The Lateral Gate of SecYEG Opens during Protein Translocation*

The SecYEG translocon of Escherichia coli mediates the translocation of preproteins across the cytoplasmic membrane. Here, we have examined the role of the proposed lateral gate of the translocon in translocation. A dual cysteine cross-linking approach allowed the introduction of cross-linker arms of various lengths between adjoining aminoacyl positions of transmembrane segments 2b and 7 of the lateral gate. Oxidation and short spacer linkers that fix the gate in the closed state abolished preprotein translocation, whereas long spacer linkers support translocation. The cross-linking data further suggests that SecYEG lateral gate opening and activation of the SecA ATPase are coupled processes. It is concluded that lateral gate opening is a critical step during SecA-dependent protein translocation.

The SecYEG translocon of Escherichia coli mediates the translocation of preproteins across the cytoplasmic membrane. Here, we have examined the role of the proposed lateral gate of the translocon in translocation. A dual cysteine cross-linking approach allowed the introduction of cross-linker arms of various lengths between adjoining aminoacyl positions of transmembrane segments 2b and 7 of the lateral gate. Oxidation and short spacer linkers that fix the gate in the closed state abolished preprotein translocation, whereas long spacer linkers support translocation. The cross-linking data further suggests that SecYEG lateral gate opening and activation of the SecA ATPase are coupled processes. It is concluded that lateral gate opening is a critical step during SecA-dependent protein translocation.
Translocation of preproteins across the cytoplasmic membrane in Escherichia coli is mediated by the Sec translocase (for a recent review see Ref. 1). Preproteins targeted for secretion contain a signal sequence that is removed upon translocation. Their synthesis and translocation are uncoupled events (2), and directly after synthesis at the ribosomes, preproteins are targeted post-translationally to the Sec translocase by the molecular chaperone SecB (3). SecB transfers the preprotein to the motor protein SecA bound at the SecYEG pore complex (4,5). SecA utilizes cycles of ATP binding and hydrolysis to bind and release the translocating protein resulting in its stepwise translocation across the membrane (6 -8). In addition, the proton motive force facilitates translocation when the preprotein is released by SecA (6,9). Various models for SecA-mediated translocation have been proposed wherein SecA functions as a power-stroke device (10) or as a directed molecular ratchet wherein SecA controls the opening and closure of the pore (11). Another view is that SecA thrusts deep into the SecYEG channel during translocation (12,13). In a recent study on the cocrystallization of the Thermotoga maritima SecA with SecYEG, it was suggested that a two-helix finger from the helical scaffold domain of SecA inserts into the cytoplasmic domain of SecY, utilizing cycles of ATP hydrolysis to push the substrate into the SecY pore (14).
The translocation pore consists of three integral membrane proteins SecY, SecE, and SecG as subunits (15), and this organization is universally conserved in all three kingdoms of life (16). The crystal structure of Methanocaldococcus jannaschii (17) demonstrates that the largest subunit, SecY, consists of an N-and C-terminal domain that comprise TMs 3 1-5 and 6 -10, respectively. These two domains are organized as a clamshelllike structure that encompass an hourglass-shaped central pore. This putative pore is closed at the periplasmic face of the membrane by a short transmembrane helix, TM2a, which has been proposed to function as a plug domain. The clamshell-like structure of SecY is embraced by SecE that in its minimal form consists of a surface-localized amphiphatic helix and a highly tilted transmembrane segment that localizes to the "back" of the SecY protein. It has been proposed that the "front" of SecY creates a lateral opening of the central pore to the membrane between TM2b and TM7 and that this gate is used to release signal sequences and transmembrane segments from the translocase (17). Cryoelectron microscopy of the E. coli SecYEG complex bound to a translating ribosome (18) suggests that the ribosome-bound SecYEG is organized as a dimer with a frontto-front organization (18). It was proposed that individual pores of the dimer have distinct functions in protein translocation, i.e. vectorial protein translocation and lateral release of TMs into the membrane (19). Freeze-fracture rotational shadowing electron microscopy has provided evidence for oligomeric forms of SecYEG, and suggest that SecA recruits SecYEG monomers to form a dimeric complex (20). Within this dimeric SecYEG complex, only a single pore seems sufficient for the translocation of preproteins (10).
