Interaction mapping of the Sec61 translocon identifies two Sec61α regions interacting with hydrophobic segments in translocating chains

Many proteins in organelles of the secretory pathway, as well as secretory proteins, are translocated across and inserted into the endoplasmic reticulum membrane by the Sec61 translocon, a protein-conducting channel. The channel consists of 10 transmembrane (TM) segments of the Sec61α subunit and possesses an opening between TM2b and TM7, termed the lateral gate. Structural and biochemical analyses of complexes of Sec61 and its ortholog SecY have revealed that the lateral gate is the exit for signal sequences and TM segments of translocating polypeptides to the lipid bilayer and also involved in the recognition of such hydrophobic sequences. Moreover, even marginally hydrophobic (mH) segments insufficient for membrane integration can be transiently stalled in surrounding Sec61α regions and cross-linked to them, but how the Sec61 translocon accommodates these mH segments remains unclear. Here, we used Cys-scanned variants of human Sec61α expressed in cultured 293-H cells to examine which channel regions associate with mH segments. A TM segment in a ribosome-associated polypeptide was mainly cross-linked to positions at the lateral gate, whereas an mH segment in a nascent chain was cross-linked to the Sec61α pore-interior positions at TM5 and TM10, as well as the lateral gate. Of note, cross-linking at position 180 in TM5 of Sec61α was reduced by an I179A substitution. We therefore conclude that at least two Sec61α regions, the lateral gate and the pore-interior site around TM5, interact with mH segments and are involved in accommodating them.

