Mechanistic Aspects of Folded Protein Transport by the Twin Arginine Translocase (Tat)*

The twin arginine translocase (Tat) transports folded proteins of widely varying size across ionically tight membranes with only 2–3 components of machinery and the proton motive force. Tat operates by a cycle in which the receptor complex combines with the pore-forming component to assemble a new translocase for each substrate. Recent data on component and substrate organization in the receptor complex and on the structure of the pore complex inform models for translocase assembly and translocation. A translocation mechanism involving local transient bilayer rupture is discussed.

The twin arginine translocase (Tat) transports folded proteins of widely varying size across ionically tight membranes with only 2-3 components of machinery and the proton motive force. Tat operates by a cycle in which the receptor complex combines with the pore-forming component to assemble a new translocase for each substrate. Recent data on component and substrate organization in the receptor complex and on the structure of the pore complex inform models for translocase assembly and translocation. A translocation mechanism involving local transient bilayer rupture is discussed.
Cells and membranes contain multiple protein translocases, i.e. the enzymes that catalyze protein translocation across or into membranes. This allows for specific targeting to distinct cellular locations and, through varied transport mechanisms, solutions to a range of translocation problems. For example, the prokaryotic and endoplasmic reticulum Sec system and several other translocases transport proteins in a mostly unfolded conformation. This permits a single mechanistic strategy for those substrates that can be unfolded for translocation and folded following transport. By contrast, the twin arginine translocase (Tat) 2 system transports proteins that are in a folded conformation during translocation. The Tat system is present in most bacteria, in some archaea, in chloroplasts of plants and algae, and in some mitochondria (1)(2)(3)(4)(5). The prevalence of Tat in prokaryotes and prokaryote-derived organelles argues that Tat is an ancient translocation system. The fact that Tat operates with as few as two membrane components of machinery (TatA and TatC) (6) and the proton motive force (PMF) implies a simple mechanism. Although it may be mechanistically simple, Tat behaves in mystifying ways and performs astonishing feats. It exists as oligomeric structures, with a multivalent receptor complex composed of TatC and TatB multimers and a "pore" complex of TatA oligomers that are thought to form the protein-conducting element. Tat can transport folded proteins from ϳ20 Å (7) to ϳ70 Å (8). It can transport heterodimers, where one subunit has the signal peptide and the other hitchhikes the ride (9), and cross-linked tetramers, where each subunit is bound via its own signal peptide (10). Tat will even trans-port engineered unstructured polypeptides (11,12). Thus, Tat has solved the daunting mechanistic challenge of transporting both large and small protein structures without opening large holes that would dissipate the PMF.
The Tat substrate repertoire depends on the organism and ranges from 1 to ϳ150 substrates (reviewed in Ref. 2). Tat plays critical roles in respiratory and photosynthetic energy production, animal and plant pathogenesis, symbiosis, etc. (reviewed in Refs. 2, 8, and 13-15). Among reasons that proteins require transport in a folded conformation include: the absence of folding and cofactor insertion machinery on the trans side of the membrane, the need to control metal cofactor specificity, and the fact that some proteins rapidly fold upon synthesis or import into the organelle (2,5).
Unraveling the mechanisms of Tat operation has been challenging because of the ephemeral nature of the translocase (16). However, mechanistic knowledge has advanced due to studies with the thylakoid membrane of plant chloroplasts and the cytoplasmic membrane of Escherichia coli. A combination of staged in vitro assays, protein cross-linking, molecular structural analyses, and fluorescence imaging in living cells has increased knowledge sufficiently that informed models are possible. Both systems employ three components, a TatC protein and two members of the TatA family (TatA and TatB). For historical reasons, the nomenclature of the two systems has not been consistent; TatA and TatB in bacteria are called Tha4 and Hcf106 in thylakoids. However, considerable research has shown that Tha4 and Hcf106 function similarly to TatA and TatB. In this review, I will refer to these proteins simply as TatA and TatB and note the system from which results were obtained. In addition, the term cis will designate the compartment of substrate origin, and trans will designate the destination compartment. The scope of this review is limited to recent insights into Tat mechanisms. The reader is referred to several other excellent reviews for a broader view of Tat (1)(2)(3)(4)(5).