The mechanism by which the translocase coordinates protein translocation is only poorly understood. SecA has been proposed to insert the signal sequence into the SecYEG pore where it may latch between TM2b and TM7 of the SecY lateral gate. This would result in a widening of the central pore constriction and a subsequent displacement of the periplasmic plug domain. Next, adjoining polypeptide segments of the preprotein may enter the opened aqueous pore, but it is not clear if under those conditions the lateral gate remains open or is closed. Despite this vast amount of experimental data available on the function of the SecYEG complex, the exact role of the putative lateral gate remains unknown. Thus far, the only study on its dynamics and role during translocation concerns a molecular dynamics simulation (21) that does not take SecA or ribosome binding into account. Here we have investigated the function and the dynamics of the proposed lateral gate located between TM2b and TM7 in protein translocation. The data demonstrates that the lateral gate needs to open to allow for SecA-mediated preprotein translocation.
Bacterial Strains and Plasmids-All strains and plasmids used are listed in Table 1. All DNA manipulations were performed using E. coli DH5␣ to maintain plasmids and constructs. Cysteine mutations were introduced into a Cys-less SecY using the Stratagene QuikChange site-directed mutagenesis kit using plasmid pEK1 as template. Mutations were confirmed by sequencing. Plasmids expressing the mutated SecYEG complex were created by exchanging the NcoI-ClaI secY fragment in pEK20 by the cysteine containing the NcoI-ClaI secY fragment of the pEK1 derivative. Plasmid pET651 4 harboring the F279C mutation, was used as template to generate the F279C/M83C and F279C/I86C double cysteine mutants. E. coli strain SF100 or NN100 (SF100, unc Ϫ ) was used for overproduction of the various SecYEG complexes.
OmpT was expressed from plasmid pND9 in strain SF100 and expressed under its own temperature-sensitive promoter (24). After overnight incubation at 37°C, outer membranes containing high levels of OmpT were isolated by differential centrifugation. Briefly, cells were harvested at 7,500 rpm for 15 min in a JLA 10.500 (Beckman) rotor, resuspended in a buffer containing 50 mM Tris-HCl, pH 8, and 1 mM EDTA, whereupon the suspension was passed twice through a cell disruptor at 8,000 p.s.i. Cell debris was removed by centrifuging at 4,000 ϫ g for 15 min in a SS-34 (Beckman) rotor. The supernatant was transferred to a Ti45 ultracentrifuge rotor (Beckman) and centrifuged for 1 h at 40,000 ϫ g and 4°C. The outer membrane pellet was resuspended in a buffer containing 50 mM Tris-HCl, pH 8, and 250 mM sucrose and stored in small aliquots at Ϫ80°C.
Chemical Cross-linking and OmpT Assay-IMVs containing overproduced levels of the SecYEG mutants were diluted to 1 mg/ml, whereupon the cross-linkers NaTT (1 mM), bBBr (1 mM), BMOE (300 M), or BMH (600 M) were added, respectively. The reducing agents DTT and TCEP were used at a final concentration of 30 and 5 mM, respectively. For optimal crosslinking, reactions were performed in a maximum volume of 30 l and afterward pooled for analysis. After 30 min at 37°C, IMVs were sedimented through a sucrose cushion in 50 mM Tris-HCl, pH 8, and 0.8 M sucrose for 20 min at 80,000 rpm in a TLA 120.1 rotor at 4°C. The pellet was resuspended in 50 mM Tris-HCl, pH 8, 20% glycerol and used for further reactions.
An OmpT protease assay was used to assess the cross-linking between cysteine residues in TM2b and TM7 of SecY. Outer membranes containing overexpressed levels of OmpT were diluted to a final concentration of 1 mg/ml in 50 mM Tris-HCl, pH 7, 0.1% Triton X-100. The OmpT solution (7.5 l) was mixed with the cross-linked IMVs (12 l) and incubated for 30 min at 37°C. Where indicated, the oxidized cysteine residues were reduced by the addition of an excess DTT (30 mM final concentration), and the samples were separated by 12% SDS-PAGE. SecY was visualized by staining with Coomassie Brilliant Blue R-250 or SYPRO Ruby (Invitrogen) where indicated. Visualization with SYPRO Ruby staining was done with a Roche Lumi-Imager F1 using a cut-off filter of 600 nm. Western blotting was performed using antibodies directed against the N-terminal hexa-histidine tag in SecY (Amersham Biosciences).
