4184 Research Article Positive charges on the translocating polypeptide chain arrest movement through the translocon

Hidenobu Fujita, Marifu Yamagishi, Yuichiro Kida and Masao Sakaguchi* Graduate School of Life Science, University of Hyogo, Kouto Ako-gun, Hyogo 678-1297, Japan *Author for correspondence ([email protected])

Accepted 27 July 2011 Journal of Cell Science 124, 4184–4193 ß 2011. Published by The Company of Biologists Ltd doi: 10.1242/jcs.086850

Summary Polypeptide chains synthesized by membrane-bound are translocated through, and integrated into, the (ER) membrane by means of the translocation channel, the translocon. Positive charges on the nascent chain determine the orientation of the hydrophobic segment as it is inserted into the translocon and enhance the stop-translocation of translocating hydrophobic segments. Here we show that positive charges temporarily arrested ongoing polypeptide chain movement through the ER translocon by electrostatic interaction, even in the absence of a hydrophobic segment. The C-terminus of the polypeptide chain was elongated during the arrest, and then the full-length polypeptide chain moved through the translocon. The translocation-arrested polypeptide was not anchored to the membrane and the charges were on the cytoplasmic side of the membrane. The arrest effect was prevented by negatively charged residues inserted into the positive-charge cluster, and it was also suppressed by high salt conditions. We propose that positive charges are independent translocation regulators that are more active than previously believed.

Key words: Endoplasmic reticulum, Positive charge, Signal sequence, Translocation

Introduction The positive charges determine the orientation of the signal Positively charged residues of membrane have long been sequences. The H-segment of the signal sequence penetrates known to function as determinants of membrane topology, which the translocon, and then one side of the H-segment moves into the is reflected in a statistical rule of membrane topology, i.e. the lumen to form the TM topology. During the insertion into the positive-inside rule of membrane proteins (Sipos and von Heijne, translocon, the orientation of the signal sequence is determined Journal of Cell Science 1993; von Heijne and Gavel, 1988). Many membrane proteins in by the flanking positively charged amino acid residues (Goder the secretory pathway in eukaryotic cells are integrated into and Spiess, 2001; Kida et al., 2006; Sakaguchi, 1997). If the N- the endoplasmic reticulum (ER) membrane by the protein terminal side of the H-segment is rich in positive charges, as translocation channel, the so-called translocon. Protein observed in signal and type II signal-anchor sequences, translocation and integration are initiated by a signal sequence the N-terminus is retained on the cytoplasmic side and the on the nascent polypeptide chain. The signal sequences are segment forms the Ncytosol and Clumen orientation (termed the primarily defined by a hydrophobic segment (H-segment). A type II orientation). The is cleaved off by signal signal recognition particle recognizes the H-segment emerging peptidase in the ER lumen, while the signal-anchor is retained in from the and induces targeting of the ribosome-nascent the final protein as a TM segment. Alternatively, if the C- chain complex to the ER. The H-segment is transferred from the terminal side is rich in positive charges, the N-terminus is signal recognition particle to the translocon by the action of a translocated through the translocon and forms the Nlumen and signal recognition particle receptor, and the ribosome attaches to Ccytosol orientation (the type I orientation) and the signal-anchor the translocon. The core part of the translocon comprises the is retained as a TM segment. For example, the topology of complex, which is composed of Sec61a, Sec61b and synaptotagmin II is gradually converted depending on the number Sec61c (Johnson and van Waes, 1999; Rapoport, 2007). In of positive charges downstream of the H-segment (Kida et al., prokaryotic cells, a similar complex, the SecY complex, is 2006). involved in the protein translocation and membrane integration The positive charges also contribute to membrane spanning into the plasma membrane. On the basis of studies of crystal of the H-segment translocating through the translocon. The structures of the SecY complex (Tsukazaki et al., 2008; Van den polypeptide chain movement through the translocon is halted at Berg et al., 2004), 10 transmembrane (TM) segments of a single the hydrophobic TM segment and then the TM segment is SecY molecule can form an aqueous pore through which a wide released into the membrane lipid. These processes are called variety of hydrophilic polypeptide chains can move toward the stop-translocation and membrane insertion, respectively. A lumen. The ten TM helices are arranged in pseudo symmetry and simple partitioning into the of the H-segment form a clam-shell-like structure that can open laterally to allow should be a major factor for stop-translocation and membrane the TM segment to be released into the membrane lipid insertion of the TM segment (Hessa et al., 2005; Hessa et al., (Rapoport, 2007). 2007). The positive charges flanking the cytoplasm also have the Positive charges arrest translocation 4185

essential function of enhancing the stop-translocation of marginal is an independent translocation regulator and provides the H-segments (Kuroiwa et al., 1990; Kuroiwa et al., 1991; Lerch- physicochemical basis underlying the fundamental function of Bader et al., 2008). In special cases, a marginal H-segment that topogenesis. is insufficient for stop-translocation by itself can span the membrane only when it is accompanied by positively charged Results residues. Interestingly, the positive charges are effective even Effect of positive-charge clusters on protein translocation when they are separated from the H-segment by more than 60 To quantitatively assay the effect of charged amino acid residues residues (Fujita et al., 2010). In this case, the H-segment can be on polypeptide chain translocation through the ER translocon, we temporarily exposed to the lumen and then slides back into the constructed systematically designed model proteins consisting of translocon. The extent of the exposure in the lumen depends on rat serum albumin (RSA) as a backbone, a query sequence as the the position of the positive-charge cluster. Such a marginal H- middle portion, potential glycosylation sites in the upstream and segment is not integrated into the membrane and remains in a downstream regions and an N-terminal signal peptide (Fig. 1A). water-accessible environment. The signal peptide induces the co-translational translocation Here, we systematically examined the effect of positive of the downstream polypeptide. Initially, the upstream portion charges of the nascent polypeptide chain on the translocation in moves into the lumen where the first potential glycosylation site the absence of an H-segment and found that the positive charges is glycosylated by the oligosaccharyl transferase, the active site arrest the movement of the polypeptide chain through the of which is located in the lumenal side (Fig. 1C). When the entire translocon, independently of the H-segment. The polypeptide protein is translocated through the translocon, the second site is chain elongates up to the C-terminus during the arrest, and the glycosylated. If the translocation is interrupted at the query full-length polypeptide is gradually translocated into the ER segment, then only the upstream site is glycosylated. The lumen. These findings demonstrate that the positive charge fully translocated polypeptide is resistant to externally added Journal of Cell Science

