Escherichia coli TatA and TatB Proteins Have N-out, C-in Topology in Intact Cells*

Background: The Tat pathway transports folded proteins across energy-coupling membranes. Results: TatA has a fixed N-out, C-in topology in intact cells that is not altered by the absence of other Tat components or overproduction of a Tat substrate. Conclusion: The TatA amphipathic helix does not re-orient during protein translocation. Significance: Topological inversion of TatA does not accompany protein transport by the Tat pathway. The twin arginine protein transport (Tat) system translocates folded proteins across the cytoplasmic membrane of prokaryotes and the thylakoid membrane of chloroplasts. In Escherichia coli, TatA, TatB, and TatC are essential components of the machinery. A complex of TatB and TatC acts as the substrate receptor, whereas TatA is proposed to form the Tat transport channel. TatA and TatB are related proteins that comprise an N-terminal transmembrane helix and an adjacent amphipathic helix. Previous studies addressing the topological organization of TatA have given conflicting results. In this study, we have addressed the topological arrangement of TatA and TatB in intact cells by labeling of engineered cysteine residues with the membrane-impermeable thiol reagent methoxypolyethylene glycol maleimide. Our results show that TatA and TatB share an N-out, C-in topology, with no evidence that the amphipathic helices of either protein are exposed at the periplasmic side of the membrane. We further show that the N-out, C-in topology of TatA is fixed and is not affected by the absence of other Tat components or by the overproduction of a Tat substrate. These data indicate that topological reorganization of TatA is unlikely to accompany Tat-dependent protein transport.

The Sec 3 and Tat pathways operate in parallel to transport proteins across the cytoplasmic membrane of bacteria and archaea and the thylakoid membrane of plants. Although the Sec machinery can only export unfolded proteins, substrates of the Tat pathway are folded prior to transport. Transport by the Tat pathway is driven solely by the transmembrane proton electrochemical gradient (⌬p) (reviewed in Refs. 1,2). Substrates are targeted to each of these transport machineries by means of N-terminal cleavable signal peptides. One of the major distin-guishing features of Tat targeting signals is that they contain a conserved twin arginine motif, which harbors consecutive and usually invariant arginine residues that are essential for efficient transport by the Tat machinery (3,4).
In Gram-negative bacteria and in plant thylakoids, the Tat machinery is made up of three membrane proteins which in Escherichia coli are termed TatA, TatB, and TatC (5)(6)(7)(8). In E. coli a fourth protein, TatE, is a minor component of the Tat pathway and has an identical function to TatA (7,9). The TatA and TatB proteins share some primary sequence homology and have evolved from a common ancestor, but they have functionally distinct roles during Tat transport (8,10). TatB is found almost exclusively as part of the TatBC complex (11). This complex, which contains multiple copies of each protein, interacts with twin arginine signal peptides and acts as the receptor for Tat substrates (11)(12)(13)(14)(15)(16).
TatA can be purified as an array of large homo-oligomeric complexes. Analysis of these complexes by negative stain electron microscopy reveals that they form a series of related channel-like structures of different sizes, with internal cavities big enough to accommodate folded proteins, consistent with the idea that TatA forms the protein-conducting channel (17,18). Large assemblies of fluorophore-tagged TatA have also been observed in vivo, and homo-oligomers containing at least 16 copies of the plant ortholog of TatA, Tha4, have been detected by cross-linking during Tat transport in thylakoid membranes (19,20). In E. coli cells, the formation of TatA assemblies is dependent upon the presence of the TatBC complex. In the absence of TatB or TatC, TatA is arranged as much smaller units, possibly tetramers, suggesting that interaction with TatBC is required to drive the polymerization of smaller units of TatA into larger assemblages (20). In resting thylakoid membranes, cross-linking studies of Tha4 are also consistent with this protein existing as a tetrameric unit (19). Transient interactions of Tha4 with the thylakoid equivalent of the TatBC complex have been detected by cross-linking, dependent upon the presence of substrate and a ⌬p (21).
