Protein translocation across the bacterial cytoplasmic membrane has been studied extensively in Escherichia coli. The identification of the components involved and subsequent reconstitution of the purified translocation reaction have defined the minimal constituents that allowed extensive biochemical characterization of the so-called translocase. This functional enzyme complex consists of the SecYEG integral membrane protein complex and a peripherally bound ATPase, SecA. Under translocation conditions, four SecYEG heterotrimers assemble into one large protein complex, forming a putative protein-conducting channel. This tetrameric arrangement of SecYEG complexes and the highly dynamic SecA dimer together form a proton-motive force- and ATP-driven molecular machine that drives the stepwise translocation of targeted polypeptides across the cytoplasmic membrane. Recent findings concerning the translocase structure and mechanism of protein translocation are discussed and shine new light on controversies in the field.
The synthesis of proteins by the ribosomal translation of messenger RNA takes place in the cytosol. Proteins with a function in another compartment of the cell or in the extracellular environment therefore have to be transported to their final location, which involves transmembrane translocation. In bacteria, secretory proteins are translocated across the cytoplasmic membrane by a protein complex termed the translocase (Wickner et al., 1991: the first and most comprehensive review on translocase). After translocation across the Gram-negative inner membrane, proteins are either retained in the periplasm or further transported to the outer membrane (Pugsley, 1993; Duong et al., 1997; Danese and Silhavy, 1998). In addition to this general secretory pathway, the bacterial cell envelope possesses protein translocation systems dedicated to the secretion of defined subsets of proteins. A cytoplasmic membrane translocation system has been described for proteins that bind complex redox cofactors (Berks et al., 2000). Gram-negative bacteria often secrete degradative enzymes or, in the case of pathogens, virulence factors and toxins via translocase-independent pathways that may cross both the inner and the outer membrane (Binet et al., 1997; Hueck, 1998).
The eukaryotic cell contains a variety of protein translocation systems. Nucleocytoplasmic transport of proteins and other macromolecules is performed by the large, but highly specific, nuclear pore complexes (Görlich and Mattaj, 1996). Protein secretion starts with the co-translational translocation across the endoplasmic reticulum (ER) membrane via ribosome–nascent chain complexes bound to a protein-conducting channel termed the translocon (Walter and Lingappa, 1986; Walter and Johnson, 1994; Rapoport et al., 1996). After translocation, proteins are sorted via an intricate vesicle transport system that connects the ER with the Golgi apparatus, the other components of the endomembrane system and the plasma membrane (Palade, 1975; Rothman and Wieland, 1994; Schekman and Orci, 1996). Proteins that bear organelle-specific targeting signals are imported by specialized protein translocation systems. Such systems are present in peroxisomes (Subramani, 1996; Erdmann et al., 1997), mitochondria (Schatz, 1996; Neupert, 1997) and chloroplasts (Schnell, 1995; Heins and Soll, 1998). Mitochondria and chloroplasts also possess protein export systems in their inner membranes that are remnants of their bacterial evolutionary origin (Schatz and Dobberstein, 1996; Settles and Martiensen, 1998).
The bacterial translocase has been extensively studied in the Gram-negative bacterium Escherichia coli. It consists of a heterotrimeric integral membrane protein complex composed of SecY, SecE and SecG, and a peripherally bound cytosolic ATPase, SecA. The SecYEG complex constitutes a highly conserved protein translocation pathway. It is homologous to the eukaryotic Sec61p complex that forms the translocon of the ER membrane (Hartmann et al., 1994). The mammalian Sec61p complex is composed of the Sec61α, β and γ subunits, which are termed Sec61p, Sbh1 and Sss1p in yeast (Hartmann et al., 1994; Panzner et al., 1995). Protein translocation in the mammalian ER is mainly co-translational and is driven by the concomitant synthesis of nascent polypeptides by Sec61p-bound ribosomes (Görlich and Rapoport, 1993). In yeasts, a post-translational translocation pathway co-exists that is mediated by Sec61p, the accessory Sec62/63 membrane protein complex and the BiP ATPase of the ER lumen (Panzner et al., 1995). Archaeal homologues of SecY (Sec61α) and SecE (Sec61γ), but not of SecA, have been identified (Pohlschröder et al., 1997), and archaeal protein translocation may therefore be co-translational. Homologues of SecY and SecE have been identified in an ancient mitochondrion by genomic analysis (Lang et al., 1997). Finally, a homologue of translocase mediates protein translocation into the thylakoid of chloroplasts (Yuan and Cline, 1994; Settles and Martiensen, 1998). In summary, protein secretion is mediated by a conserved protein-conducting channel and by pushing force (SecA or the ribosome) or pulling force (BiP) generators that interact with this channel (Fig. 1).