Fig. 1. Positive-charge clusters arrest translocation in the absence of an H-segment. (A) Top panel: the rat serum albumin-based model protein included its signal peptide (S) at the N-terminus, a query sequence of charged or hydrophobic residues in the middle portion, and glycosylation sites (circles) in the upstream and downstream portions. The model proteins are named according to the included amino acids and their numbers, e.g. 10K stands for a cluster of 10 Lys residues. 16L4A has the following sequence: LLLLALLLLALLLLALLLLA. Bottom panel: the model proteins were translated in the absence and presence of RM for 1 hour. Aliquots were treated with proteinase K (ProK) in the absence or presence of nonionic detergent (Tx). The diglycosylated forms (open circles), monoglycosylated forms (red triangles) and nonglycosylated forms (squares) are indicated. Nonspecific product is indicated by asterisks. (B) Model proteins were translated in the presence of RM and half of each aliquot was treated with proteinase K. (C) Possible membrane topology of the model protein. If the nascent chain is fully translocated into the lumen, both potential glycosylation sites are glycosylated (white circles) and the nascent chain is fully resistant to proteinase K. If translocation is interrupted by the query sequence, only the upstream site is glycosylated, the C-terminal portion is exposed on the cytoplasmic side of the membrane, and is degraded by proteinase K. The query sequence (black rectangles), glycosylated potential sites (white circles), nonglycosylated potential sites (shaded dots) and ribosomes (shaded circle and oval) are indicated. (D) Quantification of translocated polypeptide chains. The diglycosylated form and monoglycosylated form were quantified and translocation percentages were calculated using the following formula: diglycosylated form/[(diglycosylated form)+(monoglycosylated form)]6100. The experiments were performed more than three times; values are means ± s.d. Lysine (K) and arginine (R) residues produced almost the same effect, whereas aspartic acid (D) and glutamic acid (E) residues produced no effect. 4186 Journal of Cell Science 124 (24)

proteinase K, whereas the membrane-spanning polypeptide is respectively) and the nonglycosylated forms were rarely degraded (Fig. 1C). observed. In this system, the nascent chains were efficiently We examined clusters of charged amino acid residues and targeted to the RM so that the products were either hydrophobic residues as query sequences. The name of each monoglycosylated or diglycosylated. These larger forms were query sequence was created using the single letter code of the converted to the smaller form by endoglycosidase H treatment amino acid and the number of each amino acid; for example, (data not shown), confirming that the larger forms comprised ‘14K’ stands for a cluster of 14 lysine residues, and ‘16L4A’ for diglycosylated and monoglycosylated forms. The model protein 16 leucine and four alanine residues. When the models were containing the negatively charged cluster (e.g. 14D in Fig. 1A, synthesized in the reticulocyte lysate cell-free translation system and 14E in Fig. 1B) had lower mobility on the gel, for unknown for 1 hour in the absence of rough microsomal membranes (RM), reasons. As expected, the monoglycosylated forms (red triangles) prominent single bands were observed (Fig. 1A, squares). When were degraded by proteinase K, whereas the diglycosylated forms synthesized in the presence of RM, 2-kDa- and 4-kDa-larger (open circles) were proteinase K resistant. The products were forms were observed (Fig. 1A,B, red triangles and open circles, degraded by proteinase K in the presence of mild non-ionic detergent (Triton X-100), indicating that the proteinase K resistance of the diglycosylated forms was caused by the membrane. We unexpectedly found that the 14K and 14R (arginine) models produced the monoglycosylated forms as much as the diglycosylated forms (Fig. 1A). More than 50% of their full- length products spanned the membrane and their C-terminal domains were on the cytoplasmic side even though there was no H-segment. The monoglycosylated forms were largely degraded by proteinase K, eliminating the possibility that the positive- charge cluster inhibited glycosylation in the lumen. The 0K model was completely translocated, diglycosylated in the lumen and resistant to proteinase K digestion; whereas the 16L4A model was membrane-integrated, monoglycosylated in the upstream region and sensitive to proteinase K. We estimated the translocation efficiency as the percentage of the diglycosylated form in the diglycosylated form plus the monoglycosylated form (Fig. 1D). The Lys and Arg residues showed essentially the same effect, whereas the negatively charged residues, aspartic acid (Asp) and glutamic acid (Glu), had no effect on the translocation. The data clearly indicated that Journal of Cell Science the positive charges alone interrupted translocation in the absence of any H-segment and generated a full-length polypeptide spanning the membrane.

