Structural and Functional Profiling of the Lateral Gate of the Sec61 Translocon*

Background: The Sec61 lateral gate mediates transmembrane segment insertion into the ER membrane. Results: Interactions and the effects of mutations of the Sec61 lateral gating helix residues were determined. Conclusion: Key residues that modulate translocation, membrane insertion, and topogenesis of membrane proteins are identified. Significance: This study offers mechanistic and structural insights into Sec61 gating. The evolutionarily conserved Sec61 translocon mediates the translocation and membrane insertion of proteins. For the integration of proteins into the membrane, the Sec61 translocon opens laterally to the lipid bilayer. Previous studies suggest that the lateral opening of the channel is mediated by the helices TM2b and TM7 of a pore-forming subunit of the Sec61 translocon. To map key residues in TM2b and TM7 in yeast Sec61 that modulate lateral gating activity, we performed alanine scanning and in vivo site-directed photocross-linking experiments. Alanine scanning identified two groups of critical residues in the lateral gate, one group that leads to defects in the translocation and membrane insertion of proteins and the other group that causes faster translocation and facilitates membrane insertion. Photocross-linking data show that the former group of residues is located at the interface of the lateral gate. Furthermore, different degrees of defects for the membrane insertion of single- and double-spanning membrane proteins were observed depending on whether the mutations were located in TM2b or TM7. These results demonstrate subtle differences in the molecular mechanism of the signal sequence binding/opening of the lateral gate and membrane insertion of a succeeding transmembrane segment in a polytopic membrane protein.

The x-ray structure of the archaeal-bacterial SecY complex revealed that SecY subunit is divided into two bundles of five transmembrane (TM) helices (TM1-5 and TM6 -10), forming an hourglass-shaped pore with the loop between TM5 and TM6 serving as a hinge (5)(6)(7). The extracytoplasmic funnel of the pore is sealed by a short helical plug domain (TM2a). The narrow constriction in the middle of the channel, called the pore ring, consists of conserved hydrophobic residues. Together, the pore ring and plug seal the channel, preventing the flow of ions and metabolites across the membrane (8). Structural and cross-linking data suggest that TM2b and TM7, which are located opposite to the hinge, presumably form the lateral gate that allows TM segments of nascent chains to exit into the lipid bilayer and serve as the signal sequence binding site during the early stages of translocation (5-7, 9 -12).
To allow protein translocation and/or integration into the lipid bilayer, the translocon has to undergo conformational changes; that is, displacement of the plug and opening of the lateral gate. The SecY structures bound to SecA or a ribosome allow the first glimpse of the semi-open state of the lateral gate (6,7,13). The plug was shown to be important for efficient protein translocation, the orientation of signal anchor proteins, and Sec61 complex stability (14,15). Mutations that allow the translocation of proteins with defective signal sequences in Escherichia coli (prl mutations) (16 -18) are predominantly found in the plug domain, the pore ring, and the lateral gate forming TM2b and TM7 (5,19,20). Recent findings show that conserved residues in the pore ring and the lateral gate define the hydrophobicity threshold for signal sequence insertion and membrane integration (11,21,22). Furthermore, a hydrogenbond cluster between TM2, TM3, and TM7, which connects the two sides of the gate, has been proposed to play an important role in regulating the open and closed conformations of the Sec61 translocon (22,23).
Systematic measurements of potential TM segment integration (24 -27) and molecular dynamics simulations (28 -30) suggest that the process for the initial TM segment insertion from the translocon to the membrane is a thermodynamic event in which Sec61 provides a conduit through which polypeptide chains can partition into or translocate across the membrane (10). However, it is unclear whether partitioning is only passively facilitated by a "breathing" translocon (31), which continuously opens and closes toward the lipid bilayer or whether signal sequence/TM segment binding triggers an active response from the translocon and influences the kinetics of signal sequence binding and TM segment insertion. In cases of membrane insertion of multispanning membrane proteins, cross-linking and fluorescence studies have shown that associations between the TM domain(s) and the translocon and the timing of their release into the lipid bilayer vary depending on sequence context, suggesting that the Sec61 complex may play an active role in TM helix insertion (32)(33)(34)(35). A recent study by Cymer and von Heijne (36) shows that interaction between TM helices could occur co-translationally at the exit site of the Sec translocon.
