Escherichia coli preprotein translocase.

Polar molecules cross lipid-based membranes through proteinaceous transport systems. Sugars, amino acids, and small ions are recognized by transporters of great specificity and then transferred along (still ill defined) paths within the transporter and released on the opposite membrane side. Subcellular compartmentation entails the selective transport of newly made preproteins across membranes. Since preproteins are of varied sequence and can be as large as their transporter, it was likely that their transporter would have unique mechanistic features. This view has been amply confirmed for Escherichia coli preprotein translocase, a large and complex transport system with an unusual catalytic cycle. E. coli preprotein translocase has been discovered through the convergence of genetic (1) and biochemical (2) approaches. The genes themselves were identified both through the isolation of “prl” (protein localization) mutations (3), altered in their capacity to recognize mutant presecretory proteins, and “sec” (secretion) temperature-sensitive mutants (4), blocked in protein translocation at the nonpermissive temperature. The “sec” and “prl” mutants were found to be in the same genes, and the encoded proteins were found to be peripheral and integral proteins of the inner membrane. Genetic analysis established that the Sec proteins were functionally interacting (5). Physiological studies established that translocation requires metabolic energy (6–11) and showed that translocation was not coupled to translation (11, 12). The latter finding suggested that post-translational translocation might be reconstituted from purified components without the daunting task of reconstituting protein synthesis itself. In 1984, two groups reported the marriage of in vitro synthesis of presecretory proteins with targeting to purified inner membrane vesicles, resulting in translocation of the preprotein into the vesicle lumen (13, 14). Translocation in vitro was separable from protein synthesis and needed both ATP and DmH1 (10). Meanwhile, preproteins were overproduced and purified (15–17), allowing in vitro translocation to be freed from the complex protein synthesis reaction. Pure preproteins, as well as newly made preproteins in the cell, depend on chaperone proteins to maintain their translocation-competent conformation (18, 19). The stage was set to tackle the inner membrane. SecA was the first translocation component to be purified from the inner membrane in functional form (20, 21). Salt or urea treatment rendered inner membrane vesicles dependent on cytosol. The relevant soluble protein was large (;200 kDa), suggesting a SecA dimer, and this suggestion was confirmed (21) by means of an overexpressing strain. Following the early observation that translocation was blocked by membrane exposure to N3-ATP (azidoadenosine triphosphate) plus light (22), SecA was found to restore full function to these membranes, suggesting that it was the relevant ATPase (23). Not only was pure SecA then found to be a low level ATPase, but this ATPase activity was strikingly stimulated by associations with all its translocation ligands; inner membrane proteins, inner membrane lipids, and authentic preprotein (23–25). SecA, which uses both acidic lipids and a complex of Sec proteins to bind to the inner membrane, is activated as an ATPase by its lipid associations and stabilized in this activated state by association with the leader and mature domain of preproteins (24, 26). The fully activated state of SecA ATP hydrolysis is defined as the “translocation ATPase,” and it provided a rapid and quantitative assay for subsequent enzymology of the integral membrane components. Using this functional assay, a trimeric complex was isolated from detergent extracts of inner membrane (27). Fractions during the purification were assayed by forming proteoliposomes through detergent removal and measuring their capacity to support, with pro-OmpA (the precursor of outer membrane protein A), the translocation ATPase activity of SecA. Two members of this complex were the products of the secY and secE genes; the third member was not encoded by a known sec gene, but “reverse genetics” has since confirmed its role in secretion, and it was termed SecG (28–30). Though these proteins can be isolated separately and functionally recombined during proteoliposome formation (31), they are physiologically associated as a trimer, both in vivo (32) and in vitro (33). The complex of SecA bound stoichiometrically to the membrane-embedded SecYEG (34, 35) constitutes the core of “preprotein translocase.” The pure enzyme supports ATP and DmH1-dependent preprotein translocation at rates and to extents that are comparable with those seen with the starting membrane vesicles (36). Recent studies have shown that SecYEG is part of a yet larger complex that also contains SecD, SecF, yajC, and (potentially) additional subunits encoded by unknown genes. The study of this larger, holoenzyme form of translocase is just beginning.