Positive charges temporarily arrest translocation To examine the time course of the polypeptide chain synthesis and translocation, the translation was performed for 20 minutes and translation was inhibited with cycloheximide and then the

Fig. 2. Translocation is transiently arrested. (A) The mRNAs coding for the model proteins were translated in the presence of RM for 20 minutes and the chain elongations were inhibited with cycloheximide (CH). The reaction mixtures were further incubated to chase the translocation for the indicated time periods. An aliquot was removed at each incubation time and analyzed by SDS-PAGE. Monoglycosylated forms (red triangles) and diglycosylated forms (white circles) are indicated. 4L4A represents the following sequence: LALALALA. (B) Translocation percentages at each time point were calculated as described in Fig. 1. The experiments were performed twice and the mean values are presented. (C) The monoglycosylated form is sensitive to proteinase K. After 20 minutes of translation the products were treated with proteinase K in the presence of CH. The monoglycosylated forms (red triangles) were almost completely degraded, whereas the diglycosylated forms were resistant. The monoglycosylated forms span the membrane and their C-terminal region was sensitive to proteinase K. (D) In the absence of the query sequence, the nascent chain was fully translocated and diglycosylated (a–d), whereas the positive-charge cluster caused nascent chains to stall at the translocon until the full-length chain was completed (e–h) and then gradually moved through the translocon (h–j). Positive charges arrest translocation 4187

translocations were chased (Fig. 2). The full-length polypeptide of the 0K model was not detectable after the 10 minutes translation but became visible at 15 minutes. The full-length chain synthesized within the 20-minute incubation was largely diglycosylated. From the estimation that there are 65–70 residues of nascent chain from the ribosome peptidyl transferase center to the oligosaccharyl transferase active site (Whitley et al., 1996), the second glycosylation site could be accessible to oligosaccharyl transferase after translation termination. The majority of the 0K model was thus diglycosylated shortly after being synthesized. In this experiment, the distance of 192 amino acid residues between the two potential glycosylation sites caused little, if any, difference in glycosylation timing (Fig. 2A,B, 0K). When the K-cluster was included, the amount of the monoglycosylated forms observed at 20 minutes of translation varied depending on the charge numbers (Fig. 2A,B). In the case of the 12K and 10K models, less than 30% of the nascent chains were diglycosylated at 20 minutes of translation. The monoglycosylated full-length polypeptides were gradually converted to the diglycosylated form during the chase incubation. By contrast, little of the 20K model was translocated, even after a 100-minute chase. Proteinase K treatment of the 20-minute translation products completely degraded the monoglycosylated forms, whereas the diglycosylated forms were proteinase K resistant (Fig. 2C), indicating that the downstream portion of the monoglycosylated forms is exposed to the cytoplasmic side (Fig. 2D, h). In this Fig. 3. Effect of charge neutralization and charge density. (A) Asp or Ser experimental system, the full-length products of the 0K, 10K and residues were introduced into the 10K-cluster as indicated. The query 20K models became detectable after 14, 16 and 16–18 minutes of sequences were inserted upstream of the SADD sequence. The model proteins translation, respectively (supplementary material Fig. S1). The were translated as described in Fig. 2 and chased for the indicated time difference in the elongation rate does not explain the drastic period. (B) Translocation percentages were calculated. The experiments were differences in translocation. These data indicate that polypeptide performed twice and the mean values are presented. Insertion of five Asp chain movement is arrested by the positive-charge cluster trapped at residues into the 10K cluster suppressed translocation arrest and resulted in the membrane. The downstream region was on the cytoplasmic side efficient translocation as in the 0K model. Insertion of Ser partially suppressed the translocation arrest. Journal of Cell Science even after the polypeptide chain elongated up to the C-terminus, and then was translocated during the chase period (Fig. 2D). conditions. The model proteins were translated in the presence of The effect of charge neutralization and charge density RM for 20 minutes, and the translocation was then chased in the To examine the effect of charge neutralization and charge presence of cycloheximide under various ionic conditions density, Asp or serine (Ser) residues were introduced into the (Fig. 4). In the case of the 14K model, conversion of the 10K cluster (Fig. 3). The 10K cluster caused the accumulation of monoglycosylated form to the diglycosylated form was the full-length monoglycosylated form after 20 minutes of accelerated by 500 and 1000 mM NaCl. Lower ionic solutions translation. The monoglycosylated form was converted to the of 100 and 200 mM NaCl had no substantial effect on the rate of diglycosylated form. The 10K cluster used here was placed conversion to the diglycosylated form. Even in the case of the upstream of the SADD sequence. The effect of translocation 20K model, a substantial increase of the diglycosylated form was arrest was more apparent than when the 10K cluster was placed observed with the higher concentrations of NaCl. These results downstream of SADD sequence (Figs 1, 2). When five Asp were also support the conclusion that electrostatic interaction caused inserted into the 10K cluster, the product was diglycosylated as the translocation arrest independent of the flanking H-segment. efficiently as the 0K model. The effect of the 10K cluster was completely suppressed by the five Asp insertion. The 10 Asp The arrested K-cluster is on the cytoplasmic side of insertion produced the same result. When 10 Ser were the membrane introduced, the effect of the 10K cluster was largely, but not To examine membrane anchoring of the translocation-arrested completely, suppressed. The positive charge effect is neutralized polypeptide chain, the membranes were extracted under high salt by negative charges. The charge density is also crucial for the or alkaline conditions (Fig. 5). Under the high salt conditions, the arrest of translocation. products, irrespective of the glycosylation status, were recovered in the membrane precipitates, whereas the nonspecific 45-kDa The arrest by the K-cluster was suppressed in a product was in the soluble fraction. Under the alkaline high-salt environment conditions, in which the peripheral membrane proteins and The effect of the K-clusters on translocation is likely to be due to soluble proteins in the lumen are extracted into the supernatant simple electrostatic interactions. If this is the case, movement of (Fujiki et al., 1982), the 14K and 20K models were recovered in the full-length product should be accelerated under high salt the supernatant, whereas the monoglycosylated form of the 4188 Journal of Cell Science 124 (24)

Fig. 4. Effect of ionic strength on translocation rate. The 14K and 20K model proteins were translated in the presence of RM for 20 minutes. An equal volume of NaCl solution was added to produce the indicated salt concentration. The mixtures were further incubated and aliquots were removed at the indicated times and subjected to SDS-PAGE. Translocation percentages were calculated. The experiments were performed more than three times; values are means ± s.d.