Despite an abundance of data, the molecular mechanism by which the lateral gate of the Sec61 translocon functions remains elusive. We took two approaches to determine key residues that modulate lateral gating activity: systematic alanine scanning and in vivo site-directed photocross-linking of the TM2b and TM7 of the Saccharomyces cerevisiae Sec61. These approaches identified key residues modulating membrane insertion efficiency of a marginally hydrophobic TM segment and created an interaction profile between TM2b and TM7 of the yeast Sec61.
Amber stop codons and Factor Xa cleavage sites for photocross-linking experiments were also introduced into pDQ1-WT via site-directed mutagenesis. Factor Xa cleavage site was introduced into the loops 1, 2, 5, 6, or 7 at positions after Ser-57, Ser-107, Arg-207, Ile-269, or Ser-351, respectively. Residues in TM2b and TM7 were replaced with an amber codon one at a time in the background of pDQ1 plasmids containing Factor Xa cleavage sites of varying positions. The amber muta-tion and Factor Xa cleavage site carrying pDQ1 plasmids and the plasmid encoding amber suppressor tRNA synthetase and its cognate tRNA were co-transformed into W303-1␣ cells (MAT␣,ade2,can1,his3,leu2,trp1,ura3).
Proteasome Inhibition-Proteasome inhibition using MG132 was carried out as described in Liu et al. (39). RSY1293 (⌬ssh1) cells expressing either Sec61 WT or I91A were transformed with plasmids encoding Lep-H1(5L/14A) model protein. The yeast cells were grown overnight at 30°C in a synthetic medium containing L-proline as the sole nitrogen source with appropriate amino acids. The culture was re-inoculated in the same medium with 0.003% SDS at A 600 of 0.5 and grown for 3 h at 30°C until A 600 reached ϳ1.0. For pulse-chase experiment, 9 A 600 units of cells were harvested for each reaction and incubated in a proline-based medium lacking methionine supplemented with either MG132 (100 M) or the control buffer with dimethyl sulfoxide (DMSO) for 30 min at 30°C. Cells were subjected to pulse-chase experiments after incubation, and MG132 concentration was kept at 100 M during the whole chase time.
Photocross-linking-The W303-1␣ (MAT␣, ade2, can1, his3, leu2, trp1, ura3) was transformed with two plasmids, one carrying Sec61 with an amber mutation and the other carrying the amber suppressor tRNA-synthetase and its cognate tRNA (40). The yeast cells were grown at 30°C on ϪLeu ϪTrp medium to an optical density at 600 nm (A 600 ) of 1.0. Exponentially growing cultures were then diluted to A 600 ϭ 0.2, supplemented with 1 mM benzo-L-phenylalanine (Bpa) in 1 mM NaOH, and grown to an A 600 of 0.5. Cells were harvested by centrifugation at 3200 rpm. Cell pellets were resuspended in 1 ml of ϪLeu ϪTrp medium, and half of the cell pellets were mixed with an additional 2.5 ml of medium, transferred to a Petri dish (60 ϫ 15 mm) on ice, and then irradiated with ultraviolet (UV) radiation at 365 nm for 30 min. The other half of the cell pellets was kept on ice.
After cell lysis with glass beads, immunoprecipitation (IP) was performed by heating the cross-linked samples for 10 min at 55°C in the presence of 1% SDS followed by centrifugation at 16,000 ϫ g for 10 min. Four volumes of immunoprecipitation buffer (10 mM Tris-HCl, pH 7.6, 140 mM NaCl, 1 mM EDTA, 1% Triton X-100) were then added to the aliquots of the resulting supernatant. The mixture was incubated with gentle rotation in the presence of protein G-agarose, which was prewashed twice with IP buffer, and anti-HA antiserum (Covance) for 3 h at room temperature. The IP product was washed twice with IP buffer and once with Buffer C (10 mM Tris-HCl, pH 7.5, 50 mM sodium chloride).
Factor Xa Digestion-The IP products were treated with 1 l of Factor Xa (New England Biolabs, Ipswich, MA) for 2 h at 23°C in 100 l of Xa buffer (20 mM Tris-HCl, pH 8.0, 100 mM sodium chloride, 2 mM calcium chloride), which was adjusted to 0.017% SDS. After Factor Xa treatment, the same washing steps used for immunoprecipitation were performed followed by incubation with SDS-PAGE sample buffer at 55°C for 10 min before running on SDS gel.
Pulse-labeling, Pulse-chase, and IP-Cell labeling, pulsechase, lysis, and IP were performed as previously described (41). Immunoprecipitated proteins were subjected to SDS-PAGE and quantified using Fuji FLA-3000 phosphorimaging and the Image Reader V1.8J/Image Gauge V 3.45 software.