Many proteins in organelles of the secretory pathway, as well as secretory proteins, are translocated across and inserted into the endoplasmic reticulum membrane by the Sec61 translocon, a protein-conducting channel. The channel consists of 10 transmembrane (TM) segments of the Sec61␣ subunit and possesses an opening between TM2b and TM7, termed the lateral gate. Structural and biochemical analyses of complexes of Sec61 and its ortholog SecY have revealed that the lateral gate is the exit for signal sequences and TM segments of translocating polypeptides to the lipid bilayer and also involved in the recognition of such hydrophobic sequences. Moreover, even marginally hydrophobic (mH) segments insufficient for membrane integration can be transiently stalled in surrounding Sec61␣ regions and cross-linked to them, but how the Sec61 translocon accommodates these mH segments remains unclear. Here, we used Cysscanned variants of human Sec61␣ expressed in cultured 293-H cells to examine which channel regions associate with mH segments. A TM segment in a ribosome-associated polypeptide was mainly cross-linked to positions at the lateral gate, whereas an mH segment in a nascent chain was cross-linked to the Sec61␣ pore-interior positions at TM5 and TM10, as well as the lateral gate. Of note, cross-linking at position 180 in TM5 of Sec61␣ was reduced by an I179A substitution. We therefore conclude that at least two Sec61␣ regions, the lateral gate and the poreinterior site around TM5, interact with mH segments and are involved in accommodating them.
Many proteins in the secretory pathway and the plasma membrane, as well as secretory proteins, are cotranslationally translocated across and inserted into the endoplasmic reticulum (ER) 3 membrane via a protein-conducting channel called a translocon and then transferred to their final destinations by vesicular transport. Nascent polypeptide chains containing hydrophobic signal sequences are targeted to the ER by a signal recognition particle and its receptor on the ER (1). Signal sequences are also recognized at the translocon, and translocation of the following parts is initiated by their membrane insertion (2)(3)(4)(5)(6)(7). During cotranslational translocation, a transmembrane (TM) segment in a translocating nascent membrane protein stalls at the translocon and laterally moves into the lipid bilayer. The Sec61 complex, consisting of three subunits, ␣, ␤, and ␥, is the core channel of the translocon. Structural analyses of the Sec61 complex (and their ortholog archaeal and bacterial SecY complexes) revealed that 10 TM segments of Sec61␣ form a polypeptide conduit (8 -10). The channel can also open to the lipid environment between TM2b and TM7, which is termed the lateral gate. More recent structures of Sec61/SecY complexes with translocating substrates showed that signal sequences are sandwiched between these two TM segments (11,12). These structures and cross-linking between Sec61␣ and signal sequences (13,14) indicated that the lateral gate works as an exit for signal sequences to the lipid phase. In addition, prl mutants of SecY (15)(16)(17) and mutagenesis of yeast Sec61␣ (18 -20) revealed that mutations at the lateral gate, as well as the inside regions including constricted positions of the pore (called a pore ring) and TM2a (called a plug segment), cause the channel to open even by insufficient signal sequences, suggesting that these regions are critical for recognizing signal sequences for opening the channel.
Structural (21) and mutagenesis (20,22) analyses of the Sec61 channel indicated that the lateral gate and pore inside are also involved in recognition of hydrophobic (H) segments in translocating polypeptide chains. Moreover, the amino acid composition of H-segments required for translocation stalling at the translocon revealed that direct interaction with the lipid bilayer through the lateral gate mediates recognition and insertion of H-segments (23,24). The majority of TM segments possess sufficient hydrophobicity, and they must be smoothly sorted into the lipid phase via these recognition steps. In contrast, marginally hydrophobic (mH) TM segments, whose hydrophobicity is insufficient for membrane integration, are frequently found in multipass membrane proteins. Membrane insertion of such mH segments is supported by the following TM segments forming an N lumen /C cyto orientation (25)(26)(27)(28). Additionally, synthetic and natural mH segments can be stalled in the membrane (29,30), and the stalling is enhanced by the following positively charged residues (31,32). Site-specific cross-linking and chemical modification of stalling mH segments indicated that such  cro ARTICLE segments are in an aqueous location and surrounded with Sec61␣ regions (30). How the Sec61 channel accommodates these mH segments, however, remained unclear.
In the present study, to explore which channel regions associate with stalling mH segments, we used Cys-scanned variants of human Sec61␣ expressed in cultured 293-H cells. Although a TM segment in a ribosome-associated polypeptide was mainly cross-linked to positions at the lateral gate, an mH segment in a nascent chain was cross-linked to the pore-interior positions in TM5 and TM10, as well as the lateral gate. Because the crosslinking at position 180 in TM5 was reduced by an I179A substitution, we conclude that at least two regions, the lateral gate and the pore-interior site around TM5, are interactive sites for the accommodation of mH segments.

The FLAG-tagged Sec61 channel is functionally expressed in 293-H-derived microsomes
To determine the interactive sites in Sec61␣ with a translocating chain, eight endogenous Cys residues of human Sec61␣ (SEC61A1) were replaced with Ala, and 16 positions (eight original and eight new positions) were replaced with Cys (Fig. 1A). Several positions are the sites associated with a signal sequence (14), and others are in TM segments forming the pore. Stable cell lines of human 293-H cells expressing FLAG-tagged Sec61␣ proteins were established, and rough microsomal membranes (RMs) were prepared from the cells. Although expression levels varied, all FLAG-Sec61␣ proteins were expressed in RMs (Fig. 1B). When a secretory protein contain-ing an N-glycosylation site was synthesized in a cell-free translation system, it was glycosylated in a manner dependent on the dose of 293-H-derived RMs containing Sec61␣ variants (data not shown); therefore, a sufficient amount of each RM was used in the following translation reactions.
We first examined the cross-linking of Sec61␣ variants to a hydrophilic chain using a bifunctional cross-linker, bis(maleimido)ethane. Translocation of a streptavidin (SAv)-binding peptide-tagged N-terminal domain of a type I signal-anchor of synaptotagmin II was stalled in the presence of SAv, and the hydrophilic chain was cross-linked to Sec61␣ in canine pancreatic RMs (33). The FLAG-Sec61␣ was cross-linked to the stalled hydrophilic chain in an SAv-dependent manner (Fig.  2B), suggesting that the expressed Sec61␣ is functional in the RM. All 1Cys proteins were also cross-linked to the stalled chain as was endogenous Sec61␣ in the same RMs (Fig. 2C), but the percentages of cross-linking were diverse (Fig. 2D). This indicates that these Sec61␣ variants form a functional channel through which the hydrophilic chain passes.
Next, we examined the cross-linking to a TM segment forming an N cyto /C lumen orientation ( Fig. 3) that was also crosslinked to canine Sec61␣ (34). The profile of cross-linking percentages was considerably different from that of the above hydrophilic chain (Fig. 3C versus 2D), and positions 13 (in the N-terminal region) and 95 (in TM2b) were well cross-linked to the TM segment. The efficiencies, normalized by expressing extents of Sec61␣ 1Cys proteins in supplementing RMs (see "Experimental procedures" for quantification and calculation), indicate that positions 13, 85 (in TM2b), 95, and 383 (at the C terminus of TM8) were efficiently cross-linked to the TM segment, consistent with previous data showing that positions 13, 95, and 383 are cross-linked to a signal sequence trapped in the channel (14). Our data suggest that the TM segment stands by the lateral gate before moving into the lipid bilayer.