Polar molecules cross lipid-based membranes through proteinaceous transport systems. Sugars, amino acids, and small ions are recognized by transporters of great specificity and then transferred along (still ill defined) paths within the transporter and released on the opposite membrane side. Subcellular compartmentation entails the selective transport of newly made preproteins across membranes. Since preproteins are of varied sequence and can be as large as their transporter, it was likely that their transporter would have unique mechanistic features. This view has been amply confirmed for Escherichia coli preprotein translocase, a large and complex transport system with an unusual catalytic cycle.
E. coli preprotein translocase has been discovered through the convergence of genetic (1) and biochemical (2) approaches. The genes themselves were identified both through the isolation of "prl" (protein localization) mutations (3), altered in their capacity to recognize mutant presecretory proteins, and "sec" (secretion) temperature-sensitive mutants (4), blocked in protein translocation at the nonpermissive temperature. The "sec" and "prl" mutants were found to be in the same genes, and the encoded proteins were found to be peripheral and integral proteins of the inner membrane. Genetic analysis established that the Sec proteins were functionally interacting (5).
Physiological studies established that translocation requires metabolic energy (6 -11) and showed that translocation was not coupled to translation (11,12). The latter finding suggested that post-translational translocation might be reconstituted from purified components without the daunting task of reconstituting protein synthesis itself. In 1984, two groups reported the marriage of in vitro synthesis of presecretory proteins with targeting to purified inner membrane vesicles, resulting in translocation of the preprotein into the vesicle lumen (13,14). Translocation in vitro was separable from protein synthesis and needed both ATP and ⌬ H ϩ (10). Meanwhile, preproteins were overproduced and purified (15)(16)(17), allowing in vitro translocation to be freed from the complex protein synthesis reaction. Pure preproteins, as well as newly made preproteins in the cell, depend on chaperone proteins to maintain their translocation-competent conformation (18,19). The stage was set to tackle the inner membrane.
SecA was the first translocation component to be purified from the inner membrane in functional form (20,21). Salt or urea treatment rendered inner membrane vesicles dependent on cytosol. The relevant soluble protein was large (ϳ200 kDa), suggesting a SecA dimer, and this suggestion was confirmed (21) by means of an overexpressing strain. Following the early observation that translocation was blocked by membrane exposure to N 3 -ATP (azidoadenosine triphosphate) plus light (22), SecA was found to restore full function to these membranes, suggesting that it was the relevant ATPase (23). Not only was pure SecA then found to be a low level ATPase, but this ATPase activity was strikingly stimulated by associations with all its translocation ligands; inner membrane proteins, inner membrane lipids, and authentic preprotein (23)(24)(25). SecA, which uses both acidic lipids and a complex of Sec proteins to bind to the inner membrane, is activated as an ATPase by its lipid associations and stabilized in this activated state by association with the leader and mature domain of preproteins (24,26). The fully activated state of SecA ATP hydrolysis is defined as the "translocation ATPase," and it provided a rapid and quantitative assay for subsequent enzymology of the integral membrane components.
Using this functional assay, a trimeric complex was isolated from detergent extracts of inner membrane (27). Fractions during the purification were assayed by forming proteoliposomes through detergent removal and measuring their capacity to support, with pro-OmpA (the precursor of outer membrane protein A), the translocation ATPase activity of SecA. Two members of this complex were the products of the secY and secE genes; the third member was not encoded by a known sec gene, but "reverse genetics" has since confirmed its role in secretion, and it was termed SecG (28 -30). Though these proteins can be isolated separately and functionally recombined during proteoliposome formation (31), they are physiologically associated as a trimer, both in vivo (32) and in vitro (33). The complex of SecA bound stoichiometrically to the membrane-embedded SecYEG (34,35) constitutes the core of "preprotein translocase." The pure enzyme supports ATP and ⌬ H ϩ-dependent preprotein translocation at rates and to extents that are comparable with those seen with the starting membrane vesicles (36). Recent studies have shown that SecYEG is part of a yet larger complex that also contains SecD, SecF, yajC, and (potentially) additional subunits encoded by unknown genes. 1 The study of this larger, holoenzyme form of translocase is just beginning.