Western Blot Analysis-Yeast protein analysis by Western blotting was performed as previously described (41). Anti-HA antiserum (Covance) was used for detection of the model substrates.

RESULTS
Alanine Scanning of TM2b and TM7 of Sec61-To probe which residues in TM2b and TM7 of Sec61 are critical for the lateral gating function, systematic mutagenesis of these helices was performed. Residues 82-99 in TM2b and 290 -308 in TM7 were substituted with an Ala residue one at a time. Functional complementation by Sec61 variants was assessed by plasmid shuffling in a sec61 deletion strain carrying the WT SEC61 on a Ura-born plasmid. With the exception of F92A, all the mutants were viable, although I293A and L295A showed slower growth, indicating that these Sec61 mutants could functionally replace the WT protein (Fig. 1A). Sec61 has been shown rather insensitive to mutations (15,17,21,22,37). In case these Sec61 mutants disrupt the complex formation, thereby impairing the stability of individual subunits, we assessed the relative expression levels of the Sec61 mutants and Sec62 by Western blotting analysis (Fig. 1B). The protein expression levels of the Sec61 variants and Sec62 are comparable with the WT. This result indirectly indicates that the stability of the Sec61 mutants or Sec62 is not compromised.

Identifying Sec61 Mutants That Increase or Decrease the Translocation and Membrane Insertion Efficiency of a Singlespanning Membrane Protein-Each Sec61
Ala mutant strain was transformed with a set of model protein constructs to assess how each Sec61 mutant handles the targeting and membrane insertion of model proteins of varying sequence context. First, we determined the efficiency of targeting and membrane insertion for a model single-spanning membrane protein derived from E. coli leader peptidase (Lep) (27). In this protein the original two N-terminal TM segments were replaced with a hydrophobic segment (H1-segment) consisting of 5 Leu and 14 Ala residues ( Fig. 2A). We selected H1-segment with threshold hydrophobicity (5L/14A) because this segment enables sensitive detection of the insertion efficiencies of Sec61 gate mutants. The H1-segment can act as a signal sequence that targets the protein to the ER and as a TM domain (27,41). Interestingly, this model protein, Lep-H1, is integrated into the membrane with dual membrane topology (N out -C in (left) and N in -C out (right) membrane orientations; Fig. 2A). Lep-H1 contains one glycosylation site at the N terminus and two at the C terminus; thus, both membrane targeting and integration with either N in -C out or N out -C in topology can be monitored by assessing the glycosylation status: a double-glycosylated form (2G) indicates membrane insertion in the N in -C out orientation, a single-glycosylated form (1G) indicates membrane insertion in the N out -C in orientation, and an unglycosylated form (0G) indicates a non-translocated/non-targeted protein (Fig. 2B). For rare cases, we observed a triple-glycosylated form (3G) that represents a fully translocated protein.
A plasmid encoding Lep-H1(5L/14A) was transformed into the Sec61 mutant strains, and each transformant was subjected to 5 min of [S 35 ]Met radiolabeling and subsequent IP and analyzed by autoradiography. The Lep-H1(5L/14A) segment is moderately hydrophobic (⌬G of membrane insertion ϭ Ϫ1 kcal/mol) and has ϳ48% targeting efficiency (glycosylated products/total products) in WT Sec61 (Fig. 2B). Compared with WT, the targeting efficiency of the model protein was increased by ϳ10% for the V82A, I86A, Q96A, I293A, F304A, and S307A Sec61 mutants (Fig. 2C). Interestingly, in these mutants the model protein that was integrated into the membrane with an N in -C out orientation was enhanced (45-54% in the mutants compared with 33% in WT) (Fig. 2D).
We identified the second group of Sec61 mutants that decreased the insertion of the Lep-H1 model protein. In Sec61 mutants S83A, M90A, I91A, Q93A, F94A, L95A, N290A, P292A, L295A, and N302A, ϳ20% less of the Lep-H1 model protein was integrated into the membrane compared with WT (Fig. 2C). These mutations also decreased the relative amount of Lep-H1 with an N in -C out membrane topology (Fig. 2D).