Stalling mH segments interact with the pore-interior site at TM5 and TM10 as well as with the lateral gate
We previously demonstrated that designed mH segments in ribosome-bound nascent chains are stalled in surrounding Sec61␣ regions and cross-linked to them (30). Therefore, we next examined which Cys positions in Sec61␣ are cross-linked to mH segments inserted in a secretory protein, rat serum albumin (RSA). The 19A1C segment consists of 19 Ala and 1 Cys, and 2-4 Ala in the segment were replaced with either Leu or Ser to render it more or less hydrophobic (Fig. 4A). Crosslinked products between Sec61␣ (ϳ52 kDa) and nascent chains of RSA containing mH segments (ϳ35 kDa) revealed an unexpectedly large size of 100 kDa or more (Fig. 4B). Because the mobility of endogenous Sec61␣ and its recombinant 1Cys proteins was similarly shifted by the cross-linking and cross-linked products of substrate chains bound to and released from ribosomes showed a similar size (see Fig. 7), such a large size of the conjugates was due to neither further cross-linking nor attached tRNAs. Our previous study revealed that a substrate chain of the same length (321 residues without the signal peptide) also resulted in ϳ100-kDa products cross-linked to canine Sec61␣, whereas cross-linked products of a shorter chain (252

Cys-scanning analysis of Sec61␣
residues) showed the expected size, ϳ70 kDa (30). Therefore, the region between residues 251 and 319 of RSA might possess an effect that decreases the mobility of the conjugates in SDS-PAGE. Although the 4L1C segment, which was the most hydrophobic segment here, was least cross-linked to Sec61␣ variants among the mH segments because of moving into the lipid phase, it was fairly cross-linked to position 95 at the lateral gate (Fig. 4C, bottom panel). The other segments were strongly cross-linked to positions 85, 95, and 383 at the lateral gate, indicating that the lateral gate is interactive even with the segments with lower hydrophobicity. Additionally, positions 13, 163 (in TM4), and 453 (in TM10) of the channel were slightly, but meaningfully cross-linked to the 2S1C and 19A1C seg-ments but not the 2L1C segment possessing higher hydrophobicity (Fig. 4D). The normalized efficiency also reveals that position 294 in TM7 was fairly cross-linked to mH segments, whereas the cross-linking percentage was considerably low (Fig. 4C).
We next focused on position 453 in TM10 because it is located opposite to the lateral gate. Further Cys scanning at TM10 (positions 452 and 454) and TM5 (positions 179 and 180) near position 453 was performed (Fig. 5). Among them, positions 179 and 180 were similarly or more efficiently crosslinked to 19A1C and 4S1C segments than positions 85, 95, and 383 at the lateral gate. Also, position 452 was considerably cross-linked to the segments with lower hydrophobicity. These