Mechanism(s) of Preprotein Translocase
The availability of the biochemical tools (pure chaperones, translocase, and a variety of pure wild-type or synthetic preproteins) and the genetic tools (prl and sec mutants, sequence information, and plasmids bearing each of the genes) has made E. coli translocation the premier system for mechanistic analysis. Our current working model (Fig. 1) has evolved through many stages of addition and correction (17,27,37,38) and will doubtless continue to evolve.
Emerging nascent preproteins associate with chaperones. SRP 2 (bacterial signal recognition particle) (39,40) and trigger factor (41) are ribosome-bound chaperones that compete for nascent chains (42), with SRP having a preference for leader sequences of preproteins. Trigger factor delivers at least some of the non-exported proteins to GroEL/S for catalysis of folding (43), while SRP interacts with FtsY, activating their GTP cycles (44,45). The dnaJ/K/GrpE proteins can also assemble on nascent chains (46) and may precede GroEL/S in their action. Chaperones may be used in distinct fashion for integral membrane proteins, with their very different proportions of charged and apolar regions. The specificity, breadth of substrate, and mode of action of these chaperones on presecretory and membrane proteins are still undefined. Many exported proteins associate with SecB, a tetrameric, export-dedicated chaperone. Elegant studies (47)(48)(49) have established that SecB interacts with the mature domains of preproteins. One of the several identical subunits of SecB initially binds to basic residues on a preprotein, triggering a SecB conformational change that allows it to bind an adjacent short apolar preprotein domain (19). Other subunits of the same SecB molecule may then recognize additional basic and apolar regions of the preprotein. Only proteins that are not tightly folded can be captured by SecB. The SecB association stabilizes the assembly-competent conformation by preventing the final folding events, prevents aggregation by shielding the apolar regions, and targets the preprotein for export by virtue of the direct affinity of SecB for the SecA subunit of translocase (34).
SecA is a peripheral subunit of translocase with a spatially dynamic role. It is normally present in stoichiometric excess over the integral SecYEGDF domain (50,51). The cytosolic form of SecA functions as a translational repressor of its own synthesis (52) and may also function as a cytosolic chaperone (53). SecA can bind to (lipidic) low affinity sites (24) but binds functionally and with high affinity at sites of SecYEGDF (35) plus acidic phospholipid (Stage 1 of Fig. 1). This binding activates SecA, both as a receptor for recognition of the leader, mature, and chaperone domains of the SecB-preprotein complex (Stage 2 of Fig. 1) and as an enzyme for the binding and hydrolysis of ATP.
SecA is a dimeric protein (54,55), and each subunit has two nucleotide binding domains, NBD1 and NBD2 (56). Despite the absence of a classic, membrane-spanning apolar sequence in SecA, ATP binding at NBD1 (Stage 3 of Fig. 1) drives the insertion of a 30-kDa domain of SecA into, and partially across, the membrane (Stage 4 of Fig. 1) (38,57,58). 1 This SecA insertion is accompanied by the insertion of a "loop" of about 25 aminoacyl residues of the preprotein across the membrane (37,59,60), allowing cleavage by the periplasmic, catalytic domain of leader peptidase (not illustrated in Fig. 1). The insertion of SecA is accompanied by the inversion of topology of the SecG subunit (61), symbolized by the inverted G in Fig. 1, and inserted SecA is stabilized by its interactions with the SecDF subunits (58). 1 Hydrolysis of this first ATP and the further binding and hydrolysis of a second ATP are necessary (38) for SecA deinsertion (Stage 5 of Fig. 1), leaving a loop of the preprotein spanning the membrane. The second nucleotide binding site, NBD2, though not needed for SecA insertion and deinsertion (58), presumably couples the SecA cycle to net translocation of the preprotein (56). When SecA is in the deinserted state, it can exchange freely with cytosolic SecA (38). Substantial, rapid translocation is driven by ⌬ H ϩ while deinserted SecA is not bound to the transmembrane translocation intermediate at SecYEGDF (37). SecA can rebind to SecY-EGDF (Stage 1Ј), associate with the translocation intermediate (Stage 2Ј), and bind ATP (Stage 3Ј) and insert (Stage 4Ј), carrying an additional loopful of preprotein across the membrane. The unusual, "sewing machine" character of the translocase catalytic cycle has received strong, independent support from three independent lines of study. Uchida et al. (59) showed that translocation of pro-OmpA occurs in steps of approximately 25 aminoacyl residues rather than one or a few residues at a time. This model also agrees with the studies of Joly et al. (32), which showed that the preprotein chain is adjacent to SecA whenever it is also adjacent to SecY. Finally, the inversion of the topology of SecG was shown (61) to depend on ATP, SecA, and preprotein, just as does the SecA insertion event itself, and the restoration of the SecG topology requires the same factors that promote SecA deinsertion. This important study further links the functions and structures of the peripheral and integral subunits of translocase.