Identification of the E. coli translocase
Precursors of E. coli periplasmic and outer membrane proteins are synthesized with an amino-terminal signal sequence. This signal sequence has a length of around 20 amino acids and comprises a short stretch of positive charges, followed by a hydrophobic region and a proteolytic cleavage site for its removal after translocation (Randall and Hardy, 1989; Von Heijne, 1990). Signal sequences alter the folding properties of the mature part of precursors (Park et al., 1988) and interact with the lipid bilayer of the cytoplasmic membrane on account of their physicochemical properties (De Vrije et al., 1990; Keller et al., 1996). Although the presence of a signal sequence is not absolutely required for the translocation of secretory proteins (Derman et al., 1993; Flower et al., 1994; Prinz et al., 1996), mutagenesis of signal sequences has been used successfully to identify components of the protein translocation pathway in E. coli. Genes that were identified in this manner contained mutations that allowed the correct localization of precursor proteins with a defective signal sequence and were therefore termed prl genes, for protein localization. The prl and sec genes are allelic and encode the SecY (PrlA), SecE (PrlG) and SecA (PrlD) proteins (reviewed by Bieker et al., 1990; Schatz and Beckwith, 1990; Danese and Silhavy, 1998). The role of these proteins in protein translocation was unequivocally demonstrated by their purification and the subsequent reconstitution of the protein translocation reaction in vitro (Cabelli et al., 1988; Brundage et al., 1990; Akimaru et al., 1991). SecA is a soluble ATPase that binds peripherally to the cytoplasmic membrane, whereas SecY and SecE are integral membrane proteins that can be purified from solubilized cytoplasmic membranes as a stable complex. In vitro translocation is strongly stimulated by SecG (also termed band 1 or p12), a membrane protein that co-purifies with SecYE (Brundage et al., 1990; Douville et al., 1993; Nishiyama et al., 1993). Signal sequence suppressor mutations have been identified in the secG gene and are termed prlH (Bost and Belin, 1997). The SecYEG heterotrimer suffices to reconstitute efficient SecA- and ATP-dependent preprotein translocation in proteoliposomes prepared with purified E. coli phospholipids (Hanada et al., 1994; Douville et al., 1995; Van der Does et al., 1998). Importantly, the reconstituted translocase allows multiple rounds of protein translocation (Bassilana and Wickner, 1993; Manting et al. 2000) and the integration of hydrophobic segments in a translocating polypeptide into the lipid bilayer (Duong and Wickner, 1998).
Accessory membrane proteins
Although the SecYEG–A complex constitutes a functional entity, additional membrane proteins are involved in protein translocation across the cytoplasmic membrane. The SecD and SecF proteins were identified by an in vivo screen based on general protein secretion defects, and their absence renders the protein translocation reaction cold sensitive (Gardel et al., 1987; 1990). SecD and SecF form a complex together with a protein that is encoded by the yajC gene located in the same operon. This SecDF–YajC complex interacts with SecYEG and stabilizes a membrane-inserted state of the SecA protein (Kim et al., 1994; Economou et al., 1995; Duong and Wickner, 1997a,b). SecD and SecF expose large periplasmic domains (Pogliano and Beckwith, 1994). As prl mutations in these proteins have not been identified, they may act in a late stage of the translocation reaction. Antibodies against SecD prevent the release of translocated precursors from E. coli spheroplasts (Matsuyama et al., 1993), and cells depleted of SecD and SecF are unable to maintain a proton-motive force (pmf) across the cytoplasmic membrane (Arkowitz and Wickner, 1994). Homologues of the SecD and SecF proteins are present in the genome sequences of archaea that are to be devoid of a SecA homologue (Pohlschröder et al., 1997). By combining these findings, it appears as though SecDF–YajC is involved in closing the protein-conducting channel after completion of the translocation reaction. This event would relate to the release of the translocated protein on the periplasmic side of the membrane and prevent the aspecific passage of ions.