Cross-linking in the Presence of the ProOmpA-DHFR Translocation Intermediate-Folding of the DHFR domain and translocation of the proOmpA-DHFR into IMVs was performed as described (25). After 30 min of translocation at 37°C, the different cross-linkers were added to the reaction mixture and incubation was continued for another 30 min at 37°C. To analyze cross-linking of the cysteine residues, IMVs were treated with OmpT as described above.
Translocation ATPase Assay-The SecA ATPase activity during translocation was determined by measuring the amount of released free phosphate using the malachite green assay (26). Typically, proOmpA (33 g/ml) was added to translocation buffer containing SecA (20 g/ml), SecB (400 g/ml), 1 mM ATP and IMVs (12.5 g/ml). P i release was measured after 30 min. Measurements were done in triplicate and corrected for background ATPase activity in the absence of proOmpA.
Other Techniques-In vitro translocation of proOmpA was performed as described (23). Typically, fluorescently labeled proOmpA was diluted into translocation buffer containing SecA (20 g/ml), SecB (400 g/ml), ATP (1 mM), and IMVs containing SecYEG derivatives. After 8 min at 37°C, the translocation reaction was terminated on ice by treatment with proteinase K, and analyzed by SDS-PAGE and in-gel UV fluorescence using a Roche Lumi-Imager F1 station using bond pass filters of 520 (FL) and 600 nm (Texas Red), respectively. Protein concentrations were determined with the Bio-Rad RC DC protein assay kit using bovine serum albumin as a standard.

Introduction of Cysteines in the Putative Lateral Gate of
SecYEG-To date, no direct evidence exist for the functioning of the proposed lateral gate between TM2b and TM7 of SecY (17). This gate has been suggested to play a role in channel opening, release of the signal sequence of translocating proteins, as well as in the insertion of transmembrane helices of nascent integral membrane proteins into the lipid bilayer (17).
To investigate the functioning of the lateral gate we identified amino acid positions in TM2b and TM7 in E. coli SecY (Fig. 1A) that based on the M. jannaschii SecYE␤ crystal structure (17)    (17)). Indicated are amino acids Ser-87 and Phe-286 in TM2b and TM7, respectively. The figure was created using MOLMOL (38) and POV-Ray.
( Fig. 1B) would be within disulfide bonding distance of each other. As controls, we also selected amino acids that are predicted to be too far apart to form a disulfide bond. The selected amino acids were replaced by cysteine residues via site-directed mutagenesis using a cysteine-less (Cys-less) SecY as template. Subsequently, the various single cysteine mutants in TM2b were combined with single cysteine mutants in TM7 to form double cysteine mutant pairs (Table 1). From the possible combinations, we specifically selected the pair F286C/S87C, as the homologous amino acids (Asn 268 and Thr 80 , respectively) in the M. jannaschii SecYE␤ form hydrogen bonds that need to be broken to open the lateral gate (17) ( Table 1). The various SecY mutants were cloned into a secYEG expression vector and expressed in E. coli strain SF100. SDS-PAGE analysis and immunoblotting using an antibody against the N-terminal histidine tag in SecY showed that all mutants were (over)expressed at similar levels (see Fig. 2 and supplemental Fig. S2). In vitro translocation experiments using fluorescein (FL)-labeled proOmpA as substrate and IMVs from cells overproducing the different single and double cysteine mutants showed that under reducing conditions all mutants translocate proOmpA to similar levels as the Cys-less SecYEG (supplemental Fig. S1).
Chemical Cross-linking of the Putative Lateral Gate Formed by TM2b and TM7-To visualize the cross-linking between cysteines introduced in TM2b and TM7, an assay was developed based on the specific cleavage of SecY by the outer membrane protein OmpT. OmpT cleaves SecY in the fourth cytoplasmic domain (C4) between two arginine residues at positions 255 and 256 (Fig. 1A) (27). Cleavage results in a N-terminal fragment of SecY with a apparent molecular mass of 22 kDa that is readily detected by Coomasie Brilliant Blue-stained SDS-PAGE ( Fig. 2A and supplemental Fig. S2). When TM2b is cross-linked to TM7, OmpT will cleave SecY but the two halves will not be separated on non-reducing SDS-PAGE and SecY will migrate as a full-length protein.