Dissecting Mechanisms for the Increased or Decreased Efficiency of Translocation and Membrane Insertion in Sec61
Mutants-To determine whether the increase or decrease in Lep-H1(5L/14A) translocation in Sec61 mutants during 5 min of pulse-labeling would eventually reach the WT level, we performed pulse-chase experiments. WT, V82A, I86A, I91A, and N302A Sec61 mutant strains containing Lep-H1(5L/14A) were subjected to pulse-chase and IP and analyzed by autoradiography (Fig. 3A). The V82A and I86A Sec61 mutants had faster FIGURE 1. Alanine scanning of TM2b and TM7 of Sec61 had little or no effect on translocon functionality and stability. A, complementation assay in yeast cells expressing WT or mutant Sec61 by plasmid shuffling. ⌬sec61 cells carrying the Ura-born WT SEC61 were transformed with WT or sec61 variants in Leu-born plasmids. Cells were grown on ϪLeu plates containing 5-fluoroorotic acid and incubated for 3 days at 30°C. B, relative expression levels of Sec61 with single Ala substitutions in TM2b or TM7, and Sec62 were assessed by Western blotting. Tom70, a mitochondrial outer membrane protein served as a loading control.
translocation in a 5-min chase, but translocation was not increased further and eventually reached the WT level during a 15-min chase (Fig. 3A, left panel). In contrast, I91A and N302A exhibited slower translocation, and the translocation efficiency did not reach the WT level in a 15-min chase (Fig. 3A, left panel).
In the V82A and I86A Sec61 mutants, the relative amount of Lep-H1(5L/14A) protein inserted in the N in -C out orientation was ϳ2-fold more than that for the WT during a 5-min chase (Fig. 3A, right panel). Although the protein targeting in the WT eventually reached that of the V82A and I86A Sec61 mutants during a 15-min chase, the ratio of the Lep-H1(5L/14A) protein in the N in -C out orientation remained ϳ2-fold more for the V82A and I86A Sec61 mutants compared with the WT during a 15-min chase. In the case of the I91A and N302A Sec61 mutants, the relative amount of Lep-H1(5L/14A) protein inserted in the N in -C out orientation was decreased and changed little during a 15-min chase (Fig. 3A, right panel).
To determine whether unglycosylated proteins are degraded by the proteasome in the cytosol or slowly targeted, we carried out the experiment with the proteasome inhibitor, MG132 (Fig.  3B). In the presence of MG132, more unglycosylated protein was detected compared with the sample without MG132, indicating that some of unglycosylated proteins are degraded by the proteasome in the cytosol (ϳ10% for both Sec61 WT and I91A mutant), but the glycosylated products were still increased over the chase time, suggesting that some unglycosylated proteins are slowly translocated/membrane-inserted.
These data suggest that the V82A and I86A Sec61 mutants are pre-opened or have faster channel opening, whereas I91A and N302A Sec61 mutants are more prone to remain in a closed conformation or slowly open. These results further imply that the orientation of a TM domain is influenced by the status of the Sec61 channel opening when the TM domain enters into the pore.

Ssh1 Has Little Effects on the Membrane Insertion of Model Proteins in Sec61 Mutant
Strains-Ssh1, a homolog of Sec61 in yeast, mediates protein translocation, but cells are viable upon its deletion (42,43). To determine whether Ssh1 influences the membrane insertion efficiency and orientation of a model protein in the Sec61 mutant background, we tested the Lep-H1(5L/ 14A) protein in the absence of Ssh1 using an ssh1 deletion strain. Although targeting efficiency was reduced by ϳ3-17% in the absence of Ssh1, the pattern of effects due to the Sec61 mutations remained the same (Fig. 4A). Furthermore, membrane insertion in the N in -C out orientation was unchanged regardless of the presence or the absence of Ssh1 (Fig. 4B). These results suggest that Ssh1 has little influence on the observed effects of the targeting and membrane insertion of model proteins in the Sec61 mutant strains. The targeting efficiencies were calculated as the amount of glycosylated products over the total products. The average plus S.E. for at least three independent measurements is shown. D, the relative amount (%) of Lep-H1(5L/14A) with an N in -C out membrane topology was calculated as 2G/(1G ϩ 2G) ϫ 100% using the same data set in B and normalized against the WT level.