The translocation reaction is accompanied by the processing of preproteins to their mature form mediated by the signal peptidase (Dalbey et al., 1997). The signal peptide is proteolytically removed as soon as it emerges on the periplasmic side of the membrane (Schiebel et al., 1991) and does not have a further role after initiation of the translocation reaction. A newly identified component of the translocation machinery is the YidC protein (Scotti et al., 2000). It is homologous to the mitochondrial inner membrane protein Oxa1p involved in the biogenesis, a subset of inner membrane proteins with a long amino-terminal tail in the intermembrane space (Hell et al., 1998). An Oxa1p homologue in chloroplasts supports the integration of the light-harvesting chlorophyll-binding protein into the thylakoid membrane (Moore et al., 2000). YidC co-purifies with the SecYEG heterotrimer, and the cellular levels increase upon overexpression of SecYEG or SecDF(YajC). Cross-linking of YidC to transmembrane segments (TMSs) of translocating integral membrane proteins suggests that it may act either as a receptor for these domains when they leave the translocation channel laterally or function as an assembly factor (Scotti et al., 2000).
The targeting factors SecB and SRP
To ensure that newly synthesized proteins do not misfold and aggregate, the E. coli cell contains molecular chaperones. These chaperones have an important function in protein transport and prevent preproteins from folding into stable tertiary structures that are incompatible with translocation (Kumamoto, 1991). The Gram-negative bacterial protein SecB is a translocation-dedicated molecular chaperone that exists as a homotetramer of 16 kDa subunits (Fekkes and Driessen, 1999). It binds to the mature part of preproteins and facilitates their targeting to the translocase because of its affinity for the SecA protein, but is released immediately after the initiation of translocation (Hartl et al., 1990; Fekkes et al., 1997). SecB has also been implicated in the post-translational translocation of some extracellular proteins that are secreted by dedicated ABC (ATP binding cassette) transport systems (Delepelaire and Wandersman, 1998).
An alternative targeting route to the cytoplasmic membrane for precursor proteins is mediated by the signal recognition particle (SRP). The constituents of this ribonucleoprotein, a 48 kDa protein termed p48 or Ffh (fifty-four homologue) and a 4.5S RNA molecule, were identified on the basis of homology with subunits from the eukaryotic SRP involved in the translational arrest and targeting of ribosome–nascent chain complexes (RNCs) to the ER membrane (Poritz et al., 1988; Römisch et al., 1989). The E. coli SRP recognizes preprotein signal sequences and hydrophobic regions of nascent membrane proteins (Luirink et al., 1992; Valent et al., 1997). Together with its receptor, the GTPase FtsY, it is a functional substitute of the eukaryotic SRP-targeting pathway (Powers and Walter, 1997). Membrane lipid association activates FtsY for GTPase activity (De Leeuw et al. 2000), and this may regulate the SRP-mediated targeting of the RNCs. Upon the FtsY- and GTP-dependent release from SRP, RNCs can be cross-linked to SecA and the SecYEG complex present in isolated inner membranes, suggesting that the SecB- and SRP-mediated targeting routes converge at the translocase (Valent et al., 1998). The cross-linking of RNCs to SecA may result from their mutual affinity for the SecYEG complex and not represent a functional interaction, as targeting and transfer of the nascent chain from SRP to SecYEG also occurs in SecA-free membrane fractions or SecYEG proteoliposomes (Koch et al., 1999; Scotti et al., 1999). In contrast, for at least one SRP substrate, the AcrB protein, it has been demonstrated that the in vivo membrane targeting is dependent on SecA (Qi and Bernstein, 1999). These paradoxical results suggest that targeting to the cytoplasmic membrane is mediated by several, possibly overlapping interactions between preproteins, targeting factors, membrane receptors and translocase subunits. Physicochemical properties of the substrates may determine their preferred targeting route and dependency on the SecA protein for targeting and/or translocation. Several secretory proteins and, especially, cytoplasmic membrane proteins use the SRP-mediated pathway of targeting to the cytoplasmic membrane (Phillips and Silhavy, 1992; Luirink et al., 1994; Ulbrandt et al., 1997; Beck et al., 2000). Owing to their high hydrophobicity, cytoplasmic membrane proteins are more prone to aggregation and may therefore be more dependent on a co-translational targeting pathway (De Gier et al., 1997), although it should be remarked that a translational arrest of nascent polypeptide synthesis has not been demonstrated for the E. coli SRP. Direct binding of the ribosome to the SecYEG complex (Prinz et al., 2000) will also add to the proper targeting of membrane proteins. Nascent secretory proteins, on the other hand, interact with trigger factor, a cis-trans proline isomerase that competes for binding with SRP, and are routed into the SecA/SecB-mediated post-translational targeting pathway (Valent et al., 1995; 1997; Beck et al., 2000).
