Special codes are embedded in the primary sequence of newly synthesized proteins to determine their final destination. Protein translocation across biological membranes requires co-operation between the targeting and translocation machineries. A conserved membrane channel, the Sec61/SecY complex, mediates protein translocation across or integration into the endoplasmic reticulum membrane in eukaryotes and the plasma membrane in prokaryotes. A combination of recent biochemical and structural data provides novel insights into the mechanism of how the channel allows polypeptide movement into the exoplasmic space and the lipid bilayer.
Protein translocation across the ER (endoplasmic reticulum) membrane in eukaryotes and the plasma membrane in prokaryotes is crucial for the biogenesis of secretory and many membrane proteins. The code for entering the translocation pathway is a signal sequence usually comprising a stretch of 7–15 hydrophobic residues flanked by a few N-terminal charged residues and C-terminal polar amino acids . Its hydrophobicity and conformational features, rather than specific primary sequence, are required to carry out the targeting function.
Protein translocation may occur by co-translational and post-translational mechanisms, using different sets of cytosolic partners for delivering the nascent peptide to the membrane [2,3]. Partitioning of nascent polypeptides between two translocation pathways is mainly on the basis of the relative hydrophobicity of the signal sequence . A more hydrophobic signal sequence binds the SRP (signal recognition particle) as soon as the signal sequence emerges from the ribosome. The nascent polypeptide is then targeted to the membrane through the interaction between the SRP and the membrane-localized SR (SRP receptor). Subsequently, the nascent polypeptide is translocated across the membrane during translation. Nascent polypeptides with a less hydrophobic signal sequence are not selected by the SRP for express delivery. The decoupling of protein synthesis and protein translocation causes the exposure of the full-length polypeptides in the cytosol. Dedicated chaperones bind to the hydrophobic regions of the peptides to keep the peptide in a translocation-competent state and prevent aggregation. The polypeptide–chaperone complex is recruited to the membrane through the interaction with the translocation channel. The different targeting mechanisms have been summarized previously . The two translocation pathways are not mutually exclusive, at least in yeast, as cells can adapt to the loss of SRP by dynamic partitioning between the two pathways [6,7].
The central component of the translocation channel is a heterotrimeric complex of membrane proteins, called the Sec61p complex (Sec61α, Sec61β and Sec61γ) in eukaryotes and the SecYEG complex in prokaryotes. The α subunit, the major component of the complex, is highly conserved and has ten TM (transmembrane) spans [8,9]. The γ subunit (Sec61γ in mammals, Sss1p in Saccharomyces cerevisiae and SecE in bacteria) of the Sec61 complex is encoded by an essential gene . The conserved region consists of a single TM span and several surrounding residues . The γ subunit may be important for complex stability and the oligomerization of the translocon [12,13]. The β subunits do not show similarity in sequence and topology among species . Although the Sec61 β subunit is not essential for cell viability in yeast, the mammalian β subunit exhibits a large stimulatory effect in an in vitro translocation assay . This review will mainly focus on the recent discoveries regarding the structure of the channel and how the translocon allows protein translocation across and integration into the membrane.
On the basis of cryo-EM (electron microscopy) analyses of the purified Sec61p complex and Sec61–ribosome complex, it was initially proposed that the channel is assembled from 3 or 4 Sec61 heterotrimers. The centre of the oligomer was thought to form the protein-translocating channel . However, the X-ray structure of the SecYEβ complex from Methanococcus jannaschii suggests that one copy of the Sec61 heterotrimer serves as the translocation channel . The ten TM segments of the α subunit (SecY) can be viewed as two linked halves (TM spans 1–5 and 6–10) which form an hourglass-shaped interior containing a ring of hydrophobic residues at the constriction. A short helix (TM2a) occupies the luminal exit of the channel and could serve as a moveable plug for the channel. A site-directed disulfide cross-linking experiment has shown that a translocating substrate passes through the inside of SecY and makes the most contact with residues near the constriction . Photo-cross-linking experiments located the binding site for the signal sequence to the interface between TM2 and TM7 , which is proposed to be the lateral gate towards the lipid bilayer. Lateral opening of the channel would allow the intercalation of the signal sequence between TM2 and TM7 and the passage of TM spans of membrane proteins into the lipid bilayer.