To oxidize the cysteine pairs in SecY, the hydrophilic oxidizer NaTT was used to form a disulfide bond. The use of the strong and hydrophobic oxidizer copper phenantroline was avoided as this agent can cause protein aggregation, whereas there is the risk of a formation of thiol-copper-thiol bridge rather than a disulfide bond. DTT was used as a reducing agent to reverse disulfide bond formation. After NaTT treatment, IMVs were incubated with the protease OmpT and analyzed by SDS-PAGE and Coomassie Brilliant Blue staining ( Fig. 2A). Under both oxidizing and reducing conditions, OmpT treatment of the IMVs containing Cys-less SecY resulted in the complete cleavage of SecY as seen as a disappearance of full-length SecY with the concomitant formation of the 22-kDa N-terminal fragment ( Fig. 2A, lanes 15 and 17). Moreover, addition of the reducing agent DTT after OmpT digestion ( Fig. 2A, lane 18) did not affect the cleavage pattern ( Fig. 2A, compare lane 17 versus 18). It should be noted that samples containing DTT were loaded on a separate SDS-PAGE gel to prevent the DTT from diffusing and affecting neighboring cross-linked samples. For this reason there is a slight difference in migration of fulllength SecY and the derived N-terminal fragment of the same samples with and without DTT (see Fig. 2). The same results as shown for Cys-less SecY were obtained with all single cysteine mutants ( Fig. 2A, lanes 1-12, and supplemental Fig. S2, lanes [1][2][3][4][5][6][7][8][9][10][11][12][13][14][15][16][17][18]. In contrast, after oxidation of IMVs containing the double cysteine mutant SecY F286C/S87C, OmpT treatment resulted in the formation of a fuzzy band that corresponds to full-length SecY without appearance of the 22-kDa N-terminal fragment. This was not due to an incomplete digestion or NaTT-induced protein aggregation, as addition of DTT to the sample after OmpT treatment resulted in the conversion of full-length SecY into the 22-kDa N-terminal fragment ( Fig. 2A,  lane 24). This indicates that NaTT treatment results in the formation of a disulfide bond between the selected cysteine pair at 286 and 87. The fuzzy character of the protein band corresponding to OmpT cleaved but full-length SecY still indicates an altered conformational state of SecY in SDS-PAGE, a phenomenon often seen in cross-linking of membrane proteins. In comparison with the other double cysteine mutants, disulfide bond formation was very efficient with the mutant F286C/S87C and weaker for the F279C/M83C (compare Fig. 2A, lanes  19 -24 versus supplemental Fig. S2, lanes 19 -30) and I283C/ S87C (data not shown) pairs. Cross-linking was absent with the negative control, mutant F279C/M83C (supplemental Fig. S2, lanes 19 -24), confirming the SecYEG structure that indicated that these residues are not within disulfide bonding distance. For this reason, the SecY F286C/S87C mutant was used in the remainder of this study.
To verify the oxidation and OmpT digestion, formation of the N-terminal 22-kDa SecY fragment was validated by Western blotting using an antibody directed against the His 6 tag (Fig.  2B). OmpT treatment of NaTT-oxidized IMVs containing Cysless SecYEG and single cysteine mutants resulted in the complete conversion of SecY into the 22-kDa N-terminal fragment (Fig. 2B, lanes 1-18). On the other hand, in oxidized IMVs containing the F286C/S87C mutant, substantial levels of intact SecY remained after OmpT treatment (Fig. 2B, lane 21). The full-length SecY was converted into the 22-kDa N-terminal SecY fragment after DTT treatment (Fig. 2B, lane 24). The small amount of N-SecY fragment in Fig. 2B, lane 21, indicates incomplete disulfide bond formation. However, because of the blotting transfer of full-length SecY, the N-terminal fragment and the cross-linked products likely differ, data quantification was based on the full-length SecY visualized by SDS-PAGE. NaTT-mediated oxidation of the F286C/S87C SecY mutant pair results in a cross-linking efficiency of at least 80% (refer to Fig. 4A, lane 4, for cross-linking efficiency).