Structural characteristics and interactions of the translocase subunits
The SecA ATPase is an indispensable and unique bacterial protein that is functional as a homodimer of 102 kDa subunits (Cabelli et al., 1988; Akita et al., 1991; Driessen, 1993). In E. coli, SecA is the most abundant component of the translocase (Driessen, 1994) and has a balanced cellular distribution between soluble and membrane-bound states (Cabelli et al., 1991). SecA binds to the cytoplasmic membrane via low-affinity interactions with the lipid bilayer and with a high affinity for SecYEG. Once bound to SecYEG, SecA is primed for the high-affinity interaction with SecB/precursor complexes (Hartl et al., 1990). SecA also recognizes the signal sequence and the mature part of preproteins, and the interaction of preproteins with the SecYEG-bound SecA strongly stimulates its ATPase activity (Lill et al., 1990). The SecA binding sites for translocation ligands have been mapped to regions in the primary sequence (Fig. 2A). SecA contains two nucleotide binding sites (NBSs) composed of the Walker A (GXXXXGKT) and B (hXhhD; h is hydrophobic) motifs found in many ATPases (Walker et al., 1982). The amino-terminal NBS-I has a much higher affinity (Kd,ADP = 0.1–0.3 μM) than the more distal NBS-II (Kd,ADP = 300–500 μM) (Mitchell and Oliver, 1993; Den Blaauwen et al., 1996). The role of NBS-II in translocation is unclear. In vitro translocation can be driven by ATP concentrations that are far below the Kd of NBS-II (Jarosik and Oliver, 1991). On the other hand, mutations in NBS-II lower the ATPase activity of NBS-I and uncouple the translocation ATPase (Mitchell and Oliver, 1993; Economou et al., 1995). ATP-derived cross-linkers revealed that NBS-II is part of the interface between the SecA subunits in the homodimer (Van der Wolk et al., 1997a). In between the NBSs lies the region of SecA that interacts with preproteins (Kimura et al., 1991). The SecA carboxy-terminal third is involved in its dimerization and contains a region that regulates the SecA ATPase activity (Hirano et al., 1996; Karamanou et al., 1999) and the interaction with SecY (Snyders et al., 1997). At its extreme carboxy-terminus, SecA contains a stretch of 20 amino acids that binds SecB (Fekkes et al., 1997). Its size and complexity allow SecA to function both as the receptor and as an ATP-driven molecular motor of the translocase. It undergoes large conformational changes during its reaction cycle that can be measured as different thermal or proteolytical stabilities depending on its nucleotide-bound state (Den Blaauwen et al., 1996). Based on the formation of proteolytically stable fragments of SecA that are sensitive to disruption of the cytoplasmic membrane, it has been postulated that the protein inserts into the cytoplasmic membrane at SecYEG as part of the translocation reaction ( and Wickner, 1994; Economou et al., 1995; Eichler and Wickner, 1997). However, as there are data and theoretical arguments that contradict this hypothesis, membrane insertion is not the exclusive explanation for the proteolytic stabilization of SecA during translocation (see below in A mechanistic model for translocation).