Cys-scanning analysis of Sec61␣
data suggest that these inside positions, as well as the lateral gate, are in contact with stalling mH segments.
According to structural information (11), Ile-179 is a part of the pore ring and appears to protrude into the pore in the opened channel (Fig. 6A). We thus substituted Ile-179 and/or the flanking Ile-453 with Ala in the Sec61␣ 1Cys 180 variant. These mutants were also cross-linked to the hydrophilic chain stalled with SAv (Fig. 6, C and D). The cross-linking to the 19A1C segment was reduced by the I179A substitution. Because the cross-linking efficiency of the I179A/I453A variant differed little from that of the I179A variant, the I179A substitution was critical in reducing the cross-linking. These findings suggest that the pore-interior site around TM5 interacts with the stalling mH segment and that Ile-179 has an important role in the interaction.

Translational states affect the accommodation of the mH sequence within Sec61␣
The mH segment can be cross-linked to canine Sec61␣ only in a ribosome-bound state (30). We examined the cross-linking of the 19A1C segment in a nascent chain released from a ribosome to the Cys positions found in the present study (Fig. 7). In the released chain, the cross-linking to all positions was reduced (Fig. 7, B and C), suggesting that the mH segment left

Cys-scanning analysis of Sec61␣
the Sec61 channel after translation termination. Because the mH segment is retained in an aqueous environment after translation termination (30), it might be associated with another protein(s) outside the channel.

Discussion
In the present study, we explored the interactive sites in Sec61␣ with translocating segments whose cross-linking was observed previously (30,34). Site-specific cross-linking analy-

Cys-scanning analysis of Sec61␣
ses using Sec61␣ Cys-scanned variants revealed the accessibility between stalling mH segments and the pore-interior site at TM5 and TM10, as well as the lateral gate.
According to several structures of Sec61/SecY with substrates, the lateral gate opens by TM2b movement away from TM7 and a signal sequence or a TM segment in translocating polypeptides is sandwiched between these TM segments (11,12,21). This led to the hypothesis that TM2b functions as a competitor of H-segments and that the hydrophobicity and length of TM2b create a threshold for exiting via the gate. Site-specific cross-linking analyses also indicated that signal sequences stand by the lateral gate just before moving into the lipid phase (13,14). H-segments sufficient for integration into the lipid bilayer are presumably stalled at the lateral gate and immediately sorted into the lipid phase through the gate. Conversely, mH segments insufficient for membrane integration are also stalled at the translocon (29 -31). Here, our findings indicate that stalling mH segments flank at least two regions,  Fig. 3D. IP, immunoprecipitation; T, total; 61, Sec61␣; F, FLAG.

Cys-scanning analysis of Sec61␣
the lateral gate and the pore-interior site at TM5 and TM10, and the segments with lower hydrophobicity are apt to be well cross-linked to the pore interior. Structural and biochemical studies of the Sec61/SecY channel indicated that TM5 and TM10 face the path of translocating chains (12,21,35). It is thus suggested that stalling mH segments interact with the pore interior of the Sec61 channel. Furthermore, the pore ring of Sec61␣ in the closed state is formed with TM2b (Ile-81 and Val-85), TM5 (Ile-179 and Ile-183), TM7 (Ile-292), and TM10 (Leu-449), similar to archaeal and bacterial SecY (11). The hydrophobic patch, which is thought to be an initial binding site for signal sequences, is also formed with TM2b (Val-85 and Leu-89), TM5 (Ile-179), and TM7 (Ile-293). Because the I179A substitution reduced the cross-linking at position 180 to the 19A1C segment, there might be a "sticky" site around positions 179 and 180 that interacts with hydrophobic portions in translocating chains even after the channel opens. However, it is also possible that such pore-interior positions only flank mH segments binding the lateral gate and that a conformational change caused by mutation at Ile-179 only influences the accessibility of the region to passing polypeptide chains.
Nevertheless, the findings in this study suggest that stalling mH segments are accommodated at the pore inside of the Sec61 channel besides at the lateral gate. The translocation stalling and partitioning into the lipid phase of H-segments occur gradually as the hydrophobicity of the H-segments increases (30,