The TatA and TatB proteins have a common structural arrangement, comprising a single transmembrane helix (TMH), followed by an amphipathic helix (APH) and an unstructured C-terminal (Fig. 1, A and B) (18,22). Application of the positive inside rule (23) would suggest that the N termini of both proteins are located at the periplasmic side of the membrane with the C termini in the cytoplasm, and this is consistent with protease accessibility experiments that indicated that the C termini of TatA and TatB are only accessible to protease digestion in inside-out and not right side-out membrane vesicles (18,22). A similar topology was inferred from protease mapping of Tha4 in isolated thylakoids (24). The N-out topology of TatA is also supported by the observation that the TatA protein from Providencia stuartii, which is synthesized with an inactivating N-terminal extension of eight amino acids, is processed by a membrane-embedded rhomboid protease that has its active site close to the periplasmic face of the membrane (25)(26)(27). However, a recent study probing the accessibility of cysteine-substituted TatA to sulfhydryl-labeling reagents in whole cells appeared to show that the N terminus of TatA was localized at the cytoplasmic side of the membrane (28).
Some models for Tat transport assume that the APH of TatA may re-orient during transport (Fig. 1C), for example by folding into a channel assembled from TatA transmembrane helices like a trapdoor in response to a pulling force on the substrate (1,29,30). Early support for this model was provided by Gouffi et al. (31), who used compartment-sensitive marker proteins fused to the end of the APH of TatA to infer that this region of TatA was exposed at both sides of the membrane. Similar dual topology was also seen when a much smaller fusion, that of a tobacco etch virus protease cleavage sequence, was inserted between residues 53 and 54 of TatA because this site was also shown to be protease-accessible from either side of the membrane. This led the authors to conclude that the TatA APH has a dual topology and that topology changes of this region of TatA are associated with protein transport (31). Support for a helical hairpin arrangement of TatA was also provided by Chan et al. (28), who showed that in whole cells cysteine residues in the APH or C terminus of TatA were not labeled by a membraneimpermeable thiol reagent. They further showed that in the presence of an uncoupler the labeling pattern of a cysteine present in the APH of TatA was altered, suggesting that topological changes in the APH were dependent upon ⌬p.
In this study, we have re-visited the topological organization of TatA and TatB in whole cells by direct labeling of engineered cysteine residues. Our results clearly show that TatA has a fixed N-out, C-in topology that is not altered by the absence of other Tat components or by the overproduction of a Tat substrate.

EXPERIMENTAL PROCEDURES
E. coli Strains and Plasmids-Strains and plasmids used in the present work are shown in Tables 1 and 2, respectively. All strains used in this study are derivatives of MC4100 (32).
Plasmids in strain DADE-P (22) were used to overexpress the E. coli tatABC operon with alanine substitutions of all four cysteine codons in tatC and single cysteine codon substitutions in tatA or tatB. Cysteine substitutions were introduced by QuikChange TM site-directed mutagenesis (Stratagene) in plas- . The position of the invariant glycine found throughout TatA and TatB family proteins is boxed. Predicted secondary structure elements are shown above the amino acid sequences, with the transmembrane helix in dark gray and the amphipathic helix in light gray. Secondary structure predicted using PSIPRED 3.0 (59) and predicted helical regions are shown as cylinders. Note only the first 100 amino acids of TatB are shown. C, possible topological organizations for E. coli TatA based on previous experimental observations. mid pUNITATA (30) or UNITATB (22) resulting in plasmid series pUNITATAX#C or pUNITATBX#C, respectively, where X corresponds to the single letter amino acid code and # to the position of the substituted codon. Each tatA mutation was sub-cloned into the EcoRI and PmlI sites of plasmid pUNITA-TCC4 (22) giving rise to plasmid series pUNITATCC4AX#C. Each tatB mutation was subcloned into the PmlI and AflII sites of plasmid pUNITATCC4 (22) giving rise to plasmid series pUNITATCC4BX#C. Primer sequences used for Quik-Change TM mutagenesis are available on request.