Oxidative Cross-linking of the Lateral Gate Abolishes ProOmpA Translocation and SecA Translocation ATPase Activity-To determine whether cross-linking of the cysteines in the SecY F286C/S87C mutant has an effect on protein trans-location, we analyzed the in vitro translocation of FL-labeled proOmpA into IMVs that had been oxidized with NaTT or reduced with DTT. With IMVs containing Cys-less SecY, NaTT treatment had little effect on proOmpA translocation (Fig. 3A, compare lanes 3 and 4), nor did it affect proOmpA translocation into IMVs containing the SecY single Cys-286 (Fig. 3A, lane 10) and Cys-87 (Fig. 3A, lane 8), although the latter showed a reduced activity that can be attributed to the mutation. In contrast, NaTT treatment of IMVs containing the double cysteine mutant F286C/S87C was nearly completely abolished to the very low levels found for the endogenous wildtype levels of SecYEG (Fig. 3A, compare lane 6 versus 2). The activity of the F279C/I86C mutant was reduced to a lesser extent by NaTT oxidation in line with the observed weaker disulfide bonding efficiency (supplemental Fig. S3, lanes 3 and  4). As expected, proOmpA translocation into NaTT-treated IMVs containing the negative control SecY F279C/M83C was not affected (supplemental Fig. S3, lanes 5 and 6). These conditions also do not lead to the formation of a disulfide bond between the two positions.
Because proOmpA translocation is a SecA-dependent process, the effect of the lateral gate cross-linking on the proOmpAstimulated ATPase activity was determined. IMVs containing the Cys-less, single cysteine 87 and 286, and the double cysteine F286C/S87C SecYEG mutant were treated with NaTT, and analyzed for the proOmpA-stimulated SecA translocation ATPase activity. NaTT treatment of IMVs containing Cys-less SecYEG did not significantly affect the SecA translocation ATPase activity (Fig. 3B). Similar results were obtained with IMVs containing the single cysteine mutants SecY Cys-87 and Cys-286, although the activity of the SecY Cys-87 mutant was reduced as compared with the Cys-less control. The latter explains the reduced proOmpA-stimulated ATPase activity of the untreated double cysteine mutant SecY F286C/S87C (Fig.  3B), and possibly signifies a functional role of Ser-87 in translocation. Importantly, upon NaTT treatment of IMVs containing the double cysteine mutant SecY F286C/S87C, the proOmpA-stimulated ATPase activity of SecA was completely abolished. These data demonstrate that immobilization of the lateral gate of the translocon by disulfide-bonded cross-linking inhibits SecA-mediated preprotein translocation and the SecA translocation ATPase activity.
Immobilization of the Lateral Gate of SecY by Chemical Cross-linkers with Varying Length-NaTT forms a disulfide bond between the two cysteine residues with an approximate distance of 2 Å. Oxidation is reversible by the use of the reductant DTT. To analyze the effect of the cross-linking distance of the two cysteines on protein translocation, chemical cross-linkers were employed that have spacer arms with increasing lengths (supplemental Fig. S5). bBBr forms a covalent bond between thiol groups and has a rigid cross-linking distance of 5 Å (28,29). bBBr has the particular characteristic that it becomes fluorescent when both of its alkylating groups have reacted with thiols (30). BMOE is a rigid cross-linker that contains a thioether-linked spacer of ϳ8 Å, whereas BMH is a longer yet more flexible cross-linker with a maximal ϳ13-Å spacer arm (31). With the SecY F286C/S87C mutant, bBBr showed similar cross-linking efficiency as NaTT at 84 and 81%, respectively (Fig. 4A, compare lane  5 and 4). The cross-linking efficiencies of the longer cross-linking molecules BMOE and BMH were somewhat lower, i.e. 56 and 59%, respectively (lanes 6 and 7), but clearly discernable upon OmpT treatment. None of these cross-linkers interfered with the efficiency of OmpT digestion as validated with the Cys-less SecYEG complex digested by OmpT in the presence of these compounds (Fig. 4A, lanes  9 -13). In addition, the cross-link between cysteines 286 and 87 after bBBr treatment could be directly demonstrated by UV exposure of a SDS-PAGE gel showing a bright fluorescent band at the position of fulllength SecY only with the F286C/ S87C mutant (Fig. 4B, lane 4) and that is absent in the bBBr-treated single cysteine and Cys-less mutants (lanes 1-3). The direct fluorescent visualization provides further evidence for the formation of a cross-link between Cys-286 and Cys-87 in addition to the OmpT assay.