A Twin Arginine Signal Peptide Targets the Substrate
Tat substrates are targeted to Tat machinery by hydrophobic signal peptides that superficially resemble Sec signal peptides in tripartite organization: an amino proximal N domain, a hydrophobic H domain, and a polar C domain that contains the signal peptidase cleavage site (Fig. 1). The Tat recognition sequence at the N/H junction is the most distinctive feature of Tat signal peptides. It consists of two adjacent Arg residues within a broader consensus (Fig. 1). The RR motif is essential; mutation to KK eliminates interaction with the Tat machinery (17)(18)(19)(20)(21)(22). Other consensus residues are of lesser importance (22,23), with the exception of the Phe at RRϩ2, which has a critical positive effect on transport efficiency. Most thylakoid Tat signal peptides lack this Phe, but introducing a Phe into the thylakoid OE17 substrate (Fig. 1) increased the binding affinity 10-fold to ϳ1 nM K d (24,25). The resulting tOE17-20F ( Fig. 1) was used in many studies cited here. Some Tat signal peptides have an extended N domain of uncertain function. Deleting the extended N domain of several thylakoid Tat substrates improved in vitro transport efficiency (19,26). Deleting the extended N domain of the E. coli hydrogenase-1 small subunit similarly improved transport efficiency of fused reporter proteins but impaired maturation of hydrogenase 1 (27). This suggests that the extended N domain serves as a timing device, slowing translocation to allow holoenzyme folding and assembly.

Cycle of Substrate Binding, Translocase Assembly, and Translocation
Tat systems operate by a cycle in which the components ( Fig.  2A) assemble a translocase on demand (Fig. 2B). This was shown with thylakoid in vitro assays and with fluorescence imaging in living E. coli. In non-transporting thylakoids, TatB and TatC are present as a receptor complex in a 1:1 ratio; TatA is present as a separate pool (16,28). The receptor complex binds the substrate signal peptide, triggering PMF-dependent TatA assembly and oligomerization at the substrate-TatBC interface (16,29,30). The assembled complex is called the "translocase." After translocase assembly, the substrate is transported, the signal peptide is cleaved, and the TatA oligomer is disassembled in an uncertain order. The in vitro cycle is sluggish. A time course starting with substrate-bound TatBC and initiated with the PMF exhibited a 1-3-min transport lag (25) during which TatA could be increasingly cross-linked to TatBC (16). Substrate transport then occurred with first order kinetics and a transport time () of ϳ80 s (25). When all substrate was transported, the ability to cross-link TatA dropped to background. A similar in vitro rate was measured for SufI transport into E. coli vesicles (31).
Fluorescence imaging of TatA-YFP (or TatA-GFP) in E. coli cells showed a similar cycle. TatA-XFP exists as small oligomers in non-transporting cells (32)(33)(34). Substrate expression resulted in coalescence of TatA-XFP into large fluorescent foci termed "TatA assemblies" that co-localized with TatB (34) and presumably TatC. Dissipating the PMF with carbonyl cyanide m-chlorophenylhydrazone prevented the formation of TatA assemblies and resulted in disassembly when added during the protein transport phase. A transport-inactive TatA F39A-YFP mutant formed TatA assemblies but did not disassemble with carbonyl cyanide m-chlorophenylhydrazone (33). This suggests that, whereas TatA assembly requires the PMF, disassembly requires substrate transport (33), raising the possibility that the PMF may not be required for the transport step per se. The rate of assembling (20 -120 s) and disassembling TatA-YFP (5-15 s) was somewhat faster than the in vitro rate (33), but overall was still much slower than other protein transport systems (see e.g. Refs. 35 and 36).