Cys-scanning analysis of Sec61␣
36), and mH segments are not stably retained in the channel but rather fluctuate between the channel and the luminal space. The channel pore may act as an interchange between polar and apolar environments, and an mH TM segment in multipass proteins may be temporarily accommodated within the pore until its membrane integration is performed by interaction and sequestration with other TM segments (37). Also, the translocation pausing of apolipoprotein B occurs in the Sec61 channel (38,39). The SecY structure in complex with a nascent secretory protein revealed that the translocating chain is stalled, and a part of the chain forms a loop on the cytoplasmic side (40). These findings indicate that the Sec61/SecY pore is not smooth but possesses a viscous property for translocating chains, and the pore-interior site around TM5 might be involved in such a property of the channel.
In the present study, we examined the molecular mechanism of the Sec61 channel using its Cys-scanned mutants and translocation-stalling segments as an interactive probe. Although the Sec61 mutants may possess local conformational perturbations caused by substitution of endogenous Cys residues to Ala and further introduction of Cys, they are expressed in the ER membrane and functional in polypeptide translocation (Fig. 2). Because detection of delicate associations between the channel and substrate polypeptide chains by structural analyses is considered difficult, our strategy is useful to explore the nature of the translocon pore in the active state. More detailed analyses using recombinant channels to understand the properties in an active state, in addition to structural and mutagenesis analyses, are required.

Constructs
Human Sec61␣ (SEC61A1) cDNA was obtained by RT-PCR of human fetal brain total RNA (Agilent) and subcloned into pBICEP-CMV-2 vector (Sigma; between EcoRI and EcoRV restriction sites) to create the pBICEP-Sec61␣ plasmid encoding N-terminally 3ϫFLAG-tagged Sec61␣. Cys elimination from Sec61␣ by substitution with Ala was performed using the method of Kunkel (41). Further substitutions of target positions to Cys and/or Ala were performed using an inverse PCRrelated method. The constructed DNAs were confirmed by DNA sequencing. DNA constructs encoding "S-I-II" fusion proteins of the SAv-binding peptide tag (S), type I signal-anchor from mouse synaptotagmin II (I), and TM3 of human NHE6 (II) were described previously (34). Also, constructs encoding truncated RSAs containing mH segments were described previously (30). Information for all of the oligo DNAs used in this study is available from the authors.

Cell culture and transfection
293-H cells (Sigma) were cultured in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum under 5% CO 2 at 37°C. Plasmid transfection was performed using FuGENE 6 transfection reagent (Promega) or PEI MAX (Polysciences) as recommended by the manufacturers. Stable cell lines transfected with pBICEP-Sec61␣ plasmids were selected in the presence of 1 mg/ml G-418.