For the cysteine insertion between residues 1 and 2 of TatA, DNA covering the tatA start codon and upstream DNA was amplified with primers TatPromXcaBamrev (5Ј-GCGCGGATCCGTATACATGTTCCTCTGTGGTAGATG-3Ј) and TATA5 (7) and cloned into the EcoRI and BamHI sites of pBluescript KS ϩ (Stratagene) to give plasmid pBSTatAProm-XcaI. Plasmid pBSTatAins2C was constructed following amplification of tatA using primers TatAins2C (5Ј-TGTGTGGTG-GTATCAGTATT-3Ј) and TatAEcPmlBam (5Ј-GCGCGG-ATCCCACGTGTTACACCTGCTCTTTATCG-3Ј) and subsequent cloning of the PCR product into the BamHI and XcaI (blunt-end) sites of plasmid pBSTatAPromXcaI. The tatA gene with the cysteine insertion was subsequently excised by digestion with EcoRI and PmlI and cloned into similarly digested pUNITATCC4 to give pUNITATCC4Ains2C.
A cysteine insertion between codons G21 and T22 was constructed as follows. DNA covering tatA up to codon 21 was amplified using primers TatATMH1 (5Ј-GCGCGGATCCTG-GCCAAAAAGCAGTACAACGATGA-3Ј) and TATA5 and cloned into pBluescript as an EcoRI-BamHI fragment. DNA covering tatA from codon 22 onwards was amplified using TatAGly1 (5Ј-CGGCACCAAAAAGCTCGGCTCCATCGG-3Ј) and UNIA1 (30), digested with BamHI, and cloned into the above plasmid that had been previously digested with MscI (blunt end) and BamHI, to give plasmid pTatAGly1. This con-  As pUNITATCC4 tatA harboring a cysteine codon insertion between codons 1 and 2 This study pUNITATCC4Ains21C As pUNITATCC4 tatA harboring a cysteine codon insertion between codons 21 and 22 This study pUNITATCC4BX#C As pUNITATCC4 harboring single cysteine codon substitutions in tatB as indicated a This study and Ref. struct codes for TatA with a glycine insertion between G21 and T22. The extra Gly codon was subsequently changed to a Cys codon by QuikChange TM site-directed mutagenesis to give plasmid pBSTatAins21C. The tatA gene with the cysteine insertion was subsequently excised by digestion with EcoRI and PmlI and cloned into similarly digested pUNITATCC4 to give plasmid pUNITATCC4Ains21C. The tatABC operon in single copy harboring single cysteine codon substitutions in tatA and alanine substitutions of all four cysteine codons of tatC was expressed from the chromosomal phage attachment site, attB, of strain DADE (33). The tatA gene and ϳ100 bp of upstream promoter region were amplified from E. coli chromosomal DNA with primers UNIREP1 and UNIA1 and cloned into the EcoRI and BamHI sites of pBluescript KS ϩ generating plasmid pKSuniA. Single cysteine substitutions in tatA were introduced by QuikChange TM site-directed mutagenesis in pKSuniA giving plasmid series pKSuniAX#C. The tatABC operons under control of the tatA promoter in plasmids pUNICC-AX#C were subcloned into the EcoRI and BamHI sites of plasmid pRS552 (34) resulting in plasmid series pRSUNICC-AX#C. The tatABC operon under control of the tatA promoter from pRSUNICC-AX#C was subsequently integrated into the attB site of strain DADE as described (34) resulting in strains MF1-13 (listed in Table 1). A series of strains expressing only tatA were also constructed. In this case, the tatA variants with single cysteine substitutions under control of the tat promoter in the pUNITATAX#C plasmid series were subcloned into the EcoRI and BamHI site of pRS552 and subsequently integrated into the attB site of strain DADE resulting in strains SK1-13 ( Table 1).
Culture Conditions, Fractionation, and Protein Methods-Unless indicated otherwise, liquid cultures were inoculated with 1:100 volume of an overnight culture in LB medium supplemented with appropriate antibiotics and grown with vigorous shaking at 37°C (40). Growth assays to test resistance to SDS were performed as described (41). Growth assays to test Tat-dependent growth with trimethylamine-N-oxide (TMAO) as sole electron acceptor were carried out by growing strains anaerobically at 37°C for up to 4 days on solid M9 minimal The crude membrane fraction was isolated by ultracentrifugation and divided into 4 aliquots (each of 100 -150 g of protein). One aliquot was incubated with buffer alone, whereas the remaining 3 aliquots were incubated with MAL-PEG, MAL-PEG plus 1% Triton X-100, or MAL-PEG plus 1% SDS. The reaction was quenched by addition of 45 mM DTT, and a sample of each aliquot (4.5-7 g of protein) was separated by SDS-PAGE on 15% Tris-glycine gels. Proteins were transferred to nitrocellulose membrane, and TatA was detected with anti-TatA antiserum. The positions of the molecular mass markers are indicated to the left of the panels, and the positions of unlabeled and PEGylated TatA to the right of the panels. A possible TatA-lipid cross-link seen for some of the TatA variants and reported previously (30) is indicated with an asterisk. Labeling of TatA Cys substitutions shown for the N terminus and within the TMH (A), around the hinge region (B), and in the APH (C).