IMVs containing the Cys-less and F286C/S87C mutations were treated with cross-linkers NaTT, bBBr, BMOE, and BMH as well as with the reductant TCEP, respectively. TCEP instead of DTT was used here as it is capable of reducing thiol in the presence of reactive maleimides. Next, in vitro translocation assays were performed using Texas Red-labeled proOmpA as substrate. Texas Red-proOmpA has an emission spectrum that does not overlap with the emission spectrum of bBBr, thus allowing for the simultaneous detection of proOmpA translocation and cross-linking of the lateral gate in SecY F286C/S87C. At the limiting amounts of IMVs used, the activity of SecYEG is rate-determining for translocation (supplemental Fig. S4) and thus the activity can be quantified and compared. None of the chemical cross-linkers affect the translocation of proOmpA into the Cys-less SecYEG IMVs (Fig. 4C, lanes 8 -12). As shown before, NaTT treatment of SecY F286C/S87C IMVs reduced translocation to the levels observed with wild-type IMVs containing only endogenous levels of SecYEG (Figs. 3A, lane 6, and  4C, lane 2). A strong reduction in activity was also observed when SecY F286C/S87C IMVs were treated with bBBr (Fig. 4C,  lane 3), whereas the longer cross-linkers BMOE and BMH had little effect on proOmpA translocation (Fig. 4C, lanes 4 and 5). Quantification of the data shows that translocation efficiency increases with an increasing length of the spacer arm reaching  7 and 8), or Cys-87 (lanes 9 and 10) SecYEG. Prior to translocation, IMVs were oxidized or reduced with NaTT (lanes 4, 6, 8, and 10) and DTT (lanes 2, 3, 5, 7, and 9), respectively. Lane 1 shows a 20% FL-proOmpA standard. B, SecA translocation ATPase activity of IMVs containing Cys-less, F286C/S87C, Cys-286, or Cys-87 SecYEG and that had been preincubated with (black bars) or without (white bars) NaTT. Error bars represent S.D.
the same level of activity as untreated IMVs with the longest cross-linker BMH (Fig. 4D, black bars).
We also addressed the reversibility of oxidation and crosslinking. Addition of DTT after the cross-linking reaction restored translocation to normal levels only for the NaTT-oxidized F286C/S87C IMVs (Fig. 4D, gray bars). As expected, DTT had no significant effect on the translocation of proOmpA into IMVs treated with the irreversible cross-linkers bBBr, BMOE, and BMH. It should be noted that none of these compounds affected the activity of the single Cys-87 mutant (supplemental Fig. S6). Thus the inhibition of proOmpA translocation is due to a constrained lateral gate and is not caused by alkylation of a single cysteine position.
Next we performed SecA ATPase assays using SecY F286C/ S87C IMVs treated with the various cross-linkers. Whereas these compounds did not affect the basal SecA ATPase activity of the IMVs (data not shown), the proOmpA-stimulated ATPase activity increased when the length of the spacer arm of the cross-linker (Fig. 4E) matches their effect on proOmpA translocation. Although the reduced efficiency of BMH cross-linking will contribute to the observed translocation and SecA translocation ATPase activities, it does not explain the entirely undisturbed activities with the BMH cross-linked F286C/S87C SecY (Fig. 4, D and E). Therefore, these data lend strong support for the notion that SecYEG lateral gate opening and activation of the SecA ATPase are linked processes.