Mutations in TM2b and TM7 Exhibit Different Severity of Defects in the Membrane Insertion of Single-Versus Double-
spanning Membrane Proteins-The TM segment in singlespanning membrane proteins initiates translocation, opens the Sec61 lateral gate, and positions itself in the N in -C out or N out -C in orientation before membrane insertion. However, the mechanism of the membrane insertion of the downstream TM domain in multispanning membrane proteins may differ because the initial opening of the channel is not required. To test this, we used the Lep-H2 model protein, the membrane insertion of which was previously analyzed in detail both in mammalian and yeast systems (27). This protein contains two hydrophobic segments with two N-linked glycosylation sites, one at the N terminus and the other at the downstream of the second TM segment (Fig. 5A, left panel). A triple hemagglutinin (HA) tag was fused at the C terminus of the protein to facilitate immunoprecipitation against an anti-HA antibody. The first TM segment is the original first TM segment of E. coli Lep and is shown to efficiently target the protein to the ER (27). The second segment was replaced by an engineered segment (H2segment), consisting of 2L/17A (H2-2L, ⌬G ϭ 0.7 kcal/mol) or 3L/16A (H2-3L, ⌬G ϭ 0.0 kcal/mol) (Fig. 5A, left panel). If the

. Ssh1 does not significantly influence the effects of the Sec61 mutations on protein translocation and membrane integration.
A, to assess whether Ssh1 influences the translocation of Lep-H1(5L/14A), the model protein was expressed in ⌬sec61 ⌬ssh1 cells carrying WT or mutant Sec61. Cells were radiolabeled and treated as described in Fig. 2. Glycosylated products were quantified and plotted as targeting efficiency (%). B, the relative amount of Lep-H1(5L/14A) with an N in -C out topology was calculated as 2G/(1G ϩ 2G) ϫ 100%. The average of two independent experiments is shown.
H2-segment is not membrane-inserted, a single-glycosylated product results, whereas if it is membrane-inserted, a doubleglycosylated form appears (Fig. 5A, left panel). For H2-2L, the major product was a single-glycosylated protein, and the minor product was a double-glycosylated protein, both of which became an unglycosylated protein upon endoglycosidase H treatment (Fig. 5A, right panel). Albeit little, a very faint cleaved product that migrated faster than the unglycosylated Lep-H2 was detected (Fig. 5A, right panel). Because the size of this band did not change upon endoglycosidase H treatment, it was unglycosylated. This cleaved unglycosylated protein only appeared when the H2-segment was not membrane-inserted. For H2-3L, the major product was double-glycosylated, and the minor product was single-glycosylated as judged by endoglycosidase H digestion (Fig. 5A, right panel). When the doubleglycosylated product appeared, smaller size band was detected, and upon endoglycosidase H treatment the size was shifted down, indicating that it was glycosylated. It only appeared when the H2-segment was membrane-inserted, and it has been confirmed in Lundin et al. (27) that this cleavage occurs by the signal peptidase activity. We selected H2-segments with 2L/17A and 3L/16A, the H2-segment membrane insertion effi-ciency of which are ϳ20 and ϳ70%, respectively. The threshold hydrophobicity of the Lep-H1 (single-spanning) and Lep-H2 (double-spanning) model proteins was shown to be different (27), because the H1-segment must be inserted into the membrane alone, whereas H2-segment membrane insertion can be aided by the preceding TM segment (44); thus Lep-H1 requires higher hydrophobicity than the H2-segment in Lep-H2. For WT and all the mutant strains, the model protein was efficiently targeted, as little unglycosylated products were detected (Fig.  5A). For H2-2L and H2-3L, compared with the WT level, we observed two groups of mutations that either increased (V82A, I86A, S88A, and Q96A) or decreased (S83A, I91A, L95A, N290A, P292A, L295A, S301A, N302A, and Q308A) the membrane integration of the H2-segment (Fig. 5B). These gating mutant groups superimpose with those found with the Lep-H1 model protein with the exception of I293A, S301A, F304A, and Q308A. In the I293A and F304A Sec61 mutants, targeting and membrane insertion of the Lep-H1 protein were defective, but membrane insertion of the Lep-H2 protein was only mildly impaired (Figs. 2, C and D, and 5B). In the S301A and Q308A Sec61 mutants, membrane insertion of the H2-3L model protein was reduced by Ͼ30% compared with the WT (Fig. 5B), whereas the targeting defect with the Lep-H1 model protein was moderate (Fig. 2C). We noticed that the targeting and membrane insertion of Lep-H1 was more severe in the TM2b mutants (S83A, M90A, I91A, Q93A, F94A, and L95A; 16 -36% or an average of 23% reduced insertion relative to the WT) compared with the TM7 mutants (N290A, P292A, L295A, S301A, N302A, and Q308A; 7-22% or an average of 15% reduced insertion relative to the WT) (Fig. 2C). However, the membrane insertion of the H2-segment in H2-3L construct was more severely affected in the TM7 mutants (N290A, P292A, L295A, S301A, N302A, and Q308A; 26 -46% or an average of 35% reduced insertion relative to the WT) than the TM2b mutants (S83A, M90A, I91A, Q93A, F94A, and L95A; 4 -48% or an average of 16% reduced insertion relative to the WT) (Fig. 5B). These results may indicate that the initial binding of a signal anchor sequence may be mediated more by TM2b, whereas the membrane insertion of a succeeding TM segment is more influenced by TM7.