16L4A model was in the membrane precipitate. The 16L4A model was anchored to the membrane by hydrophobic interaction, whereas translocation-arrested monoglycosylated forms of the 14K and 20K models were not anchored to the membrane. Journal of Cell Science To address the geometry of the translocation-arrested polypeptide chain on the membrane, we examined the distance between the 14K cluster and oligosaccharyl transferase active site by glycosylation site scanning (Fig. 6). The glycosylation site located 106 residues upstream of the 14K cluster was silenced and a new one was created 33 residues upstream of the 14K cluster. When the model protein was translated for 60 minutes, a nonglycosylated form was not observed and the glycosylation status was similar to that of the model used above, indicating that the 33-residue upstream site was accessible to oligosaccharyl transferase. Thirty-five percent of the product was translocated into the lumen and 65% of the product was stalled at the membrane. As the distance between the glycosylation site and the 14K cluster was shortened by serial deletions, the amount of the monoglycosylated form decreased and the amount of the Fig. 5. Alkaline extraction of the translocation-arrested polypeptide. nonglycosylated form increased (Fig. 6B,C); the distance (A,B) The model proteins with the indicated sequences were translated for producing half maximum monoglycosylation was 29–30 60 minutes and the reaction mixtures were treated with high-salt solution (A) residues from the 14K cluster. The distance of the or alkaline buffer (B). The mixture was subjected to ultracentrifugation to oligosaccharyl transferase active site from the hydrophobic TM separate membrane precipitates (P) and supernatant (S). These fractions and segment was estimated as a control. The monoglycosylation was the total translation product (T) were analyzed by SDS-PAGE. Diglycosylated decreased depending on the distance; the distance causing half forms (white circles) and monoglycosylated forms (black triangles) are indicated. Nonspecific products are indicated by asterisks. The maximum glycosylation was 18–19 residues for the 7L7A monoglycosylated forms of the 14K and 20K models were not anchored to the segment and 15–16 residues for the 16L4A segment. These membrane, whereas the 16L4A model was anchored in the membrane. The values are consistent with the value of 15 residues between the trace amount of an apparently diglycosylated form of the 16L4A-model in the oligosaccharyl transferase active site and the authentic TM P fraction was not examined further. segment (Nilsson and von Heijne, 1993; Popov et al., 1997). In Positive charges arrest translocation 4189

Fig. 6. Geometry of the translocation-arrested polypeptide chain. (A) To assess membrane topology of the lysine cluster (14K), a potential glycosylation site was introduced 33 residues upstream of the 14K cluster. The distance from the 14K cluster was altered by serial deletion, as indicated. As a control, the glycosylation site was placed at various points upstream of the H-segments (7L7A and 16L4A), as indicated. (B) The models were translated in vitro in the presence of RM for 60 minutes and analyzed by SDS-PAGE. Diglycosylated forms (white circles), monoglycosylated forms (black triangles) and nonglycosylated forms (white squares) are indicated. The distance between the glycosylation site and the query sequence is indicated at the top of the gel. The 14K cluster did not completely stop the translocation. Under the condition used, 40% of the nascent chain was fully translocated after the 60 minutes translation and consequently

Journal of Cell Science diglycosylated irrespective of the positions of glycosylation site. (C) Percentage of the monoglycosylated form in the nonglycosylated plus the monoglycosylated form was calculated. The experiments were performed twice and the mean values are presented. (D) The 14K cluster is more than 30 residues away from oligosaccharyl transferase (OSTase), whereas the authentic TM segment is 15 residues away from the enzyme. The hydrophilic 14K cluster is on the cytoplasmic side of the membrane.

the case of the short H-segment of 7L7A, an additional spacer Val320 that is 152 residues downstream of the 12K cluster. If should be required for glycosylation. We concluded that the the nascent chain was retained in the ribosome, the Cys(+145) positive charges were 30 residues away from the oligosaccharyl should be included in the ribosome tunnel, the Cys(+121) near transferase. A hydrophilic segment of approximately 10 residues the exit site of the ribosome and the Cys(+61) out side the is likely to span the translocon in an extended conformation, ribosome (Fig. 7B). When the RNAs with a termination codon whereas a hydrophobic TM segment of approximately 20 were translated in the presence of RM, all the Cys residues were residues spans the membrane in a helix conformation. These highly reactive; the reaction with PEGmal caused a shift up by data indicated that the 14K cluster was on the cytoplasmic side of 2 kDa (Fig. 7C, lanes 6, 12 and 18, and Fig. 7D). However, the membrane (Fig. 6D). when the truncated RNA was translated, Cys(+145) and Cys(+121) showed low reactivity (Fig. 7C, lanes 8 and 14, The arrested polypeptide chain is released from and 7D), whereas the Cys(+61) was highly reactive (Fig. 7C, the ribosome lane 2, and 7D). Immediately after the puromycin treatment, To examine whether the translocation-arrested polypeptide both of Cys(+145) and Cys(+121) became highly reactive. As a chain is released from the ribosome tunnel, we performed a control we performed the modification reaction with the nascent chemical modification experiment where the cysteine residue of chain synthesized in the absence of RM; the Cys(+61) fully- the nascent chain was reacted with a high-molecular mass exposed from the ribosome, the Cys(+121) at the exit site of the (2 kDa) reagent, polyethylene glycol maleimide (PEGmal), for a ribosome and the Cys(+145) retained in the ribosome were short period on ice (Fig. 7). A single cysteine (Cys) residue was highly, moderately and poorly reactive, respectively (Fig. 7C, added at the indicated position of the 12K model: 61, 121 and lanes 20, 22 and 26, and 7D). After puromycin treatment, the 145 residues downstream of the 12K cluster. To make a Cys(+121) and Cys(+145) became fully reactive. All data polypeptide chain that is retained in the ribosome as a ribosome- indicated that the nascent chain was fully released from the nascent chain complex, the RNA was truncated just after the ribosome tunnel during translocation arrest. 4190 Journal of Cell Science 124 (24)