Cross-linking of the Lateral Gate of a Translocon Containing a Protein Translocation Intermediate-A pre-existing disulfide bond between positions 286 and 87 of the lateral gate of SecY prevents preprotein translocation. To access the proximity of these positions during translocation, a translocation intermediate was generated using a proOmpA-DHFR fusion protein (25,32,33). Upon addition of ligands NADPH and methotexrate, the DHFR moiety folds and blocks further translocation yielding a translocation intermediate that is stably arrested in the translocase. To ensure that proOmpA-DHFR efficiently blocks the F286C/S87C SecYEG pore, translocation was performed with proOmpA-DHFR in excess of the translocation sites whereupon IMVs were re-isolated and incubated with Texas Red-proOmpA for a second round of translocation. Unlabeled proOmpA translocated during the first incubation period did not interfere with subsequent translocation of Texas Red-proOmpA (Fig. 5A, lane 2). However, the folded DHFR domain in the proOmpA-DHFR fusion protein nearly completely blocked the translocation sites as translocation of Texas Red-proOmpA in the second round was greatly reduced (Fig.  5A, lane 3). Next, the SecYEG F286C/S87C IMVs loaded with the proOmpA-DHFR translocation intermediate were incubated with different chemical cross-linkers and formation of a bond between Cys-286 and Cys-87 was analyzed using the OmpT protease assay and gel staining with SYPRO Ruby. Oxidation with NaTT or addition of the short cross-linker bBBr to the IMVs containing the translocation intermediate yielded almost similar levels of cross-linked SecY as observed with the unoccupied pore (Fig. 5B, lanes 3 and 4, compare with Fig. 4A,  lanes 4 and 5). This demonstrates that the cysteines are within disulfide bonding distance when the pore is occupied with a translocation intermediate. On the other hand, cross-linking with the longer spacer length BMOE and BMH was very inefficient with a dramatic reduction of the amount of full-length SecY upon OmpT treatment concomitant with the formation of a prominent 22-kDa SecY fragment (Fig. 5B, lanes 5 and 6). This is in marked contrast to the efficiency of cross-linking in the absence of a translocation intermediate (see Fig.  4A, lanes 6 and 7). These data demonstrate that Cys-87 in TM2 and Cys-286 in TM7 are in close proximity when the pore is occupied by a translocation intermediate. Likely because of geometrical constraints, these positions can no longer be cross-linked by the chemical reagents with longer spacer arms.

DISCUSSION
In this study, we have addressed the role of the lateral gate in translocation of preproteins across the cytoplasmic membrane. On the basis of homology and sequence alignment with M. jannaschii SecY (17) we have engineered pairs of cysteine residues into the putative TM2b/TM7 lateral gate region of E. coli SecY. This allowed site-specific disulfide-bonded cross-linking of positions that were predicted to be in close vicinity, whereas more remotely introduced cysteines did not yield cross-links. Recently, two additional SecYE translocon structures have been described, notably from bacteria known to contain SecA as opposed to the archaeon M. jannaschii (14,34). The high resolution structure of SecYE from Thermus thermophilus is in a antibody-stabilized pre-open state (34), whereas the structure of SecYEG from T. maritima is with SecA bound in an intermediate state of ATP hydrolysis (35). Strikingly, compared with the M. jannaschii SecYE␤ structure, the T. maritima SecA-bound SecYEG structure shows a partial opening of the lateral gate region around TM2b and TM7. Our biochemical cross-linking data for E. coli SecYEG on the lateral gate region are in line with this structural observation, and importantly, we demonstrate that the opening of this lateral gate is required for protein translocation and activation of the SecA ATPase.
The introduction of double cysteine mutants in TM2b and TM7 allowed an efficient cross-linking of the lateral gate. When fully oxidized, translocation is completely abolished. However, when a cross-linker is introduced with sufficient spacer length (larger than 5 Å), translocation occurred unrestricted. These observations support two main conclusions: (i) translocation of proOmpA occurs via a single translocon in line with a previous report (10); and (ii) the lateral gate region needs to open up to allow translocation to take place. It should be noted that formation of a functional consolidated pore between two opposing SecYEG translocons seems unlikely because the presence of, for instance, the BMH cross-link in the lateral gate of one of the pores will prevent diffusion of the polypeptide substrate to the other pore.