The Model and Mechanistic Questions
How does this current model address the mechanistic questions? Preprotein recognition is a stepwise cascade. Chaperones recognize the leader sequence (SRP) and the mature domain (SecB) and deliver the preprotein to translocase. There, SecA recognizes the leader and mature domains (24); there are also genetic data indicating a SecY-mediated "proofreading" event (62,63). Finally, the leader region is recognized by leader peptidase (64). The recognition of the mature domain as well as the leader is underscored by Bassford's studies, showing that SecB can be titrated by an excess of mature maltose binding protein (48) and by work from the Beckwith (65) and Silhavy (63)  and ⌬ H ϩ are used in different parts of the catalytic cycle (37), also accounts for how one transport system can use two such energy sources. The initial translocation of an amino-terminal loop, with linear N to C continuance of its progress, also explains how folded proteins can be translocated. This mechanism can drive protein unfolding (66) and reduces the problem to one of translocating linear polypeptide segments. The variation in aminoacyl side chain structure is presumably accommodated in a SecA binding site as SecA weaves back and forth across the membrane like a needle in a sewing machine.
There are, of course, new structural questions raised by this model. Does the SecA expose a large portion of its 30-kDa domain surface to the lipid phase during its transit, or is it largely buried within the helices of SecYEGDF? What are the nearest neighbors of the preprotein during transit in addition to SecA and SecY? Since even nonionizable preprotein can be driven across the membrane by ⌬ H ϩ (67), what is the mechanism of action of this energy source?

Conserved and Divergent Themes of Translocation
Other translocation reactions, reviewed in the other minireviews in this series, share common themes with bacteria. Bacterial and endoplasmic reticulum leader (signal) sequences are virtually interchangeable. The SecYEG and Sec61 trimeric complexes show striking similarities, and in each case, this complex is the "core" of a larger, more complex holoenzyme structure (68). Unlike bacteria, mammalian endoplasmic reticulum is believed to function almost exclusively in a translationcoupled mode of translocation. As reviewed by Schekman (second minireview in this series), yeast is somewhere in between, with some proteins following a post-translational path and others apparently coupled to a cotranslational transit. Endoplasmic reticulum and mitochondrial matrix translocation also couple ATP energy to movement but via an hsp70 (DnaK homolog) on the trans side of the membrane. Though this entails a conformational change that may be as profound as that of SecA, there is no evidence for an intramolecular subunit translocation event in any other protein transport system. Mitochondria and chloroplasts also use both ATP and the membrane electrochemical potential.

What Comes Next?
Powerful experimental systems are never "solved" but provide the basis for asking ever more fundamental questions. Analysis of translocation will be profoundly aided by information on the physical structure of SecA and of SecYEGDF and any remaining subunits. The latter must be isolated, cloned, and studied. We have little information as to how the potential drives translocation, how membrane-spanning proteins escape laterally from translocase into the lipid, or the dynamics of subunit exchange among SecDFyajC, SecYEG, and the complex of SecYEGDFyajC. Questions of orderly protein release into the periplasm and further transit into the outer membrane are also just beginning to be addressed. Preprotein translocase will continue to command attention in the future, as an elegant transport reaction, as a vital tool in the export of cloned proteins in biotechnology, and as a potential drug target.