Fig. 7. The nascent polypeptide chain was fully released from the ribosome during translocation arrest. (A) To assess reactivity with PEGmal, a single Cys residue was created at the indicated position of the Cys-less model protein. The positions relative to the 12K cluster of the Cys residue, glycosylation sites and C-terminus are indicated in parentheses. The 12K cluster was inserted between AS and SA, as in Fig. 3. (B) Geometry of Cys residues of the nascent chain. In the ribosome-nascent chain complex (RNC), Cys(+145) is in the ribosome and not reactive with PEGmal, whereas Cys(+121) is near the exit site and partially reactive. Cys(+61) is outside the ribosome and freely accessible to PEGmal. In the presence of a termination codon, the nascentchain was released from the ribosome immediately after synthesis and even the Cys(+145) reacted with PEGmal. (C) The RNAs, including those with a termination codon (Ter) or truncated after the Val codon (RNC and Puro) were translated in the presence or absence of RM for 20 minutes and subjected to PEGmal treatment on ice for 5 minutes. Where indicated, the nascent chain was released from the ribosome with puromycin (Puro) before the PEGmal reaction. The PEGylated form (red squares), monoglycosylated form (red triangles) and nonglycosylated form (open squares) are indicated. Nonspecific

Journal of Cell Science products are indicated by asterisks. (D) The efficiency of the PEGmal reaction. The experiments were repeated twice with each model protein and the mean PEGylation efficiency was plotted.

Translocation-arrested polypeptide was crosslinked with Discussion the Sec61a subunit Positive charges are known to determine the transmembrane To examine the environment of the translocation-arrested topology of membrane proteins. We demonstrated here that polypeptide that spanned the membrane, we performed a the positive charges of translocating nascent polypeptides chemical crosslinking experiment (Fig. 8). A single Cys residue temporarily arrest chain movement through the translocon even in was created at the fourth or fifth residue upstream of the 14K the absence of the H-segment. The effect depends on the number of cluster using a Cys-less model protein [Cys(–4) and Cys(–5), positive charges; in certain cases, the downstream portion of the respectively]. The models were translated in the presence of positive charge is retained on the cytoplasmic side of the membrane, RM and the crosslinking reaction was performed with a even after the nascent chain elongates up to the C-terminus and is homobifunctional crosslinker bismaleimidoethane (BMOE), the released from the ribosome. The translocation-arrested full-length crosslinking distance of which is 8 A˚ . Crosslinked products were nascent chain is then gradually translocated into the lumen. In the subjected to immunoprecipitation with anti-Sec61a antibody. A case of the 10K cluster, translocation of the majority of the nascent Cys(–5) gave a substantial crosslinked band of ,85 kDa that was chain was arrested 20 minutes after the start of translation, and it immunoreactive with anti-Sec61a antibody, and the Cys(–4) also was then moved through the translocon during the chase incubation. gave weak but distinct crosslinked form of the same molecular With the 20K cluster, the translocation was almost completely mass. Given that the nascent chain is ,40 kDa, the size of the arrested at 20 minutes of translation and little translocation was crosslinking partner is consistent with that of Sec61a. In the observed even after a chase of 100 minutes. The effect of the K- absence of the 14K cluster, the crosslinking with Sec61a was not clusters was neutralized by negatively charged residues inserted observed, indicating that the crosslinking was specific. These into the cluster and was greatly suppressed by high salt results demonstrated that the translocation-arrested polypeptide concentrations. The positive charges stall on the cytoplasmic side chain is adjacent to the translocon. of the membrane, probably at the ribosome–translocon junction. Positive charges arrest translocation 4191