Interaction sites for the cytosolic partners
After being targeted to the ER membrane, transfer of the RNC (ribosome–nascent chain complex) from the SRP to an unoccupied translocon is mediated by the interaction between the ribosome, the SR and the cytosolic domains of the translocon. The ribosome-binding sites on the translocon were mapped to the two large cytosolic loops in the α subunit [L6 (loop 6) and L8 (loop 8)] by trypsin digestion . Further systematic mutagenesis of those two loops in yeast revealed that a few highly conserved basic residues are essential for co-translational translocation pathway . In the X-ray structure of SecYEβ, L6 and L8 are highly exposed in the cytosol and project approx. 20 Å away from the membrane surface . The docking of the archeal SecY structure into a cryo-EM structure of Escherichia coli ribosome–SecY complex revealed that those two loops form a rod-like feature which extends into a depression on the docking surface of the ribosome . L6 and L8 also form the main ribosome–translocon connections, as shown in a recent cryo-EM structure of eukaryotic Sec61 complex interacting with the translating ribosome  (Figure 1A). In both cryo-EM studies [21,22], L6 and L8 seem to adopt a more extended conformation upon ribosome binding. Thus L6 and L8 constitute a conserved functional domain which serves as the docking site for the ribosome. Charge reversal mutations in those conserved basic residues abolished ribosome-binding activity, suggesting the electrostatic nature of interaction. Indeed, extensive contact between those two loops and a ribosome RNA was observed in the cryo-EM structure and the conserved basic residues are at the tip of those loops [21,22] (Figure 1B).
L6 and L8 of the α subunit form the ribosome-binding site
One puzzling observation was that although mutations of conserved basic residues in both L6 and L8 cause severe defects in co-translational translocation, only the latter dramatically decreases ribosome-binding affinity in yeast [20,22]. A hint of an explanation was provided by the cryo-EM analysis of the RNC in yeast and mammals, which shows that L6 is in very close proximity to the nascent chain . One interesting possibility would be that L8 is essential for ribosome binding, whereas L6 interacts with the nascent peptide and guides its insertion into the channel. In contrast with the yeast Sec61 complex, mutations in either loop result in a loss of ribosome binding in E. coli . The longer and more positively charged L6 in E. coli SecY may be involved in both steps mentioned above. Whether L6 serves as an usher for the nascent peptide in vivo requires further investigation.
Although Sec61β does not form the primary binding site for the ribosome , the cytosolic domain of Sec61β binds ribosomes with an affinity comparable with the Sec61 heterotrimer . Cross-linking studies of in vitro translocation revealed that Sec61β is very close to rpL17, a mammalian ribosomal protein located near the polypeptide exit site . Ubiquitin translocation assays show that yeast cells lacking Sbh1 and Sbh2 (yeast Sec61β) have a severe co-translational translocation defect . A synthetic growth defect was observed when Sbh2 was deleted in cells expressing soluble SRβ, suggesting that the Sec61β subunit may provide a transient docking site to recruit the SRP receptor to the vicinity of the channel. Taken together, these findings suggest that the β subunit interacts with both the ribosome and the SRP receptor to facilitate the transfer of the RNC from the SRP to a vacant translocon.
Opening of the channel
The closed conformation of the channel (Figure 2A) observed in the SecYEβ crystal structure may represent the stable ground state of the channel, as indicated by a molecular dynamics simulation . The closed state could be destabilized upon binding of the ribosome or another cytosolic partner to facilitate the insertion of nascent polypeptides. Separation of the lateral gate interface to accommodate a signal sequence or a TM span could reduce the interaction between the plug and the pore-ring residues, in turn leading to opening of the luminal gate. Based on cysteine-cross-linking experiments, the plug domain (TM2a) was proposed to move out of the centre of the channel and towards the C-terminal end of SecE in a fully open channel [27,28]. A partially open lateral gate was observed in a recent crystal structure of the Thermatoga maritima SecYEG–SecA complex . Upon SecA binding, a relative movement between the two halves (TM spans 1–5 and 6–10) expands the lateral opening to allow insertion of the two-helix finger of SecA (Figure 2C). The plug was partially dislocated towards the luminal side of the channel, supporting the concept that opening of the two gates is intimately connected. Binding of an anti-SecY Fab fragment to SecY mimics the SecY–SecA interaction without the two-helix insertion into the channel . Only a crack was observed on the cytoplasmic side of the lateral gate in the crystal structure of SecY–Fab complex (Figure 2B). In this case, no movement of the plug domain was detected. The binding sites for SecA and the ribosome partially overlap . It is likely that the two conformations may represent sequential conformational changes in the opening of the channel in both the co- and post-translational modes of translocation.