TatC Scaffolds TatB, TatA, and Substrate to Form the Translocase
Recent NMR and crystal structures provide valuable insights into the individual Tat components, whereas several other approaches yield information on the associations of components and substrate. The accumulated findings support a model in which TatC is the initial specific receptor for the signal peptide RR motif and also the organizing scaffold for the other components. TatB and then TatA, while bound to TatC, sequentially contact the substrate signal peptide and folded domain. TatC spans the membrane six times with N and C termini exposed to the cis compartment ( Fig. 2A). The crystal structure of Aquifex aeolicus TatC (37, 38) portrays the TM helices folded into the shape of a cupped hand or glove. This arrangement produces a "palm" with a strikingly broad concave cavity (Fig. 3A) in which the TatA oligomer is thought to form. The six TMs appear to be stabilized by an overhanging cap structure formed by trans loops 1 and 2.

TatB and TatA
TatB and TatA are highly homologous proteins with a short N-terminal TM, a hinge, an amphipathic helix (APH), and a natively unstructured C-tail ( Fig. 2A). Solution and solid-state NMR structures of bacterial TatB (39) and TatA (40 -42) show that the TM and APH bend at the hinge to form an L-shaped molecule. Both of the E. coli proteins have a kink in their APH domain, with the TatB APH showing a more acute change in direction (39). TatB also has a longer C-tail than the TatA C-tail, both of which can be deleted without eliminating function (43,44).
Suppressor analysis (45), site-directed photocross-linking (46), and disulfide scanning cross-linking (37,45) in non-transporting conditions indicate that the E. coli TatB TM is associated with the C-proximal helix of TatC TM5 (colored cyan in Fig. 3). Because the TatB TM is matched in length to the TatC TM5 helix, it is thought that they align along their length (37). Photocross-linking also showed that the TatC cis 1 domain and trans 2 loop cross-link to TatB (46,47). Considering TatB topology (48 -51), a tentative arrangement of TatB association with TatC is as shown (Fig. 3B).
In non-transporting membranes, TatA exists primarily as a large pool of small homo-oligomers (29,32). Some TatA is also associated with isolated E. coli TatBC when components are overexpressed (52,53), although TatA is absent from isolated thylakoid TatBC (16,28). Nevertheless, cross-linking studies in E. coli and thylakoids demonstrate TatA contact with TatC in non-transporting conditions in a remarkably similar location as that occupied by TatB. Disulfide scanning (54) and photocross-linking (46,47) demonstrated contact of the TatA N terminus, TM, and APH with the corresponding TatC trans loops 2 and 3, TM5, and TatC cis regions, respectively. Photocross-linking from TatB suggests juxtaposition of TatA and TatB in the nontransporting Tat complex (46). Together, these results support the arrangement of TatA with TatC shown in Fig. 3B. TatA is placed distal to the TatC cavity with respect to TatB because TatB, but not TatA, contacts the substrate H domain in the absence of the PMF (20, 21) (see below).
Upon assembly of the thylakoid Tat translocase, TatA makes a new contact in the center of the concave cavity of TatC such that TatA Glu-10 is aligned with TatC Gln-234 on TM4 (Fig.  3B, Gln-234 colored magenta in Fig. 3, A and B) (54). This TatA-TatC contact exhibits the same requirements as translocase assembly: an RR signal peptide, the PMF, and the TatA TM Glu-10. These results support the proposal (37) that TatA initiates assembly of the transport-active oligomer by hydrogen bonding between a glutamine and a glutamate of partner proteins. The thylakoid TatC Gln-234 and TatA Glu-10 are conserved in the same relative positions among virtually all chloroplast and cyanobacterial TatC and TatA proteins. Bacterial TatC and TatA proteins commonly have a glutamate at the Gln-234 position and a polar residue at the Glu-10 position (5).