establish the membrane topology of polypeptide chains. We found only two mammalian secretory proteins in a protein databases (human extracellular sulfatase Sulf-1 and human R- spondin-2) that have a cluster of 10 positive charges within a 12- residue window. Such a positive-charge cluster should be detrimental for translocation and thus excluded from secretory proteins during evolution. The arrest of polypeptide chain translocation causes various topogenic events. If the N-terminal side of an H-segment of a signal sequence is arrested, then the opposite side moves into the lumen. In this case, the positive charges only have to retain the N- terminal side until its downstream sequence comprising, at most, 30 residues emerges from the ribosome. After that, the H- segment is settled as the TM a-helix with an Ncyto orientation. If the C-terminus of an H-segment of a signal sequence is arrested on the cytoplasmic side until the opposite N-terminal side is flipped into the lumenal side, the H-segment becomes a TM helix with an Nlumen orientation. For example, in the case of synaptotagmin II, there are eight positive charges just after the Fig. 8. Crosslinking of translocation-arrested polypeptide chains with H-segment. When they are moved more than 20 residues Sec61a. A single Cys residue was added at the indicated position of downstream, they still affect the orientation of the H-segment, Cys-less 14K or 0K model proteins. The position relative to the 14K-cluster whereas those that are moved more than 30 residues downstream was indicated in the parenthesis. Supposed geometry of the Cys residues is no longer affect the orientation (Kida et al., 2006). In many cases, also shown. The model proteins were translated in the presence of RM for the TM topology of a signal sequence is likely to be determined 60 minutes, and subjected to BMOE treatment on ice for 1 hour (BMOE + within a short range. The positive charges flanking an H-segment lanes). Aliquots of the products were subjected to immunoprecipitation with enhance stop translocation (Jaud et al., 2009; Kuroiwa et al., anti-Sec61a antiserum (IP lanes). The bands immunoreactive with Sec61a 1990; Kuroiwa et al., 1991; Lerch-Bader et al., 2008). The TM antibody (arrowheads) and the nascent chains (triangle) are indicated. topology is achieved as soon as the positive charges reach the translocon. There is a considerable amount of data to support the Our data strongly suggested that a simple electrostatic contributions of positive charges to the membrane topology of H- interaction causes the translocation arrest independent of segments, as indicated by the positive-inside rule (Sipos and von flanking hydrophobic sequences. The Coulomb force on the Heijne, 1993; von Heijne and Gavel, 1988). The positive charges positive charges increases the energy barrier for the nascent chain have been discussed only in relation to the flanking H-segment to move into the translocon and consequently decreases and have been recognized as a topology modulator of H- frequency of the forward movement. The possible interaction Journal of Cell Science segments. Recently, we found a longer range effect of positive partners are translocon subunits, ribosomal subunits, ribosomal charges than previously detected (Fujita et al., 2010). A positive- RNA and charges of the membrane lipids. Goder et al. have charge cluster located more than 60 residues downstream of a indicated that charged residues of Sec61p of the yeast translocon marginal H-segment can contribute to its membrane spanning contribute to the positive-inside rule (Goder et al., 2004). The (Fujita et al., 2010). After the upstream H-segment passes charged amino acid residues within the Sec61a subunit might be through the translocon, the positive charges at the 60 residues involved in the interaction. In the ribosome, the positive charges downstream arrests the translocation and the H-segment slides of the nascent chain might slow down polypeptide chain back into the translocon. In the present study, we found that the elongation through electrostatic interaction with negative positive charges arrest polypeptide chain movement even in the charges of the ribosome tunnel (Lu and Deutsch, 2008). In the absence of an H-segment. Tentative translocation arrest by present study, the effect of the positive charges on the protein positive charges provides the nascent chain in the translocon with synthesis rate was not drastic; similar amounts of the 0K model a time interval that might allow the polypeptide segments to and the 20K model were observed after 20 minutes of translation move back and forth, reorient and assemble with each other. (Fig. 2 and supplementary material Fig. S1). In contrast to the Our findings suggest that polypeptide chain movement is time range of the translation delay, the positive charge caused a arrested during cotranslational translocation and that the forward much more drastic delay in translocation, which resulted in movement resumes after completion of elongation. The positive accumulation of translocation-arrested full-length polypeptide. charges stall at the translocon, even in the presence of the For the accumulation of the monoglycosylated full-length pushing force of the chain elongation, and can then continue to polypeptide, the translocation must be arrested until the slowly advance in the absence of the pushing force. Even if the downstream polypeptide chain of 140 residues is synthesized. chain is not elongating, translocation tendency generated in the The 6K cluster induced substantial accumulation of arrested full- lumen should similarly contributes to the movement, e.g. binding length product. We expect that such a translocation arrest also of chaperones, folding of the polypeptide chain, glycosylation occurs with only a few positive charges, although they might not and disulfide bond formation, etc. In addition, there might be induce a much accumulation of the full-length polypeptide. favorable conditions for translocation after termination of the Considering the time scales involved in polypeptide chain translation. The cotranslational forward movement is driven by synthesis and topology determination of the nascent chain, chain elongation on the ribosome but the rate is limited by the short and temporary translocation arrest should be sufficient to elongation rate. The degree of freedom of the polypeptide chain 4192 Journal of Cell Science 124 (24)