Multiple stages in channel opening
Surprisingly, a complete deletion of the plug domain is not lethal . Deletion of the plug allows the transport of proteins with defective signal sequences [32,33], suggesting that one role of the plug is to stabilize the closed conformation of the channel and maintain translocation fidelity. Crystal structures of the plug deletion mutants revealed a conformational change in adjacent SecY segments to form new plug domains . It is unclear how the membrane permeability barrier is maintained in the larger plug deletion mutants. Additional proteins may be involved in gating of the translocon, including luminal BiP (immunoglobulin heavy-chain-binding protein), as suggested by fluorescence-quenching experiments .
Oligomerization of the channel
Whether a functional translocon is formed by more than one Sec61/SecY complex in vivo is still unclear. Three-dimensional EM reconstructions of the ribosome–Sec61 complex and the RNC–Sec61 complex revealed four stalk-like connections between the channel and the ribosome [35,36], which were thought to emerge from the three or four Sec61 heterotrimers per native channel. A two-dimensional crystal structure of the E. coli SecYEG complex is a dimer . Oligomerization of the channel was also supported by the observation that a defective SecY monomer that is covalently linked to a wild-type monomer can translocate nascent peptides in vitro .
However, a 16 Å cryo-EM structure of an E. coli ribosome–SecYEG complex shows one monomer of the SecYEG complex bound to a non-translating ribosome . Recently, a higher resolution cryo-EM structure was obtained for a translating ribosome docking on to the yeast Ssh1 complex (9 Å) or the mammalian Sec61 complex (6.5 Å) . Again, only a monomer can be accommodated in the observed density for the channel and the extra density surrounding the fitted monomer is probably contributed by protein-bound lipids and detergents  (Figure 3). Moreover, the four connections between the ribosome and the monomeric Sec61 complex corresponded to L6, L8 and the N- and C-termini of the Sec61α subunit respectively.
A monomer of mammalian Sec61 complex binds the RNC
The arrangement of the Sec61 complexes in the proposed oligomer is also under debate. The exit of the TM domains from the channel would suggest the lateral gate of the Sec61 complex faces the phospholipids. A back-to-back orientation was observed in the two-dimensional crystal structure of SecYEG dimer , whereas an earlier cryo-EM reconstruction of the E. coli SecYEG bound to an RNC favours a front-to-front arrangement with lateral gates in contact .
What is the significance of oligomerization if an oligomer of the Sec61 complex does exist? Oligomerization may be required to provide a high-affinity binding site for SecA  or the ribosome . Another possibility is to recruit important accessory components, such as the TRAP (translocon-associated protein), OST (oligosaccharyltransferase) and TRAM (translocating chain-associating membrane) complexes, to the translocation site .
INTEGRATION OF MEMBRANE PROTEINS DURING TRANSLOCATION
Integration of membrane proteins is considerably more complicated than translocation of secretory proteins. Challenging tasks for the translocon include orientation of TM spans, integration of TM spans into the lipid bilayer and localization of cytoplasmic and luminal domains. How does the translocon recognize and interact with different domains of nascent polypeptides and undergo conformational changes to ensure the efficiency and fidelity of membrane protein integration? Those questions have been intensely investigated primarily using photo-crosslinking of in vitro-assembled translocation intermediates and the results have been interpreted in the context of known translocon structures.