Cultures for TMAO reductase assays were grown anaerobically in 50 ml of LB medium supplemented with 0.4% (w/v) TMAO and 0.5% (v/v) glycerol at 37°C overnight (8). Cells were separated into spheroplast and periplasmic fractions using lysozyme/EDTA treatment in a high sucrose buffer followed by centrifugation (42). TMAO:benzyl viologen oxidoreductase activity was measured in periplasmic fractions of TMAOgrown cells as described previously (43).
SDS-PAGE and immunoblotting analysis was performed as described (44,45), and immunoreactive bands were detected using a chemiluminescent horseradish peroxidase (HRP) substrate kit (Millipore). Antisera against E. coli TatA and TatB (46) were used to detect the proteins, with an anti-rabbit IgG HRP conjugate (Bio-Rad) used as secondary antibody. SufI was detected using a polyclonal anti-SufI antiserum (41). Hemagglutinin-tagged SoxY was detected with anti-hemagglutinin tag (HA tag) HRP conjugate (Sigma).
Sulfhydryl Labeling-For sulfhydryl labeling with crude membrane fractions, cells were grown in 25 ml of LB medium at 37°C overnight, harvested by centrifugation at 5000 ϫ g, and washed in Buffer K (50 mM triethanolamine-HCl, pH 7.5, 250 mM sucrose, 1 mM Na 2 EDTA). Cell pellets were resuspended in 1 ml of Buffer K supplemented with protease inhibitors (Complete Mini, EDTA-free protease inhibitor mixture tablets; Roche Applied Science). Cells were disrupted on ice by five cycles of sonication with 15-s pulses and 15-s intervals in between. Cell debris was pelleted by centrifugation at 16,000 ϫ g for 10 min, and the crude membrane fraction was pelleted from the supernatant by ultracentrifugation at 278,000 ϫ g for 30 min. The crude membrane fraction was resuspended in 1 ml of HEPES/NaCl buffer (50 mM HEPES, pH 6.8, 50 mM NaCl) and diluted to 10 -15 g l Ϫ1 of total protein. 10 l of the membrane fraction were labeled with 5 mM methoxypolyethylene glycol maleimide (5000 Da, Sigma) in a final volume of 50 l of HEPES/NaCl buffer at room temperature for 1 h. Control samples were incubated with buffer alone or were treated with 1% Triton X-100 or 1% SDS prior to labeling. The sulfhydryl labeling reaction was stopped by addition of 5 l of 0.5 M dithiothreitol (DTT) and mixed 1:1 with 2ϫ Laemmli sample buffer (Bio-Rad).
For sulfhydryl labeling with intact cells, cultures were grown at 37°C to mid-logarithmic phase and normalized to an A 600 of 0.3. Cells were harvested by centrifugation at 5000 ϫ g and washed with HEPES/MgCl 2 buffer (50 mM HEPES, pH 6.8, 5 mM MgCl 2 ), and cell pellets were resuspended in 1 ml of HEPES/MgCl 2 buffer. 80 l of cell suspension were incubated with 5 mM methoxypolyethylene glycol maleimide (MAL-PEG) at room temperature for 1 h in the presence of 5 mM EDTA and in a final volume of 100 l of HEPES buffer. The sulfhydryl labeling reaction was stopped with 25 l of 0.5 mM DTT, and proteins were precipitated as described (47). Protein precipitates were resolubilized in 70 or 100 l 2ϫ Laemmli sample buffer depending if TatA variants of samples were expressed from the chromosome or from plasmid, respectively.