SecY is the largest subunit of the SecYEG heterotrimer, with a molecular mass of 48 kDa and 10 hydrophobic segments that span the cytoplasmic membrane (Akiyama and Ito, 1987; Fig. 2B). SecE (14 kDa) has three TMSs, of which only the most carboxy-terminal TMS is required for protein translocation (Schatz et al., 1991; Fig. 2B). SecY is stabilized by an equimolar amount of SecE, and excess SecY is rapidly degraded by the membrane-bound protease FtsH (Taura et al., 1993; Kihara et al., 1995). SecY and SecE form a stable complex that does not dissociate in the cytoplasmic membrane or upon solubilization (Brundage et al., 1990; Joly et al., 1994). Mutational analyses of SecY and SecE have revealed several regions of interaction between the two proteins. The fourth cytoplasmic domain (C4) of SecY (Baba et al., 1994) and the highly conserved C2 of SecE (Murphy and Beckwith, 1994; Pohlschröder et al., 1996) are sensitive to mutations that affect a stable SecY/E interaction. Allele-specific synthetic lethality between prlA and prlG mutations suggests interactions of SecE TMS3 with SecY TMS7 and 10, and between SecY periplasmic domain 1 (P1) and SecE P2 (Flower et al., 1995). The latter contact has been verified by cysteine-scanning mutagenesis and in vivo cross-linking (Harris and Silhavy, 1999). Specific contacts have also been mapped between SecE TMS3 and the interface of TMS2 (Kaufmann et al., 1999) and TMS7 of SecY (A. K. Veenendaal and A. J. M. Driessen, unpublished results). SecG (11.5 kDa) traverses the membrane twice and possesses a cytosolic hydrophobic segment, making the total protein highly hydrophobic (Nishiyama et al., 1996; Fig. 2B). In vivo studies have shown that SecG is exchanged between different SecYE complexes in the cytoplasmic membrane (Joly et al., 1994), and that it associates with SecY (Homma et al., 1997). Although signal sequence suppressor (prl) mutations have been identified in all three subunits of the SecYEG complex, direct interaction with a translocating preprotein has only been demonstrated for SecY (Joly and Wickner, 1993).
SecA binds to the SecYEG complex via a direct interaction with SecY. This interaction can be cross-linked in vivo (Manting et al., 1997) and comprises at least the first 107 amino acids of SecY (Snyders et al., 1997). SecY may have multiple interaction sites with SecA, as mutations in carboxy-terminal domains of SecY (C5 and C6) disturb the SecA binding and reaction cycle (Matsumoto et al., 1997; Taura et al., 1997), and a single amino acid substitution in TMS7 improves SecA binding and translocation (Manting et al., 1999). Improved binding of SecA to SecY is responsible for the reduced signal sequence specificity of the prlA4 mutant (Van der Wolk et al., 1998). PrlA4 harbours mutations in TMS7 and TMS10, of which the TMS10 mutation is responsible for the Prl phenotype and causes the tighter binding of SecA (J. de Keyser and A. J. M. Driessen, unpublished results), whereas the TMS7 mutation elevates the stability of the SecYEG heterotrimer (Duong and Wickner, 1999). Finally, the proteolytic stabilization (membrane insertion) of SecA that occurs during translocation coincides with a reversal of the SecG topology (Nishiyama et al., 1996; Suzuki et al., 1998) and an increased cross-linking between cysteine residues that were introduced in TMS3 of SecE (Kaufmann et al., 1999), suggestive of interactions of SecE and SecG with SecA, and between SecE subunits of neighbouring SecYEG heterotrimers.
Translocation is a stepwise process
In vitro protein translocation is a stepwise process that results in the formation of translocation intermediates. These intermediates are coupled to the reaction cycle of SecA, as they are more abundant at low ATP concentrations and because their stability depends on the presence of both SecA and ATP (Schiebel et al., 1991). The major translocation intermediates of the precursor of outer membrane protein A (proOmpA) are determined by short hydrophobic stretches in its primary sequence (Sato et al., 1997a). The slowed translocation of hydrophobic sequences may be an important feature of the translocase with respect to membrane protein integration. Hydrophobic sequences in cytoplasmic membrane proteins serve as ‘stop-transfer’ sequences, which halt the translocation reaction to allow their integration into the lipid bilayer (Von Heijne, 1994, 1997). The hydrophobic segments in proOmpA indeed serve as stop-transfer signals when placed at adjacent positions (Sato et al., 1997b). Importantly, membrane integration of segments with a ‘threshold’ hydrophobicity depends on the kinetics of the translocation reaction (Duong and Wickner, 1998). The stop-transfer event thus seems to be a well-orchestrated process, determined by the translocating polypeptide as well as the translocase. Independent of the proOmpA primary sequence, its translocation proceeds in steps of approximately 5 kDa (Van der Wolk et al., 1998). In contrast, disulphide-bonded loops in proOmpA with a maximal size of only 20 amino acids can be handled by the translocase (Uchida et al., 1995). The step size in translocation actually represents two distinct catalytic events in the SecA reaction cycle. These two events can be distinguished by the addition of the SecA inhibitor sodium azide, which traps the reaction cycle in an intermediate stage and reduces the step size of translocation to 2–2.5 kDa (Van der Wolk et al., 1997b).