Strikingly, there is a strong cooperativity between the ability of SecA to hydrolyze ATP and the opening of the lateral gate. Oxidative cross-linking of the lateral gate resulted in a complete loss of the SecA translocation ATPase activity, whereas chemical modification of the lateral gate with bimolecular cross-linkers with longer spacer arms (8 -13 Å) supported SecA translocation ATPase activity up to the level observed with an unconstrained lateral gate. A recent study on the role of the E. coli SecA ATPase two-helix finger suggested that the helical scaffold domain, that harbors the two-helix finger, moves up and down inside the translocon with ATP hydrolysis cycles resulting in the pushing of the preprotein substrate into the translocon (35). Two cysteine positions on proOmpA could be cross-linked to both position 282 of SecY together with a cysteine on the tip of the two-helix finger (35). This is very close to cross-linked positions 286 and 87 of the lateral gate as shown in this study. This further highlights the importance of the region. Tsukazaki et al. (34) suggested that SecYE predominantly assumes the closed conformation, whereas binding of SecA induces a conformational change of the translocon from the closed to the pre-open state where a swinging of TM8 possibly induces an opening of the lateral gate region. Taken together, our data shows that a tight physical interaction between TM2b and TM7 is incompatible with preprotein translocation lending strong support for the hypothesis that the clamshell structure of the SecYEG translocon needs to open for the SecA-dependent initiation of protein translocation.
A recent molecular dynamic simulation study modeled conformational changes in SecYE␤ upon translocation initiation where a force of 2-3 nN is required to move TM2b and TM7 apart (21). Two independent events are proposed to require this amount of force. The first event is the opening of the gate to around 2 to 5 Å, which involves a latch-like movement of SecE and which requires a large force. The second event is the displacement of the plug domain (TM2a), which allows the lateral gate to open to 6 -9 Å. Cross-linking studies suggested that the signal sequence from a translocating protein binds the region between TM2b and TM7 (36) and based on this observation it was proposed that the signal sequence inserts between TM2b and TM7 resulting in a displacement of the plug and the concomitant translocation of the mature preprotein domain. Our biochemical data are consistent with the molecular dynamics simulation as a SecY channel in which the gate is fixed by a cross-linker with a 8-Å or longer spacer arm allows normal translocation of proOmpA, whereas a cross-linker of 5 Å or shorter abolishes protein translocation. Interestingly we also observed that TM2b and TM7 can be efficiently cross-linked utilizing the oxidizer NaTT and the short cross-linker bBBr even when the pore contains a preprotein translocation intermediate. As the major fraction of the preprotein translocation intermediate is processed by leader peptidase (32) (data not shown), the signal sequence is no longer present in the complex. In contrast, the longer cross-linkers BMOE and BMH were no longer able to cross-link the cysteine positions in TM2b and TM7 when the pore was occupied by a translocation intermediate, despite their hydrophobicity that may facilitate access of the cysteine positions via the lipid phase. This suggests that the lateral gate region is in a near to closed conformation with the translocation intermediate present. Most likely when SecA encounters the DHFR domain at the C terminus of the proOmpA-DHFR fusion, SecA remains in an ADP-bound state and is unable to unfold the DHFR domain (37). The latter may correspond to a state of the SecA ATPase in which the twohelix finger is not inserted into SecY (35). Possibly, under these conditions the SecA-SecY interaction corresponds to a more energetically favorable pre-open state (34). In the T. maritima SecA-SecYEG complex structure (14), the opening in the lateral gate between TM7 and TM2b is about 5 Å. Likely, longer crosslinkers such as BMOE and BMH will not be able to efficiently cross-link these selected positions when the flexibility of this region is constrained by the SecA transition state and/or the protein translocation intermediate. When the folding ligands of DHFR (NADPH and methotrexate) are removed further translocation of the DHFR domain occurs in an ATP and SecAindependent manner (33). Therefore, with a stalled and processed translocation intermediate, the SecY translocon most likely returns to a pre-open state with bound SecA, which corresponds to a narrowed lateral gate region.
In conclusion, for the first time we have provided experimental evidence for a catalytic requirement for the opening of the lateral gate at the interface of TM2b and TM7 of SecY during SecA-dependent protein translocation. Importantly, SecA activation and lateral gate opening are coordinated events consistent with an allosteric mechanism of channel opening. Eventually, signal sequences may leave the translocon via the TM2b and TM7 gate. Taken together, the SecYEG translocon seems to be a highly flexible structure that undergoes specific conformational changes during protein translocation. Future studies should address the function of the lateral gate in the insertion of membrane proteins.