RESULTS
Cysteine-substituted TatA Proteins Can Be Labeled with MAL-PEG-Previous studies looking at cysteine accessibility of E. coli TatA and TatC proteins have used an indirect method to detect sulfhydryl-labeled proteins (28,48). N ␣ -(3-Maleimidylpropionyl)-biocytin (MPB) is a membrane-permeable maleimide reagent with a molecular mass of around 500 Da that can subsequently be detected by binding of streptavidin. 4-Acetamido-4Ј-maleimidylstilbene-2,2Ј-disulfonic acid is a membrane-impermeable maleimide of similar molecular mass to MPB. Periplasmically located cysteines in TatA and TatC were identified by pretreating cells with 4-acetamido-4Ј-maleimidylstilbene-2,2Ј-disulfonic acid, which will only react with external cysteines, followed by MPB, which will react with any cysteines that have not been previously labeled with 4-acetamido-4Ј-maleimidylstilbene-2,2Ј-disulfonic acid (28,48). Because these reagents do not alter the apparent mass of the protein with which they react, labeling can only be detected once the protein has been purified and incubated with streptavidin. According to this method, periplasmic cysteines are identified because they show a lower level of MPB labeling than cytoplasmic cysteines.
Because this approach is rather time-consuming and requires significant amounts of material (because of losses at the purification step), we employed a more direct approach to assess cysteine accessibility. MAL-PEG is a membrane-impermeable maleimide of much larger mass (ϳ5,000 Da) that results in a clear size-shift of the labeled protein, which can be detected by Western blotting (38,49). To assess whether cysteine substitutions in TatA that might be expected to lie close to the membrane were accessible to labeling by the relatively large MAL-PEG reagent, we first isolated membrane fractions from E. coli strains producing plasmid-encoded TatA variants along with TatB and a cysteine-less variant of TatC. Cysteine residues close to the N terminus of TatA (i.e. a G2C substitution or a variant with a Cys insertion between M 1 and G 2 termed Ins2C) clearly reacted with MAL-PEG to give a species that migrated more slowly than unlabeled TatA (Fig. 2A). Also as expected, a TatA protein lacking Cys residues did not label with MAL-PEG, and a Cys residue that is predicted to be buried in the TMH (I12C) did not label with MAL-PEG unless detergent was also included in the labeling reaction.
Cysteine residues in the hinge region between the TMH and APH showed poor labeling in the absence of detergents (Fig.  2B). Thus, Cys residues at positions 18 -20 did not react with MAL-PEG in isolated membranes, nor did a Cys residue that had been inserted between G21 and T22. The first residue in this hinge region to become accessible to PEGylation was a Cys substitution at position 22. This suggests that the preceding residues are buried within the membrane, which is in accordance with the findings of a solid state NMR study on the Bacillus subtilis TatA d protein (50). All of the Cys substitutions that were tested along the APH were freely accessible to MAL-PEG labeling in isolated membranes (Fig. 1C), indicating that for at least a proportion of the TatA molecules the APH must lie along the membrane rather than being in a transmembrane orientation.
Labeling of Accessible Cys Residues with MAL-PEG in Intact Cells-The outer membranes of Gram-negative bacteria are generally permeable to hydrophilic molecules Ͻ600 Da due to the presence of porins, but they do not allow passage of molecules larger than this. Therefore, to test whether any of the membrane-extrinsic Cys substitutions of TatA were accessible to MAL-PEG in whole cells, it was necessary to devise a method that would allow the compound to permeate across the outer membrane. EDTA chelates divalent metal ions that are associated with the lipopolysaccharide, offsetting their stabilizing effect and increasing permeability (51). We therefore added 5 mM EDTA to cell suspensions prior to addition of MAL-PEG. Control experiments showed that the EDTA treatment did not affect the viability of cells (supplemental Fig. S1).
MAL-PEG accessibility was tested for TatA variants harboring Cys substitutions at position 2, Ins2 and 22 to determine the topology of the TMH, and 33, 35, 39, 41, 45, 47, 49, 60, 78 and 89 to determine the localization and topology of the APH and C-terminal tail. Some of these Cys substitutions have been described previously (30), and it was shown that the G2C, S35C, and K41C variants supported a high level of Tat transport activity, and T22C supported a low but detectable level of Tat transport, whereas G33C and F39C inactivated the function of TatA. When we assessed the activity of all of the new substitutions constructed here (Ins2C, D45C, E47C, K49C, T60C, T78C, and V89C) each of them supported high Tat transport activity (supplemental Fig. S2A) indicating that TatA function was preserved.