Translocation is stimulated by the presence of a pmf both with IMVs and with the purified and reconstituted translocase (Geller et al., 1986; De Vrije et al., 1987; Yamane et al., 1987; Brundage et al., 1990). The pmf affects several aspects of the translocation reaction via mechanisms that are mostly unclear. The binding of signal sequences to the cytoplasmic membrane and their subsequent insertion is altered by the presence of a pmf, and this optimizes the initiation of translocation (De Vrije et al., 1990; Nouwen et al., 1996a; Van Dalen et al., 1999). The pore size of the translocation channel may be affected, as the translocation of a disulphide-bonded loop in proOmpA is dependent on the pmf (Tani et al., 1990). Upon the removal of SecA from translocation intermediates, they can be further translocated by the pmf (Schiebel et al., 1991). In the absence of SecA, ATP and the pmf, reverse translocation occurs (hysteresis movement of the polypeptide chain), and one of the contributions of the pmf to the translocation reaction is to provide unidirectionality (Driessen, 1992). Finally, the pmf optimizes the SecA reaction cycle. The rate-limiting release of ADP from SecA is promoted by the pmf (Shiozuka et al., 1990), and conformational transitions (membrane deinsertion) of the SecA molecule during translocation are accelerated (Nishiyama et al., 1999). The influence of the pmf on the SecA reaction cycle is a dominant factor in the translocation reaction, as excess SecA renders the translocation reaction pmf independent (Yamada et al., 1989). In prlA mutants, translocation is pmf independent because of an altered gating of the translocase (Nouwen et al., 1996b) and/or the increased affinity of SecA for the SecYEG complex (van der Wolk et al., 1998; Manting et al., 1999). The latter will lead to a more permanent SecA binding to the translocation channel, optimize SecA- and ATP-driven translocation and thus prevents the translocation of preprotein segments by the pmf.
The protein-conducting channel
SecYEG provides the structural basis for the proteinaceous channels across the cytoplasmic membrane, which can be observed as ion-conducting pores upon the addition of translocation ligands (Simon and Blobel, 1992; Kawasaki et al., 1993). The inside of the protein-conducting channel is not in contact with the surrounding lipid bilayer, as translocating preproteins are shielded from phospholipids and are mainly in contact with SecA and SecY (Joly et al., 1993). Like the homologous Sec61p complex, SecYEG is likely to oligomerize to constitute the protein-conducting channel. Electron microscopic analysis of purified Sec61p revealed quasi-pentagonal structures with a 2 nm central pore or indentation. These structures represent oligomers of two to four Sec61p complexes (Hanein et al., 1996). The central cavity in the ribosome-bound Sec61p aligns with the putative protein-conducting channel of the large ribosomal subunit, suggesting an alignment of conduits in co-translational translocation (Beckman et al., 1997). The SecYE complex from the Gram-positive bacterium Bacillus subtilis forms structures that resemble Sec61p (Meyer et al., 1999). Although B. subtilis SecG stimulates translocation in a manner similar to E. coli SecG (Van Wely et al., 1999), it seems dispensable for the formation of the translocation channel. This is consistent with data in E. coli that indicate that SecYE is the minimal constituent of the integral membrane translocase domain (Duong et al., 1997). The formation of the purified Sec61p oligomer is dependent on the addition of ribosomes or co-reconstitution with the yeast Sec62–63 complex (Hanein et al., 1996), whereas the B. subtilis SecYE structure is visible without any preincubation with translocation ligands (Meyer et al., 1999). Thus, B. subtilis SecYE is either more stable during purification and reconstitution, or oligomerization is regulated differently from Sec61p.