The Effects of Individual Mutations Are Additive-We tested whether a combination of individual mutations would have an additive effect on TM segment insertion. Sec61 double mutants with two Ala mutations that increased (V82A/I86A, V82A/ Q96A, I86A/Q96A) or decreased (S83A/I91A) membrane insertion and mutants with both effects (V82A/I91A, S83A/ Q96A, L295A/F304A) were prepared. The Lep-H2 model proteins (Fig. 5A) were expressed in these double mutants, and the membrane insertion efficiency of the H2-segment was assessed (Fig. 5C). The V82A/I86A, V82A/Q96A, and I86A/Q96A double mutants further enhanced the membrane insertion efficiency of the H2-segment compared with the respective single mutants, whereas S83A/I91A significantly decreased membrane insertion efficiency, indicating that the effect of Ala mutations in gating helices is cumulative. The combination of two Ala mutations that increased and decreased membrane insertion (V82A/I91A, S83A/Q96A, L295A/F304A) counterbalanced the opposing effects and showed membrane insertion levels between those of the individual mutants (Fig. 5C). In the I83A/I91A double mutant, the membrane insertion of H2-2L and H2-3L was significantly reduced, and the membrane insertion of the hydrophobic H2-5L (⌬G ϭ Ϫ1.0 kcal/mol), which was efficiently inserted in the WT (ϳ96%), was also reduced to 58%, indicating that the lateral gate is either stabilized with the closed conformation or more difficult to open (Fig. 5D).
Bpa-induced Photocross-linking-To investigate which residues in TM2b and TM7 of Sec61 face one another to mediate the opening and closing of the gate, an in vivo photocross-linking approach was performed (40,45). Each residue in TM2b (82-96) and TM7 (290 -308) was systematically substituted with the amber codon (TAG), and tandem repeats of the Factor Xa protease site (5Ј-IEGRIEGR-3Ј) were introduced into the first (loop 1), second (loop 2), fifth (loop 5), sixth (loop 6), or seventh loop (loop 7) by site-directed mutagenesis (46,47). We confirmed that the introduction of Factor Xa cleavage sites into the Sec61 loops did not disrupt the function of Sec61 by plasmid shuffling (Fig. 6A).
Our experimental procedures consisted of two steps: UV cross-linking and Factor Xa-mediated cleavage of Sec61. Bpa is a photoreactive unnatural amino acid and generates a cross-link to a molecule in close proximity under UV irradiation. Bpa was biosynthetically incorporated into the amber suppressor codon in Sec61. Whole cell lysates were subjected to UV irradiation at 365 nm for 30 min followed by IP and Factor Xa treatment and analyzed by SDS-PAGE and Western blotting.
The rationale behind this approach assumes that if TM2b and TM7 are cross-linked, the full-length product of Sec61 with a Factor Xa cleavage site in loop 6 (Sec61-Xa(loop 6)) will still be present after Factor Xa treatment (Fig. 6, B and C, upper panels). In contrast, without cross-linking, Sec61-Xa(loop 6) will be cleaved into two peptides (TM1-5 and TM6 -10) (Fig. 6, B and  C, lower panels). Because the triple HA tag for IP and Western blotting is located at the C terminus of Sec61, only the cleaved C-terminal portion of Sec61 will be detected.
Site-directed Photocross-linking Identified residues in TM2b and TM7 of Sec61 That Are in Close Proximity to One Another-All Bpa-incorporated Sec61 variants were subjected to UV irradiation and subsequent IP and Factor Xa treatment. By comparing the relative amounts of full-length Sec61 over the total product (full-length plus cleaved Sec61) in UVϩ/Factor Xaϩ versus UVϪ/Factor Xaϩ, the efficiency of Sec61 intramolecular cross-linking was assessed. The efficiency of Factor Xa cleavage usually reached ϳ95%, and UV-activated cross-linking efficiency was maximal at 10 -20%.