A TM span may assume a final topology with either a Ncyt–Cexo (cytoplasmic N-terminus–exoplasmic C-terminus) or Nexo–Ccyt orientation. The orientation of a TM span is primarily influenced by the uneven charge distribution of the flanking residues. As proposed in the ‘positive inside’ rule , the more positively charged flanking sequence is often found on the cytosolic end of a TM span. The uneven charge distribution probably affects the orientation of a TM span by electrostatic interaction with the negatively charged phospholipids , as well as charged protein residues in the translocon . In addition, rapid folding of an N-terminal domain may lead to a type II (Ncyt–Cexo) orientation , while extending the hydrophobic core of a single-TM protein would favour the translocation of the N-terminus [45,46]. Different models have been proposed to explain how a signal sequence or a TM span enters the translocon and adopts the correct orientation . In vitro experiments using model membrane proteins indicate that the TM span can undergo dynamic reorientation in the translocon prior to adoption of the final orientation [47–49]. After insertion into the channel, a TM span of nascent membrane protein binds in the signal-sequence-binding site and can be cross-linked to phospholipids . The hydrophobicity of a TM may determine how fast it can be released from the channel. A more hydrophobic TM partitions into the lipid rapidly, whereas a less hydrophobic TM stays close to the channel for an extended time period .
Little is known about the integration of polytopic membrane proteins. The insertion of multi-TM-spanning proteins occurs sequentially and the final topology is primarily determined by the insertion orientation of the initial TM span. Individual TM spans must sequentially move from the transport pore to the signal-sequence-binding site, and exit through the lateral gate of the channel into the membrane bilayer [49,50]. Conformational changes in the Sec61 heterotrimer that facilitate the lateral moment of TM spans between these three distinct environments and control lateral and luminal gating in the translocon are not well understood. Moreover, cross-linking studies revealed that TM spans of polytopic membrane proteins remain within or adjacent to the translocon for a prolonged period of time and traverse through different environments before final integration into the lipid [49,51–53]. An intriguing hypothesis is that the translocon and the associated proteins may provide a more favourable environment than the lipid for the folding of polytopic membrane proteins .
Although photo-cross-linking analysis of integration intermediates provides mechanistic insight into membrane protein integration, it does not provide kinetic information due to the long period between the translation of the nascent polypeptide and the photoactivation of the cross-linking reagent. In vivo kinetics of membrane protein integration was measured using UTA (ubiquitin translocation assay) reporters which are derived from a yeast type II membrane protein . Interestingly, integration of the UTA reporters is slow relative to elongation of nascent polypeptides. Several slow steps that may contribute to the delay of translocation include the aforementioned dynamic orientation of a TM span and opening of the lateral gate to allow exit of TM spans. Interestingly, a closed lateral gate was observed after the insertion of a type II TM span into the signal-sequence-binding site , which may explain the prolonged retention of a TM span within the channel and could delay translocation of the subsequent nascent chain. Fluorescence-quenching experiments have shown that the luminal gate of the translocon is in the closed state during synthesis of a cytosolic loop . A delay could also occur when adjacent TM spans co-operatively exit the translocon by passing through the lateral gate to enter the lipid bilayer . The delay in membrane protein integration may provide a time window that ensures correct topology for the integration of TM spans into the membrane. On the other hand, it also leads to cytosolic exposure of C-terminal TM spans and luminal domains, which could have a negative impact on the fidelity of membrane protein integration. Cytosolic chaperones may help prevent premature folding of TM spans and luminal proteins.
Although these recent discoveries provide tremendous insight into the translocation mechanism, many questions remain unaddressed. Marking a vacant translocon with the SR should minimize the search time for the RNC–SRP complex. However, the interaction between the SR and the translocon is not totally understood. How is the channel opened and gated differently in co- and post-translational translocation reactions? Do Sec61/SecY monomers oligomerize in the membrane and if so what are the biological functions of oligomerization? A high-resolution structure of a RNC–translocon complex is needed to provide a view of the translocon during active translocation. Deciphering the role of the translocon in the translocation and folding of membrane proteins is challenging. A combination of structural, biophysical and biochemical analysis of integration intermediates of more model and native membrane proteins is required to provide a general model for membrane protein biogenesis.
I thank Reid Gilmore (University of Massachusetts Medical School, Worcester, MA, U.S.A.) for critical reading of the manuscript prior to submission and for discussion.
Z.C. is currently a post-doctoral research associate at the Howard Hughes Medical Institute.