It is questionable whether the structures observed with either Sec61p or SecYE in the absence of translocation conditions represent the active protein-conducting channel. The Sec61p structure cannot account for the 4–6 nm pore through which nascent chains pass the ER membrane (Hamman et al., 1997). These pores are formed upon incubation with RNCs and are observed in addition to smaller, 1.5–2.5 nm pores that are also present without such preincubation (Hamman et al., 1998). The recruitment of additional subunits may underlie the increase in channel size of the active translocon. Electron microscopic analysis of the purified E. coli SecYEG heterotrimer revealed the presence of two channel-like structures, of which one resembles Sec61p or B. subtilis SecYE and corresponds to a dimer of SecYEG. Another, larger structure with an increased central indentation is composed of four SecYEG complexes (Manting et al., 2000). The latter was observed only upon membrane insertion of SecA or trapping of a precursor in the translocase (Manting et al., 2000). The protein-conducting channel is thus a highly plastic structure constituted by oligomers of Sec61p or SecYEG complexes, with a tetrameric assembly as the active conformation. Based on the SecA-dependent assembly of the tetrameric SecYEG from smaller, dimeric complexes as well as the observation of SecA–SecYEG dimer subcomplexes in statistical mass analysis, we propose a two-step model for the formation of the translocase. First, two dimeric SecYEG units are brought together by their affinity for the membrane-bound SecA dimer. Next, they are assembled into one active, tetrameric SecYEG channel upon SecA membrane insertion, which can be enforced by AMP–PNP binding but normally coincides with the ATP-dependent initiation of translocation (Fig. 3).
The prlA and prlG mutations result in a loosened association among the SecYEG subunits (Duong and Wickner, 1999). This may increase the pore size or flexibility of the translocation channel, resulting in altered gating, pmf independency or an improved SecA membrane insertion. The last is associated with the formation of the active, tetrameric conformation of SecYEG. Owing to the high affinity of prlA mutants for SecA, the process of oligomerization may also be affected favourably by the optimized SecA-mediated assembly of smaller SecYEG complexes before the formation of active translocation channels. Cause and effect are not easily distinguished in this manner as, alternatively, increased stability of the prlA mutant SecYEG tetramer may improve SecA binding and membrane insertion. Remarkably, a recent study demonstrated that prlD mutations affect the SecA molecule in a manner that resembles the effects of prlA/G mutations on the SecYEG complex. SecA molecules that contain prlD mutations have a higher affinity for the translocation channel and are in a more relaxed conformation than wild-type SecA, resulting in accelerated exchange of ADP for ATP at low temperature and a higher membrane ATPase activity (Schmidt et al., 2000).
SecA appears to be an integral subunit of translocase, as cysteines in the SecA carboxy-terminus can be labelled from the periplasmic side of the membrane (Kim et al., 1994; Van der Does et al., 1996; Eichler and Wickner, 1998). This periplasmic exposure of the SecA molecule is not dependent on translocation and, remarkably, the labelled cysteines are part of the SecB binding site (Fekkes et al., 1997; 1998). Upon the introduction of additional cysteines in SecA, multiple regions of the molecule have been shown to be periplasmically accessible for thiol-reactive probes (Ramamurthy and Oliver, 1997). The topology studies on SecA must be interpreted with caution, as SecA is bound to the transmembrane structure formed by the SecYEG complex that encompasses a pore through which labelling may occur.
A mechanistic model for translocation
SecA is a highly dynamic molecule that undergoes extensive nucleotide-modulated conformational changes in solution (den Blaauwen et al., 1996). When bound to the SecYEG complex, these conformational changes may be responsible for the transmembrane movement of a translocating polypeptide. Economou et al. (1994) used 125I-labelled SecA in the in vitro protein translocation reaction and showed that a radiolabelled proteolytic 30 kDa fragment of the SecA arises during translocation conditions. The fragment is susceptible to proteolysis upon the addition of detergent or after repeated cycles of freeze–thawing and can be chased by excess cold SecA. From this, it was concluded that the 30 kDa fragment is protected from the proteolysis by the cytoplasmic membrane and represents cycles of SecA membrane insertion and deinsertion at SecYEG during translocation. The 30 kDa fragment is a carboxy-terminal domain of SecA (Price et al., 1996) and is shielded from phospholipids, presumably by the SecYEG complex (Eichler et al., 1997; Van Voorst et al., 1998). Several findings support the correlation between the 30 kDa fragment and translocation. Formation of the 30 kDa fragment requires a functional SecA NBS-I (Economou et al., 1995; Rajapandi and Oliver, 1996) and a productive interaction between SecA and SecY (Matsumoto et al., 1997). The SecDF–yajC complex modulates the translocase in a manner that stabilizes both the 30 kDa fragment and translocation intermediates (Economou et al., 1995; Duong and Wickner, 1997b). The membrane-inserted state of SecA represented by the 30 kDa fragment is trapped by non-hydrolysable ATP analogues (Economou and Wickler, 1994; Economou et al., 1995) or sodium azide (Van der Wolk et al., 1997b). The pmf reduces 30 kDa and optimizes the transition of SecA from its inserted to its deinserted state (Nishiyama et al., 1999). The carboxy-terminal region of SecA that comprises the 30 kDa fragment is the only part of the molecule that is iodinated. By using other detection techniques, various other proteolytic fragments of soluble and membrane-bound SecA can be visualized, of which some are nucleotide modulated (Chen et al., 1996; Den Blaauwen et al., 1996; Price et al., 1996). An amino-terminal 35S-labelled 65 kDa proteolytic fragment of SecA correlates with translocation in a manner similar to the carboxy-terminal 30 kDa fragment (Eichler and Wickner, 1997) and, by these criteria, it has been suggested that the entire SecA protein may insert into the membrane at a certain stage of the translocation reaction.