Substrate
Substrate binding triggers transition from receptor complex to the assembled translocase. Substrates exhibit a wide range of affinity for TatBC. Some bind in a highly reversible manner, their association only detected by cross-linking (16,26). Others remain bound during detergent solubilization and purification (25, 26, 28, 55-57). The signal peptides of tightly bound substrates are deeply inserted into the TatBC complex (24, 46), an FIGURE 2. Tat operates by a cyclical mechanism with a signal-assembled translocase. A, three components of Tat machinery in chloroplasts and E. coli. Chloroplast TatC has a very long N terminus of unknown function that is absent from bacterial TatC. APH domains are shown with a striped pattern. The relative molar quantities of components in situ are indicated in parentheses. B, cyclical mechanism for Tat protein transport. The TatBC receptor complex binds the substrate signal peptide in an energy-independent step. The receptor complex is depicted in the figure as a TatBC heterodimer, but it is actually a multimer estimated to contain up to eight TatBC units. Signal peptide binding triggers PMF-dependent assembly and oligomerization of TatA. The resulting complex is the translocase. Changes in the TatA oligomer are thought to facilitate protein transport, after which the translocase dissociates. arrangement considered to be an advanced binding stage. Interestingly, Gerard and Cline (24) found that a weakly binding substrate similarly inserted into TatBC when thylakoids were energized with the PMF, suggesting that all substrates achieve the advanced stage in actively transporting membranes. Some studies suggest that substrate can initially bind to the lipid bilayer before accessing TatBC (58,59). By contrast, binding of at least some thylakoid substrates to the lipid bilayer appears to be a dead end (25).
Binding of the high affinity tOE17-20F saturated at one substrate per TatC (25). With sufficient TatA, all bound tOE17-20F was efficiently transported. This initially suggested that each TatBC pair constitutes one binding and transport site as illustrated in Fig. 2A (but see below). A synthetic SufI signal peptide also bound with a 1:1 stoichiometry to purified A. aeolicus TatC, although with dramatically reduced affinity (37).
Knowledge of the precise location of bound signal peptide is critical for understanding how signal peptide alone can induce PMF-dependent translocase assembly (16,29,30). Early studies showed that TatC binds to the RR motif and TatB contacts the H domain (20,21). Recent extensive and complementary data indicate that residues of the TatC cis 1 domain and cis 2 loop form a specific RR binding site (37, 47, 56, 60 -63). Fig. 3A shows residues of chloroplast TatC (colored orange and red) that, when mutated to alanine, impair signal peptide binding. Residues colored red (glutamate) are thought to coordinate the arginine guanidinium groups; residues colored green direct disulfide cross-linking to RR proximal residues of the signal peptide (56). Complementary evidence also suggests that signal peptide H domain uniformly binds to the TatB TM, which is simultaneously associated with TatC TM5. Reciprocal photocross-linking experiments indicate that several regions of the signal peptide H domain contact the TatB TM (20,21,46). Signal peptides and the early mature domain bind in an inverted hairpin configuration, placing the H domain at the same membrane height as the TatB TM (64 -66). When TatB was absent from E. coli membranes, the signal peptide of some substrates was cleaved (without substrate transport) by the trans facing signal peptidase (66), implying that the TatB TM binding to the H domain prevents premature C domain exposure to the trans compartment. This is supported by photo-and disulfide cross-linking analyses that place the C-terminal end of the bound signal peptide in proximity with the trans-proximal end of TatC TM5 (46,54). Finally, suppressors of defective TorA signal peptides map to the TatB N terminus (62,63). It is apparent that all of these conditions can be met only if the signal peptide H domain binds to the TatB TM attached to a different TatC subunit (37,54) as depicted in the head-to-tail TatC arrangement in Fig. 3C. This will be further explored below.