should be restricted during chain elongation. However, after the adjust the final concentrations of salt and then incubated at 30˚C for the indicated time. For alkaline or high salt extraction, the samples were treated as described polypeptide chain is released from the peptidyl transferase center previously (Ikeda et al., 2005); briefly, the translation mixtures were treated with of ribosome, the nascent chain would be unrestricted and could high salt (500 mM NaCl) or alkali (100 mM NaOH) and subjected to thus fluctuate largely and freely. Such free motion might ultracentrifugation for 5 minutes at 4˚C (Hitachi S100AT4 Rotor, 50,000 r.p.m.), occasionally overcome the arrest caused by positive charges. to separate membrane precipitates and supernatant. For a PEGmal reactivity assay, the RNAs were translated in the presence of RM for 20 minutes at 30˚C. To stop the translation, cycloheximide (2 mM) or Materials and Methods puromycin (2 mM) was added and the reaction mixture immediately chilled on ice. Materials Reaction with PEGmal was performed essentially as described (Kida et al., 2010). Rough microsomal membrane (RM) (Walter and Blobel, 1983) and rabbit Then 1 ml 100 mM PEGmal dissolved in water was added to the mixture (9 ml) reticulocyte lysate (Jackson and Hunt, 1983) were prepared as previously and incubated on ice for 5 minutes. The maleimide reaction was quenched with described. RM was treated with EDTA and then with Staphylococcus aureus 1 ml 100 mM dithiothreitol (DTT), the mixture was treated with RNaseA and nuclease as previously described (Walter and Blobel, 1983). Castanospermine 20 ml SDS-PAGE sample buffer containing 100 mM DTT was added. (Merck, Tokyo, Japan), proteinase K (Merck, Tokyo, Japan), RNaseA (Wako, For chemical crosslinking, the RNAs were translated in the presence of RM for Osaka, Japan), DNA manipulating enzymes (Takara and Toyobo, Tokyo, Japan), 60 minutes and the reaction was terminated with 2 mM cycloheximide. The 2 kDa PEGmal (Sunbright ME-020MA, NOF Corporation, Tokyo, Japan), chemical crosslinking was performed essentially as previously described (Kida bismaleimidoethane (BMOE, Thermo Scientific, Yokohama, Japan), puromycin et al., 2007). The products were treated on ice for 60 minutes with 10 mM BMOE (Sigma, Tokyo, Japan) and cycloheximide (Sigma, Tokyo, Japan) were obtained or its solvent dimethyl sulfoxide. The crosslinking reaction was quenched for from the indicated sources. 10 minutes on ice with a sixfold volume of dilution buffer (30 mM HEPES, pH7.4, 150 mM potassium acetate, 2 mM magnesium acetate) containing 20 mM DTT. Construction of model proteins For immunoprecipitation, RM were isolated by centrifugation at 100,000 g for In the following DNA construction procedure, DNA fragments were obtained by 5 minutes, solubilized with 1% SDS for 5 minutes at 95˚C, and then diluted with a PCR using primers including the appropriate restriction enzyme site (indicated in 30-fold volume of immunoprecipitation buffer [1% Triton X-100, 50 mM Tris- parentheses). The fragments were subcloned into plasmid vectors that had been HCl, pH7.5, 150 mM NaCl]. After a centrifugation at 20,000 g for 5 minutes, digested with the restriction enzymes. At each junction, the six bases of the the supernatant was incubated for 30 minutes with protein-A–Sepharose (GE restriction enzyme site were designed to encode two codons. Point mutations were Healthcare) alone to remove materials nonspecifically bound to resin. The introduced using the method of Kunkel (Kunkel, 1985) or the QuickChange unbound fractions were incubated for 2 hours with anti-Sec61a antiserum and then procedure (Stratagene, La Jolla, CA). All the constructed DNAs were confirmed with protein-A–Sepharose for 14 hours. The resin was washed three times with by DNA sequencing. immunoprecipitation buffer and the isolated proteins were eluted by boiling with The model protein, based on RSA, was previously described (Fujita et al., 2010). SDS-PAGE sample buffer. Briefly, potential glycosylation sites were created at Asn67 and Asn259, each query sequence was introduced in the middle portion and the Val and termination Acknowledgements codon were inserted after H319 (Fig. 1). Various query sequences, as indicated in the figures, were inserted into the middle portion of the model protein using the We thank Shigeki Mitaku and Runcong Ke (Nagoya University) for QuickChange procedure. The model proteins shown in Fig. 3 were created by instruction on database searching. inserting synthetic oligonucleotides encoding the indicated sequences between AS and SA (NheI–Aor51HI). For the glycosylation site scanning constructs (Fig. 6), Funding the upstream regions of the query sequence were serially deleted or glycosylation This work was supported by Grants-in-Aid for Scientific Research sites were newly created by the mutations (T149SF to N149ST, S150FQE to N150STA, F151QE to N151ST, Q152ENP to N152STA, E153NP to N153ST, E154NPT from the Ministry of Education, Culture, Sports, Science and to T154NST, N155PTS to A155NST and P156TSF to A156NST). The newly created Technology of Japan, and Japan Society for the Promotion of potential sites were confirmed to be efficiently glycosylated in the absence of the Science [19058013, 20370041 and 23370055 to M.S.; 21770123 and query sequences. For chemical modification of the model proteins with PEGmal 23770151 to Y.K.; 23-6629 to H.F.]; the Global COE program; the Journal of Cell Science (Fig. 7), a single Cys residue was included at Cys229 Cys284 or Cys313 and DNA Sumitomo Foundation [grant number 080178 to Y.K.]; and the fragment encoding the 12K cluster was inserted between AS and SA (NheI– Aor51HI). For the construction, we exploited the Cys-less model protein which had Hyogo Science and Technology Association [grant number 229035 glycosylation site 12 residues downstream of the 12K cluster. To make truncated to Y.K.]. mRNA, the restriction enzyme site (BamHI) was created just after the Val320 codon. The models for crosslinking (Fig. 8), a single Cys was included at Cys164 Supplementary material available online at for Cys(–5) or Cys165 for Cys(–4) and DNA fragment encoding the 14K cluster was inserted between AS and SA (NheI–Aor51HI). http://jcs.biologists.org/lookup/suppl/doi:10.1242/jcs.086850/-/DC1

In vitro synthesis and membrane translocation References Cell-free transcription and translation were performed essentially as described Fujiki, Y., Hubbard, A. L., Fowler, S. and Lazarow, P. B. (1982). Isolation of previously (Sakaguchi et al., 1992), except that translation reactions with RM intracellular membranes by means of sodium carbonate treatment: application to contained 20 mg/ml castanospermine to prevent trimming of the sugar chain, endoplasmic reticulum. J. Cell Biol. 93, 97-102. which results in heterogeneity of the products. The plasmids harboring RSA model Fujita, H., Kida, Y., Hagiwara, M., Morimoto, F. and Sakaguchi, M. (2010). Positive proteins were linearized with XhoIorBamHI. The plasmids were then transcribed charges of translocating polypeptide chain retrieve an upstream marginal hydrophobic with T7-RNA polymerase. The RNA was translated in the reticulocyte lysate cell- segment from the endoplasmic reticulum to the translocon. Mol. Biol. Cell 21, 2045- 2056. free system for 1 hour at 30 C in either the absence or presence of RM. The ˚ Goder, V. and Spiess, M. (2001). Topogenesis of membrane proteins: determinants and translation reaction included 90 mM potassium acetate, 1.2 mM magnesium dynamics. FEBS Lett. 504, 87-93. acetate, 32% reticulocyte lysate, 20 mg/ml castanospermine and 15.5 kBq/ml Goder, V., Junne, T. and Spiess, M. (2004). Sec61p contributes to signal sequence EXPRESS 35S protein-labeling mix (Perkin Elmer, Waltham, MA). Where orientation according to the positive-inside rule. Mol. Biol. Cell 15, 1470-1478. indicated, 2 mM cycloheximide was included at the indicated time points. After Hessa, T., Kim, H., Bihlmaier, K., Lundin, C., Boekel, J., Andersson, H., Nilsson, I., the translation reaction, aliquots were treated with proteinase K (200 mg/ml) for White, S. H. and von Heijne, G. (2005). Recognition of transmembrane helices by 1 hour on ice. Radiolabeled polypeptide chains were analyzed by SDS-PAGE, the endoplasmic reticulum translocon. Nature 433, 377-381. visualized with an imaging analyzer (BAS1800, Fuji Film) and quantified using Hessa, T., Meindl-Beinker, N. M., Bernsel, A., Kim, H., Sato, Y., Lerch-Bader, M., Image Gauge software (Fuji Film). Where truncated mRNA was translated, the Nilsson, I., White, S. H. and von Heijne, G. (2007). Molecular code for product was treated with RNase (42 mg/ml) before SDS-PAGE to completely transmembrane-helix recognition by the Sec61 translocon. Nature 450, 1026-1030. remove tRNA from the nascent chain. Ikeda, M., Kida, Y., Ikushiro, S. and Sakaguchi, M. (2005). Manipulation of Where indicated, the translation reactions were terminated by the addition of membrane protein topology on the endoplasmic reticulum by a specific ligand in cycloheximide (2 mM) and were then further incubated for the indicated times. At living cells. J. Biochem. 138, 631-637. each time point, aliquots were removed and proteins were precipitated by TCA, Jackson, R. J. and Hunt, T. (1983). Preparation and use of nuclease-treated rabbit and then analyzed by SDS-PAGE. Other aliquots were treated by proteinase K, in reticulocyte lysates for the translation of eukaryotic messenger RNA. Methods the absence or presence of detergent (Triton X-100) as previously describe (Sakaki Enzymol. 96, 50-74. et al., 1999). For chase reaction in the presence of NaCl, translation reaction Jaud, S., Ferna´ndez-Vidal, M., Nilsson, I., Meindl-Beinker, N. M., Hu¨bner, N. C., mixture was diluted with an equal volume of appropriate NaCl buffer solution to Tobias, D. J., von Heijne, G. and White, S. H. (2009). Insertion of short Positive charges arrest translocation 4193