Although the proteolytic protection of the 30 and 65 kDa fragments correlates with translocation, experimental and hypothetical arguments contradict their membrane insertion. First, detergent-solubilized SecYEG supports the formation of the 30 kDa SecA fragment upon the addition of the non-hydrolysable ATP analogue AMP-PNP. In this case, SecYEG is degraded to near completion, showing that the 30 kDa fragment is intrinsically stable to proteolysis and does not require insertion into the SecYEG channel (Van der Does et al., 1998). Secondly, the SecA dimer is an elongated molecule with a maximal width of 8 nm and a length of 15 nm (Shilton et al., 1998). The size of the tetrameric SecYEG assembly or the thickness of the lipid bilayer is far from sufficient to accommodate the entire SecA molecule or large domains of it. Conformational changes in SecA that lead to proteinase protection may therefore not represent membrane insertion, and other (e.g. hinge- or diaphragm-like) mechanisms of SecA-mediated translocation should not be excluded (Den Blaauwen et al., 1996; Karamanou et al., 1999).
A two-step reaction model that includes SecA membrane insertion is shown in Fig. 4. The model is based on translocation of 2–2.5 kDa of the preprotein upon SecA binding and a similar step upon ATP binding (Tani et al., 1989; 1990; Schiebel et al., 1991). Because it is unlikely that SecA will be present in the cell in a nucleotide-unbound state, we assume that SecA binding to the precursor takes place with ADP-bound SecA. The ADP is than replaced by ATP (at NBS-I), and this will result in a further translocation of the precursor protein. The total step size of a complete reaction cycle is 5 kDa (Van der Wolk et al., 1997b). The exchange of ADP for ATP after ATP hydrolysis and SecA deinsertion (Shiozuka et al., 1990; Economou et al., 1995) is rate limiting and determines the size of translocation intermediates. Intermediates associated with SecA membrane insertion can only be visualized when trapped with sodium azide (Van der Wolk et al., 1997b).
How the translocase works mechanistically is still largely unclear but, as a total enzyme, it is highly dynamic. SecG undergoes inversions of its membrane topology during translocation, and this optimizes the SecA reaction cycle (Nishiyama et al., 1996; Suzuki et al., 1998). SecE molecules are rearranged within the SecYEG oligomer, such that they can be cross-linked with high efficiency via unique cysteines positioned in one helical face of TMS3 (Kaufmann et al., 1999; A. Veenendaal and A. J. M. Driessen, unpublished results). The co-ordinated interaction and dynamics of all translocase subunits make it the unique molecular machinery that allows protein translocation across the cytoplasmic membrane of bacteria in a fashion that is still far from understood. More insight into the mechanism of translocation will be obtained from the thermodynamics of the translocation reaction, detailed structural analysis of SecA, SecYEG and translocase, as well as the surface topography of the SecA molecule during translocation. Other unresolved questions concern the role of the SecDF–YajC complex, YidC and the mechanism of membrane protein integration and assembly.
We thank past and present members of the team who contributed to this review. Nico Nouwen and the reviewers of the manuscript are thanked for critical reading and helpful suggestions. This work was supported by the Earth and Life Sciences Foundation (ALW) and the Netherlands Foundation for Scientific Research (NWO).