dependent inhibition of translocation, we used UV-inducible cross-linking to probe the interaction between RNCs and components of the translocation machinery. RNCs were prepared by translation of truncated mRNAs encoding the first 86 residues of the secretory model protein preprolactin (pPL). Lysyl-tRNAs carrying a modified ⑀ANBlysine residue were added to the translation mix to incorporate the photoreactive ⑀ANB-Lys probe in place of lysine at positions 4, 9, 72, and 78 in pPL (Fig. 5, panel A). The pPL-86nascent chain is long enough to be efficiently targeted to the Sec61 translocon in the microsomal membrane, but is too short for signal peptide cleavage to take place (40,41).
As shown in Fig. 5, panel B, when translated in the absence of microsomes, pPL-86 can be efficiently cross-linked to SRP both in the presence and absence of 0.14 g/l of added cholesterol (lanes 4 and 5). When microsomes are present, crosslinking to SRP is only seen in the presence of added cholesterol (lane 9 versus lane 10). In contrast, cross-linking to the translocon components Sec61␣ (panel C, lanes 4 and 5), TRAM (panel C, lanes 6 and 7), and Sec61␤ (panel D, lanes 4 and 5) is only seen in the absence of added cholesterol (the band just above Sec61␣ photoadduct in panel C, lanes 2 and 4, probably results from nascent chain cross-linking to Sec61␣ at two different sites, as has been observed previously (42)). In a separate control experiment, we verified that CyD alone (4.9 g/l) did not inhibit cross-linking of pPL-86 to either Sec61␣, TRAM, or SRP (data not shown).
We conclude that cholesterol does not affect the interaction between RNCs and SRP, but efficiently prevents targeting of RNCs to the Sec61 translocon in the microsomal membrane.
Inhibiting Effects of Other Sterols on Protein Translocation-It has been established that sterol analogues having altered double bond position or number can be added to L-cell mouse fibroblasts without significant cellular toxicity (43)(44)(45). Still, a number of enzymes involved in lipid homeostasis are sensitive to the structure of cholesterol, for example CTP:phosphocholine cytidylyltransferase and ACAT (47)(48)(49). Based on this knowledge we wanted to examine whether altering the cholesterol double bond structure would affect protein translocation. We thus tested the inhibiting effect on protein translocation of a set of other sterols (dihydrocholesterol, cholestenone, lathosterol, allocholesterol, and 8-sterol, Fig. 6, panel A) in the same way as had been done for cholesterol. Of these sterols, lathosterol and dihydrocholesterol have been detected in animal cells, while allocholesterol, cholestenone, and 8-sterol, to our knowledge, are purely synthetic (50,51). Dihydrocholesterol and allocholesterol inhibited translocation of Lep at roughly the same concentrations as did cholesterol (Fig. 6, panel B, whereas cholestenone, lathosterol, and 8-sterol had no effect up to the highest concentration tested (0.22 g/l).

DISCUSSION
To be able to maintain a high concentration of cholesterol in the plasma membrane, the transport pathways for both newly synthesized and lipoprotein-derived cholesterol are directed toward the cell surface (52)(53)(54)(55). The amount of cholesterol that can be solubilized in the plasma membrane has been shown to depend on the sphingomyelin content (56 -58). When the solubilization capacity of the plasma membrane is exceeded, cholesterol is transported from the plasma membrane to the ER for esterification by ACAT (19,59). Low density lipoprotein-derived cholesterol can also be transported directly from lysosomes to the ER by a pathway that is independent of the plasma membrane (16). Cholesterol transport from the plasma membrane to the ER seems to occur by a different mechanism than the transport of newly synthesized cholesterol from the ER back to the plasma membrane, but the mechanism for this transport is not yet clearly defined (60). The transport has been suggested to be mediated by a "classical" vesicular pathway or by a novel energy-independent vesicular transport mechanism (16,61,62). Caveolae have also been suggested to play an important role in cholesterol transport (16,(63)(64)(65). The pool of cholesterol esters in the ER is in rapid equilibrium with the regulatory pool of free cholesterol in the cell through the activity of neutral cholesterol esterase, so that the level of free cholesterol in the cell is maintained constant (64).
It seems likely that the modulation of ER cholesterol levels controls the activity of various regulatory elements of cholesterol homeostasis located in the ER (55). A small increase in the level of plasma membrane cholesterol above a critical threshold increases the transfer of cholesterol to the ER and signals the control proteins to reduce the amount of free cholesterol in the cell (66). The regulatory elements in the ER include enzymes of sterol biosynthesis, ACAT, and a family of transcription factor precursors called sterol regulatory element-binding proteins that control the expression of other regulatory elements (20,22,67). Cholesterol is known to decrease the effective free volume available for molecular motion in the hydrophobic core of lipid bilayers (5), and this change in membrane properties has been shown to affect the function of proteins residing in the membrane (6,68,69). Possibly, some ER proteins need free volume in the membrane to be able to undergo conformational changes upon activation and therefore can sense the cholesterol-induced changes in the properties of the ER.
In this study, we have found that cholesterol reversibly inhibits an early step in protein translocation across the microsomal membrane when present at a concentration that is sim- ilar to its concentration in the Golgi and significantly lower than in the plasma membranes of eukaryotic cells. The effect seems to be specific for sterols that can interact with phospholipids and decrease the free volume of the membrane (cholesterol, dihydrocholesterol, and allocholesterol), since cholestenone does not inhibit the translocation reaction (Fig. 6). It is well accepted that cholestenone represents a membrane-inactive steroid and that it cannot order phospholipid acyl chains to the extent that cholesterol does (70,71). Lathosterol and 8-sterol also do not significantly inhibit the translocation reaction. Both of them have been shown to order membranes to a lesser extent than cholesterol (50). It is interesting that lathosterol behaves differently than cholesterol, allocholesterol, and dihydrocholesterol, because of these sterols only lathosterol is unable to activate membrane-bound CTP:phosphocholine cytidylyltransferase (49). The different effect of lathosterol on cytidylyltransferase as compared with the other sterols used in the study was suggested to be caused by the inability of lathosterol to stabilize the lamellar-to-hexagonal phase transition of the membrane bilayer (49). Unfortunately, 8-sterol was not included in that study.
As shown by photocross-linking using truncated pPL-nascent chains, cholesterol does not inhibit the interaction between ribosome-nascent chain complexes and SRP, but fully blocks the interaction between ribosome-nascent chain complexes and components of the translocon (Fig. 5). Furthermore, truncated Lep-nascent chain translocation intermediates can be fully translocated across the microsomal membrane and glycosylated when released from the ribosome by treatment with puromycin-high salt even in the presence of cholesterol (Fig. 4). Cholesterol thus exerts its effect either at the level of the interaction between SRP and the SRP receptor or at the level of binding of ribosome-nascent chain complexes to the translocon itself, but does not inactivate the translocon once targeting has been completed. The translocon-induced, SRP receptor subunit ␤-mediated release of the signal sequence from SRP may be a key step in this regard (46).
At present, we can only speculate on the biological relevance of these findings. One interesting possibility is that cholesterolmediated inhibition of protein translocation may be a way to inactivate translocon complexes that have leaked out of the ER (23). Another possibility is that the inhibition may be important under conditions when increased plasma membrane cholesterol increases ER cholesterol levels (18,59). In any case, our results show that the protein translocation machinery in mammalian cells is well adapted to its normal low cholesterol environment in the ER membrane.