Contacts between TatC and the folded substrate domain have not been detected. However, the TatB APH makes extensive contact with the folded domain of receptor-bound SufI (67), suggesting that the TatB APH serves as a platform for the folded domain of bound substrates (67). TatA contacts the signal peptide (20) and folded domain of the substrate later in the reaction cycle. Photocross-linking showed that TatA TM, APH, and C-tail are in contact with substrate only in the presence of PMF (68). These contacts precede transport and may represent translocase assembly events. Disulfide exchange cross-linking with thylakoids captured interactions between the substrate folded domain and the TatA APH and the N terminus (69). The latter contacts occurred during protein transport, implying that the substrate encounters the TatA TM during passage across the membrane. A, TatC is labeled for the cis (stroma) 1-4 exposed loops and tails, the TMs 1-6, and the trans (lumen) 1-3 loops. The pivotal C-proximal helix of TM5 is colored cyan. Residues of cis 1 and 2 important for RR binding are colored orange and red to designate loss of binding upon mutation to alanine, with red also designating essential glutamate residues that may coordinate the arginine guanidinium groups, and green designating residues that direct disulfide cross-linking to RR proximal residues of the signal peptide (56). B, arrangement of TatB (37,45,46) and TatA (54)

Models for TatA-facilitated Translocation
Several observations suggest that TatA forms the translocation pore. It contacts substrate late in the transport reaction, is required only for the transport step (28), is present in superabundance over TatB and TatC (Fig. 2), forms large oligomeric assemblies (29,30,32) (Fig. 2A), and was found associated with an apparent transport intermediate (70).
Two general pore models have been proposed. The first, invoking form-fitting channels lined with TatA subunits (71,72), was inspired by the appearance of channel-like structures of varying diameter when TatA was detergent-extracted from non-transporting membranes (71). Such channels might be formed by concerted bilayer insertion of TatA APHs (73) to produce a water-filled passageway, similar to the action of amphipathic antimicrobial peptides (74). However, this model is less attractive because recent analyses show that TatA exists as small oligomers in non-transporting membranes (29,(32)(33)(34). In addition, the expected topology inversion entailed by APH bilayer insertion was not found in two recent studies that analyzed TatA topology before and during translocation (50,51).
A second model proposes that TatA assemblies "weaken" the bilayer (75). Such weakening might occur by severe bilayer thinning due to hydrophobic mismatch between the very short TatA TM and the lipid bilayer. This model is supported by a structure and modeling analysis of E. coli TatA (42) showing that, whereas an oligomer of 25 TatA subunits is stable in detergent, small oligomers would be unstable in a lipid bilayer. An oligomer of between four and nine subunits would produce severe distortions that destabilize intermolecular TM contacts and lead to local bilayer rupture (42) (Fig. 3D). The translocaseassociated TatA assemblies are estimated to contain ϳ25 TatA subunits by fluorescence imaging (32) and kinetic measurements (25), and disulfide cross-linking places oligomer size at 8 -16 TatA subunits (29). Thus, the TatA oligomers deemed completely unstable in lipid bilayers have been shown to exist in the translocase. This evokes a model whereby the translocase creates an environment that permits the assembly of bilayerdestabilizing TatA oligomers and harnesses this instability to produce transient pores for substrate passage (29).
The strength of this model is that it agrees with and can explain experimentally determined characteristics of the protein transport reaction. Topology studies with thylakoid membranes (51) confirm structural studies of TatA in bicelles (40) that the APH is angled in non-transporting conditions, effectively relieving hydrophobic mismatch. The same topology studies suggest that the APH is more evenly partitioned into the bilayer interface in the translocase. Similarly, cross-linking studies indicate that the translocase TatA docking site on TatC places it higher in the membrane than the constitutively docked TatA (Fig. 3B) (54). Both observations support the bilayer thinning arrangement of translocase-associated TatA (Fig. 3D). The model also explains the observation that translocase-associated TatA oligomers are the same size whether induced by a synthetic signal peptide or by a complete substrate protein (29), and it gives a rationale for the translocase being assembled anew for each substrate (16,33). Collapse of a metastable TatA oligomer might be subject to stochastic thermal fluctuations, which would be consistent with the slow first order transport kinetics following TatA assembly (25).