Sec61 variants with Bpa incorporated at residues Ile-86, Met-90, and Gln-93, with a Factor Xa site in loop 5, had ϳ10 -20% cross-linking efficiency, indicating that these residues are cross-linked to a region between TM6 and TM10 of Sec61 (Fig.  7A). In the case of Ser-89, ϳ20% cross-linking efficiency was observed when the cleavage site was in loop 2, but no crosslinked products were detected when the cleavage site was in loop 5, indicating that Ser-89 is cross-linked to a region between TM3 and TM5.
To further narrow down the region where the Ile-86, Met-90, and Gln-93 residues were cross-linked, Factor Xa sites were engineered in loop 6. For Ile-86, the full-length product was dropped to Ͻ10%, whereas Met-90 and Gln-93 maintained ϳ10% of full-length products when loop 6 was cleaved (Fig. 7A). This result indicates that Met-90 and Gln-93 are cross-linked to a region between TM7 and TM10.
For Sec61 variants with Bpa incorporated in TM7 and a Factor Xa site located in loop 1 or loop 2, Met-294, Leu-295, and Ser-301 showed ϳ20% full-length products when the cleavage site was in loop 2, but the amount of full-length product was decreased for Leu-295 and Ser-301 when the cleavage site was in loop 1 (Fig. 7A). These data indicate that although Met-294 faces TM1 and TM2, Leu-295 and Ser-301 preferentially interact with TM2. Asn-302 showed ϳ10 -15% cross-linking efficiency when the cleavage site was in loop 1, and Ile-303 showed ϳ15% cross-linking efficiency when the cleavage site was in loop 2, indicating that Asn-302 may weakly interact with TM1 and Ile-303 may interact with the TM2. Although Gln-93 showed ϳ10 -15% cross-linking efficiency when the cleavage site was in loop 5 or loop 6, its cross-linking was abolished when the cleavage site was in loop 2. This contradictory result is puzzling but may indicate that local structural changes can affect cross-linking efficiency.
Overall, the cross-linking efficiency was low, which may reflect the flexible nature of gating helices because cross-linking would be more efficient if nearby residues were in constant contact rather than dynamically interacting. Thus, failure to cross-link does not mean that a residue is not in close proximity with other Sec61 TM segments. For the same reason, crosslinked residues may indicate that they are in close proximity and have a firm contact. The residues in TM2b that were crosslinked to TMs 7-10 or the residues in TM7 that were crosslinked to TM2b are the residues that showed a decrease in membrane insertion for model proteins when mutated to Ala (Met-90, Gln-93, Leu-295, and Ser-301). These results suggest that when the critical residues lining the interface of the lateral gate are mutated, two helices may collapse toward one another and stabilize the closed lateral gate; thus, these residues show relatively efficient cross-linking.
Our data are summarized in Fig. 7B. The corresponding homologous residues of cross-linked residues are indicated in blue. Ile-83 and Leu-261 in the Methanocaldococcus jannaschii SecY (Met-90 and Leu-295 in yeast, respectively) are located in closest proximity, facing each other. In the case of Gln-86 in SecY (Gln-93 in yeast), it is located near TM8 (Fig. 7B, left and  middle panels). Moreover, the corresponding residues in Thermotoga maritima, Ile-90 and Ile-274 (Met-90 and Leu-295 in yeast), also face each other, even in the SecA-bound state (Fig.  7B, right panel). Finally, when we mapped Ser-281 in T. maritima (Ser-301 in yeast), it was located near the plug domain (colored in orange). These results indicate that the gating helices have high structural homology to the archaeal-bacterial TM2b and TM7 in terms of helix orientation and proximity of individual residues and thus suggest a high structural similarity between the eukaryotic Sec61 translocon and its archaeal-bacterial homolog SecY.

DISCUSSION
The structure of the archaeal and bacterial SecY translocon and earlier cross-linking studies indicated that TM2b and TM7 form the site where signal sequences bind and TM domains partition from the translocon channel into the lipid bilayer and has been referred to as the "lateral gate" (5-7) (9, 10). Guided by these earlier studies, we attempted to determine an interaction profile for the residues in the lateral gate of yeast Sec61 using an in vivo site-directed photocross-linking approach, and analyzed the effects of mutations of those residues using membrane insertion assays with a set of model proteins. Three main findings are summarized in the following.