transmembrane helices by the Sec61 translocon. Proc. Natl. Acad. Sci. USA 106, Popov, M., Tam, L. Y., Li, J. and Reithmeier, R. A. (1997). Mapping the ends of 11588-11593. transmembrane segments in a polytopic membrane protein. Scanning N-glycosylation Johnson, A. E. and van Waes, M. A. (1999). The translocon: a dynamic gateway at the mutagenesis of extracytosolic loops in the anion exchanger, band 3. J. Biol. Chem. ER membrane. Annu. Rev. Cell Dev. Biol. 15, 799-842. 272, 18325-18332. Kida, Y., Morimoto, F., Mihara, K. and Sakaguchi, M. (2006). Function of positive Rapoport, T. A. (2007). Protein translocation across the eukaryotic endoplasmic charges following signal-anchor sequences during translocation of the N-terminal reticulum and bacterial plasma membranes. Nature 450, 663-669. domain. J. Biol. Chem. 281, 1152-1158. Sakaguchi, M. (1997). Eukaryotic protein . Curr. Opin. Biotech. 8, 595-601. Kida, Y., Morimoto, F. and Sakaguchi, M. (2007). Two translocating hydrophilic Sakaguchi, M., Hachiya, N., Mihara, K. and Omura, T. (1992). Mitochondrial porin segments of a nascent chain span the ER membrane during multispanning protein can be translocated across both endoplasmic reticulum and mitochondrial membranes. topogenesis. J. Cell Biol. 179, 1441-1452. J. Biochem. 112, 243-248. Kida, Y., Kume, C., Hirano, M. and Sakaguchi, M. (2010). Environmental transition Sakaki, K., Sakaguchi, M., Ota, K. and Mihara, K. (1999). Membrane perturbing of signal-Anchor sequences during membrane insertion via the endoplasmic reticulum factor in reticulocyte lysate, which is transiently activated by proteases. FEBS Lett. translocon. Mol. Biol. Cell 21, 418-429. 454, 345-348. Kunkel, T. A. (1985). Rapid and efficient site-specific mutagenesis without phenotypic Sipos, L. and von Heijne, G. (1993). Predicting the topology of eukaryotic membrane selection. Proc. Natl. Acad. Sci. USA 82, 488-492. proteins. Eur. J. Biochem. 213, 1333-1340. Kuroiwa, T., Sakaguchi, M., Mihara, K. and Omura, T. (1990). Structural requirements for interruption of protein translocation across rough endoplasmic Tsukazaki, T., Mori, H., Fukai, S., Ishitani, R., Mori, T., Dohmae, N., Perederina, reticulum membrane. J. Biochem. 108, 829-834. A., Sugita, Y., Vassylyev, D. G., Ito, K. et al. (2008). Conformational transition of Kuroiwa, T., Sakaguchi, M., Mihara, K. and Omura, T. (1991). Systematic analysis Sec machinery inferred from bacterial SecYE structures. Nature 455, 988-991. of stop-transfer sequence for microsomal membrane. J. Biol. Chem. 266, 9251-9255. Van den Berg, B., Clemons, W. M., Jr, Collinson, I., Modis, Y., Hartmann, E., Lerch-Bader, M., Lundin, C., Kim, H., Nilsson, I. and von Heijne, G. (2008). Harrison, S. C. and Rapoport, T. A. (2004). X-ray structure of a protein-conducting Contribution of positively charged flanking residues to the insertion of transmem- channel. Nature 427, 36-44. brane helices into the endoplasmic reticulum. Proc. Natl. Acad. Sci. USA 105, 4127- von Heijne, G. and Gavel, Y. (1988). Topogenic signals in integral membrane proteins. 4132. Eur. J. Biochem. 174, 671-678. Lu, J. and Deutsch, C. (2008). Electrostatics in the ribosomal tunnel modulate chain Walter, P. and Blobel, G. (1983). Preparation of microsomal membranes for co- elongation rates. J. Mol. Biol. 384, 73-86. translational protein translocation. Methods Enzymol. 96, 84-93. Nilsson, I. M. and von Heijne, G. (1993). Determination of the distance between the Whitley, P., Nilsson, I. M. and von Heijne, G. (1996). A nascent secretory protein may oligosaccharyltransferase active site and the endoplasmic reticulum membrane. J. traverse the ribosome/endoplasmic reticulum translocase complex as an extended Biol. Chem. 268, 5798-5801. chain. J. Biol. Chem. 271, 6241-6244. Journal of Cell Science