The Oligomeric TatBC Structure May Enable Gated TatA Assembly to Form the Translocase
Regardless of the precise translocation mechanism, it is clear that the environment in the translocase allows the TatA oligomer to form. Stability may come from interactions between the folded substrate and the TatA APH, but TatC scaffolding surely provides considerable stability. It is also likely (42) that the TatA oligomer would assemble by stepwise addition of TatA monomers or very small oligomers.
The higher order arrangement of Tat (A)BC probably plays a critical role in assembly of the TatA oligomer. Until now, I have discussed the TatB-TatC heterodimer as a functional unit ( Figs.  2A and 3, A and B). However, TatBC actually exists as a multimer and appears to function as one. TatBC is estimated to contain 6 -8 copies of each protein based primarily on the electrophoretic and chromatographic migration of the detergentsolubilized complex (28,52,57). Several studies show that this large complex is present in situ (33,34,56). The functional necessity of multimeric TatBC is indicated by the fact that TatC mutations that impair TatBC assembly also impair substrate binding and transport (56,60), whereas mutations in the RR binding site do not affect assembly (56,60). Multimeric TatBC also appears capable of collectively binding and transporting multiple substrates. Specifically, disulfide cross-linking of individually bound substrates via their signal peptide H/C (54,56) and early mature domains (10) produced substrate dimers and tetramers, respectively, indicating a common compartment for binding multiple substrates. Efficient transport of the dimers and tetramers without breaking the cross-links suggests that TatBC units can combine to form a common translocation pore. Genetic data suggest that the TatBC complex is built from dimeric TatC units (76). In particular, a fused TatC dimer was fully functional when one, not both, RR binding sites was mutated. This is consistent with the positioning of the signal peptide between two TatC subunits as illustrated in Figs. 3C and Fig. 4.
The next major challenge will be to determine how TatBC subunits are arranged to produce such a functional structure. Photo-and disulfide cross-linking provide some insight. Certain TatC-TatC cross-links support a face-to-face dimer arrangement of TatBC units (Fig. 4A). Disulfide scanning of thylakoid TatC (56) produced TatC dimers from three different concave face positions, whereas dimers were not obtained from any of eight different convex side (back) positions (54,56). Disulfide scanning with E. coli TatC produced four physiologically relevant dimers (77): one in trans 1 loop on the concave face (46), one in trans 2 loop that may face front or back, and two in TM1 at the edge of the TatC cup. These latter two crosslinks are consistent if face-to-face TatC dimers are arranged end-to-end. This configuration might allow a pair of dimers to unfold into a tetramer (Fig. 4). In fact, certain cross-links produced by bismaleimide cross-linking of asymmetric pairs of Cys residues support dynamic arrangements of TatBC dimers, tetramers, and hexamers in which the concave faces of all TatCs point inward (46) (Fig. 4). The exact pairing geometry of TatBC units is not entirely clear, but certain TatC-TatB photocrosslinks suggest a possible arrangement. Blümmel et al. (46) observed photocross-linking to TatB directed from TatC TM2 and TM4 as well as from TM5. Their interpretation is that the TM of each TatB subunit in the multimeric TatBC complex bridges TM5 of one TatC subunit and the TM2-TM4 cleft of an adjacent or opposing TatC subunit (Fig. 4, A and B). This arrangement can explain the observed cross-linking of E. coli TatB trimers and tetramers (67,78). It is also an intriguing pairing because it places the TatB TM in the approximate site where TatA appears to initiate translocase oligomer assembly. Considering the high level of homology between the TMs of TatA and TatB, this supports a role for TatB as regulatory surrogate for TatA. In this regard, the simplest Tat systems function with just a TatC and a TatA.