First, data from cross-linking experiments show that Met-90 and Gln-93 in TM2b face TM7-10 and Leu-295 and Ser-301 in TM7 face TM2b. Corresponding residues in SecY structure show that they are in close proximity. When these residues were mutated to Ala one at a time, the membrane insertion efficiency of both single-and double-spanning membrane proteins was reduced. These results suggest that the mutation of the residues in the interface of the lateral gate may cause the collapse of the lateral gate and narrow the distance between two helices, hence making the gate harder to open. Thus, these residues may be the key residues guarding the lateral gate. It has been shown that disulfide bond formation between Ile-86/Phe-279 in E. coli SecY (Ile-86/Leu-295 in yeast Sec61) causes a translocation defect, suggesting that these residues are in close proximity and mediate the opening of the lateral gate (48).
Second, alanine scanning of Sec61 TM2b and TM7 identified mutants that increase the efficiencies of targeting and membrane insertion of model proteins. These residues are scattered across helices TM2b and TM7, and cross-linking data show that they face away from the lateral gate interface or are not cross-linked. We observed that these mutants allow membrane insertion to occur more quickly and facilitate membrane insertion of a single-spanning membrane protein in an N in -C out topology. These mutants exhibit similar phenotypes as prl mutants in E. coli (16 -18), which allow the translocation of substrates carrying defective signal sequences. When Bondar et al. (23) examined the effect of prl mutations by molecular simulation, the interaction between the plug and pore region of SecY was loosened; thus the closed conformation was destabilized. They also suggested that when the closed conformation is destabilized, the channel pore allows more water molecules to enter (23). This polar environment of the channel pore may promote a single-spanning membrane protein to enter as a hairpin or insert into the ER membrane heads-in and then flip. Alternatively, the premature opening of the lateral gate may allow faster or increased contact with Sec62, which is located at the lateral exit site of Sec61. Recently, we have shown that Sec62 facilitates the membrane insertion of a TM segment in an N in -C out membrane topology (41). Therefore, faster or increased contact with Sec62 by the premature opening of the Sec61 lateral gate may enhance the membrane insertion of a TM domain in the N in -C out orientation.
Third, the degree of defects in targeting and membrane insertion differs depending on the type of a model protein between TM2b and TM7 Ala mutations. In the TM2b Sec61 variants M90A, Q93A, and F94A, the membrane insertion efficiency of a single-spanning membrane protein was reduced by ϳ20% compared with the WT. Interestingly, the membrane insertion of the second TM segment of a double-spanning membrane protein was only reduced by ϳ5%. In comparison, the TM7 Sec61 variants S301A, N302A, and Q308A showed a relatively greater defect in the membrane insertion of the downstream TM segment in a double-spanning membrane protein (H2-3L, reduced by ϳ35%) than a single-spanning membrane protein (Lep-H1(5L/14A), reduced by ϳ12%). These results may illustrate subtle differences in molecular mechanisms of signal sequence binding/initial opening of the lateral gate and membrane insertion of the downstream TM segment in multispanning membrane proteins that does not require initial opening of the channel.
An earlier photocross-linking study suggested that the hydrophobic core of a signal sequence (ϳ12 residues) forms a FIGURE 7. Cross-linking efficiency of each residue in TM2b and TM7. A, the cross-linking efficiency of each residue was calculated by comparing the amount of full-length UVϩ/Factor Xaϩ versus UVϪ/Factor Xaϩ proteins. Factor Xa cleavage sites were present in the indicated Sec61 loops. The average of at least three independent measurements plus S.D. is shown. B, corresponding residues of the cross-linked residues in yeast Sec61 are indicated in the SecY structure (blue) (PDB code 1RH5) (5), PDB code 3DIN) (13). The corresponding homologous residues in yeast are shown in parentheses. TM2b and TM7 are colored in yellow and green, respectively. TM2a (plug domain) is colored in orange (right). M. ja, M. jannaschii; T. ma, T. maritima. helical structure and binds between helices TM2 and TM7 as well as Sec62/71 (9). Twelve residues form an alpha helix with an estimated length of ϳ18 Å, which is roughly long enough to traverse the membrane half way. Based on these earlier studies and our present study, we speculate that the residues on the cytoplasmic side of TM2b may be critical for the recognition and binding of a signal (anchor) sequence and/or initiating the opening of the lateral gate, whereas TM7 has more influence on the membrane insertion of the succeeding TM segment in multispanning membrane proteins. Alternatively, our results may indicate that the first TM segment initiates opening of the lateral gate and remains near TM2b until the succeeding TM segment is translated and exits together through the lateral gate. If so, the succeeding TM segment would have less of an opportunity to interact with TM2b than TM7.