Preparation of microsomes from cultured cells and immunoblotting
Cells stably expressing Sec61␣ recombinant proteins were collected, suspended in homogenization buffer (50 mM triethanolamine, pH 7.4, 250 mM sucrose, 50 mM potassium acetate, 1 mM magnesium acetate, 2 mM EDTA, 1 mM DTT, 0.5 mM phenylmethylsulfonyl fluoride, and cOmplete TM EDTA-free protease inhibitor mixture (Merck)), and homogenized using a Potter-Elvehjem homogenizer. After centrifugation of the homogenate at 10,000 ϫ g, the supernatant was placed on a cushion solution (1.2 M sucrose in homogenization buffer without EDTA) and ultracentrifuged at 150,000 ϫ g for 3 h. The membrane precipitate was suspended in suspension buffer (50 mM triethanolamine, pH 7.4, 250 mM sucrose, and 1 mM DTT) using a Dounce homogenizer. The RM was treated with micrococcal nuclease (Merck) in the presence of 1 mM calcium chloride at 23°C for 10 min and then extracted with 25 mM EDTA on ice. The RM was sedimented through a cushion (0.5 M sucrose in suspension buffer) by ultracentrifugation at 150,000 ϫ g for 1 h and suspended in suspension buffer by pipetting and using a sample tube-shaped pestle. The dose of each RM for the translation reactions was determined to be sufficient for membrane targeting of RSA (Ͼ70%).
Proteins in 293-H-derived RMs were separated by SDS-PAGE and transferred to a nitrocellulose membrane. The membranes were blocked with blocking solution (2% nonfat dry milk in PBS and 0.2% Tween 20) and then incubated with primary (anti-FLAG M2 (Sigma) or anti-Sec61␣ (42)) and appropriate secondary antibodies. The blots were developed with ECL Western blotting detection reagents (GE Healthcare) and imaged using an ImageQuant LAS4000 mini (GE Healthcare).

In vitro transcription and translation
For the synthesis of S-I Cys-87 truncated mRNA (Fig. 2), the plasmid was linearized with NheI located just downstream of TM II. For the synthesis of S-I-II Cys-258 and Cys-266 truncated mRNAs (Fig. 3), plasmids were linearized with BspHI as described previously (34). For the synthesis of truncated and full-length mRNAs of mH segment-containing RSAs, plasmids were linearized with PmaCI (just upstream of the termination codon) and BamHI (downstream of the coding sequence), respectively. Template DNAs were transcribed with T7 RNA polymerase (TakaraBio). mRNAs were translated in a reticulocyte lysate cell-free system as described previously (30) except in the presence of 293-H-derived RMs instead of canine pancreatic RM.

Chemical cross-linking, immunoprecipitation, and image analysis
After the cell-free translation, chemical cross-linking using 1,2-bis(maleimido)ethane (Tokyo Chemical Industry) and sample preparation for immunoprecipitation were performed as described previously (30). The products conjugated to Cysscanned Sec61␣ proteins were then sedimented from lysates with anti-FLAG M2 antibody and protein G-Sepharose (GE Healthcare). Subsequently, the same lysates were subjected to immunoprecipitation with anti-Sec61␣ antiserum to isolate the Cys-scanning analysis of Sec61␣ products conjugated to endogenous Sec61␣ proteins. Sedimented resins were extracted with SDS-PAGE sample buffer.
Proteins in lysates or immunoisolated were separated by SDS-PAGE and visualized on a bioimage analyzer (Typhoon FLA7000, GE Healthcare). Radiolabeled and immunodetected protein bands were quantified using ImageQuant TL software, version 7.0 (GE Healthcare). Cross-linking efficiencies were calculated as follows: products cross-linked to FLAG-Sec61␣ recombinants (a) and total radiolabeled products (the main band (ϳ32-kDa band of SBP-SAI; ϳ37-kDa band of S-I-II; ϳ35-kDa band of RSAs with mH segments) and all upper bands in 12.5% total cross-linked products) (b) were quantified, and the cross-linking percentage (X) was estimated using the following formula: X ϭ a/8 ϫ 100/b. These X values are presented in Figs. 2D, 3C, 4C, 5C, and 7C. Immunodetected FLAG-Sec61␣ proteins expressed in 293-H-derived RMs (c) were also quantified. The expression equivalents in the RMs added to the translation reactions (Y) were estimated as c ϫ d/e (where d is the volume added to translation reaction and e is the volume applied to immunoblotting) and normalized to the largest amount, which was defined as 1. Finally, the cross-linking efficiencies were calculated as X/Y and are presented in Figs. 3D, 4D, 5C, and 6D.