Substrate binding appears to "subtly" change the arrangement of TatBC subunits. Two of three disulfide cross-links between thylakoid TatC subunits were greatly diminished by prior binding of substrate. Substrate binding also increases the stability of detergent-solubilized thylakoid TatBC complex on blue native PAGE, increases stability of the TatB-GFP-TatC complex in living E. coli cells (34), and changes the appearance of TatBC in single particle EM imaging (57). Nevertheless, substrate binding does not change the face-to-face arrangement of TatBC units. As evidence, a study with thylakoids produced a disulfide cross-linking product in which two bound substrates were linked to each other via their signal peptide C domains and were independently linked to separate TatC subunits via the RR and RR binding site domains (54). Suggested arrangements for TatABC units and substrate are shown in Fig. 4, C and D. These arrangements preserve the head-to-tail arrangement of TatC subunits for signal peptide binding and also create an enclosed chamber for assembly of the TatA translocation oligomer.

PMF and Signal Peptide Binding as Assembly Triggers
Mechanisms for how the PMF and signal peptide trigger TatA assembly are speculative. For thylakoids, the ⌬pH component of PMF probably protonates the TatA Glu-10, making it energetically feasible to move up in the membrane to its docking site on TM4 and hydrogen-bond with TatC Gln-234 ( Fig.  3B) (54). It is unclear how the PMF facilitates assembly of E. coli TatA, where only the electrical component of the PMF is required (79). Signal peptide binding must alter the association of components to provide a TatA pathway to the interior as well as an open assembly site on the concave TatC face. As depicted in Fig. 4, this would involve altering the association of TatB with TatC, which appears to block TatA entry into and assembly within the central compartment (46,54). Perhaps signal peptide H domain binding displaces the TatB TM from the TatC TM2-TM4 meshing site and also counterbalances the attraction of TatB TM for TatC TM5, opening the way for TatA movement in C, which shows N-terminal tails of TatB, which are known to collectively cross-link. A and B, the face-to-face arrangement of dimers and tetramers is based on disulfide cross-linking, photocross-linking, and bismaleimide cross-linking (46,54,56) of TatC to TatC and photocross-linking between TatB and TatC (46). C and D, the arrangement of the substrate-bound TatABC is based on signal peptide cross-linking to TatC (20,21,47,54,56) and TatB, and the decrease of certain disulfide cross-links when signal peptide is bound. The signal peptide is depicted as a helical black line, the early mature domain is depicted as a dashed line, and the folded domain of the substrate is depicted as a sphere. Note that substrate binding increases the size of the chamber created by TatBC units where the TatA pore is proposed to assemble. It is also apparent that TatB controls TatA entry into the chamber. See the text for additional detail.
to the interior. Opening of the TatA pathway might be infrequent and transient, considering the time scale of 1-3 min for assembling the translocase (25). Such a transient "breathing" of the Tat assembly might also be facilitated by an increased lateral pressure on chamber walls due to the insertion of signal peptides and early mature domains into the chamber. At present, the nature of the TatBC enclosed compartment is not known and has been speculated to be either aqueous or polar lipid-filled (37,38).

Perspectives
The mechanistic models presented here are preliminary, with some aspects based on a single study; clearly these await replication. However, they emphasize that a detailed map of the subunit organization of TatABC with and without substrate is crucial to further progress. Especially important are the interactions of components and signal peptide that occur at TatC TM5 because these appear to regulate assembly of the translocase. Molecular structures of relevant complexes would be a major advance toward deciphering Tat mechanism, but even additional cross-linking with functionally substituted components and substrate would refine current maps. Certainly, other approaches to these questions are essential.
The core issue, i.e. how do TatA assemblies enable passage of the substrate across the membrane, is the most challenging because transmembrane passage may occur quite rapidly. Indirect clues to the mechanism may come by manipulating membrane composition of bilayer-forming and non-bilayer-forming polar lipids. The recently acquired ability to stabilize the translocase (54) may permit freezing the translocase for structural analysis. Nevertheless, even at this relatively early stage, it is apparent to me that Tat transports proteins by the novel strategy of organizing an unstable lipid/protein combination into the nano-environment of the translocase. This creates a restricted permeation pathway, whereas on a cellular level, the same instability would be catastrophic.