As a control for each of the labeling reactions, the E. coli strains co-produced a cytoplasmic form of the P. pantotrophus SoxY protein carrying an N-terminal hemagglutinin epitope tag. This protein contains a highly exposed single Cys residue on a flexible arm at the C terminus of the protein (35), and any MAL-PEG labeling of this indicated that the integrity of the cytoplasmic membrane was compromised.
TatA variants containing Cys substitutions at the N terminus were clearly accessible to PEGylation in intact cells, whereas the T22C variant was not (Fig. 3A). The T22C variant was, however, PEGylated in the presence of SDS. Analysis of SoxY labeling indicated that the cytoplasmic membrane was not breached during the procedure. These data show that the TatA TMH has an N-out, C-in topology. We noted that the detectability of SoxY and TatA appears to increase dramatically after PEGylation. This may be due to better transfer and/or increased binding of the PEGylated proteins to the nitrocellulose membrane during electroblotting. Fig. 3, B and C, shows the labeling patterns for Cys residues located in the APH (Fig. 3B) and C-tail (Fig. 3C). Each of the positions gives a labeling pattern that is identical to that for T22C, with the single exception of the V89C substitution, which showed some labeling in the absence of SDS. However, in this case, we noticed that the integrity of the inner membrane was compromised resulting in PEGylation of SoxY. We rou- tinely observed that the plasmid-produced V89C substitution induced leakage of the inner membrane toward MAL-PEG, for reasons that are not clear. However, taken together, the results clearly indicate that the APH and C-tail of TatA have a cytoplasmic localization, with no evidence that the APH can flip across the membrane.
TatA Shows the Same Topology When Produced at Native Levels and in the Absence of TatBC-All of the results obtained above and in the experiments undertaken by Chan et al. (28) studied TatA variants that were overproduced from a multicopy plasmid (along with TatB and TatC). It has been shown that even a relatively low level of tatABC overexpression can result in an aberrant localization of TatA, which accumulates in the cytoplasm as tube-like structures (52). The presence of cytoplasmic forms of TatA might be expected to interfere with labeling experiments because a proportion of the TatA protein will have its N terminus located in the cytoplasm rather than in a transmembrane orientation. We therefore repeated our PEGylation experiments with TatA proteins produced at chromosomal levels.
DNA encoding TatA variants G2C, ins2C, T22C, G33C, S35C, F39C, K41C, D45C, E47C, T60C, T78C, and V89C along with TatB and cysteine-less variant of TatC, under the control of the tatA promoter, were placed into the phage attachment site, attB, on the chromosome of strain DADE (which is deleted for all natively encoded tat genes). The activity of the Tat system producing these variants at the chromosomal level was assessed by determining the activity of the Tat substrate (TMAO reductase) TorA in the periplasmic fraction. As shown in Fig. 4, the S35C, D45C, E47C, T78C, and V89C variants had levels of periplasmic TorA activity that were indistinguishable from the strain carrying wild type tatABC at the attB site. The other strains had very low levels of periplasmic TorA activity, close to that seen for the negative control. These strains were FIGURE 6. TatA has an N-out, C-in topology in the absence of other Tat components. Cells of strains SK2-SK13 producing cysteine-substituted TatA at chromosomal levels in an otherwise tat Ϫ background were grown in 50 ml of culture until mid-exponential phase was reached. Cells were harvested and resuspended in 1 ml of HEPES/MgCl 2 buffer. 80-l aliquots of cell suspension were incubated with buffer alone or 5 mM MAL-PEG in the presence or absence of 1% SDS for 1 h at room temperature. Reactions were quenched with 45 mM DTT; proteins were precipitated with chloroform and methanol and separated by SDS-PAGE. After electroblotting, TatA was detected with anti-TatA antiserum and SoxY using an anti-hemagglutinin HRP conjugate. The positions of the molecular mass markers are indicated to the left and the positions of PEGylated and non-PEGylated protein to the right. Labeling of TatA Cys substitutions shown for either side of the TMH (A), the APH (B), and the C-terminal tail (C). APRIL 27, 2012 • VOLUME 287 • NUMBER 18 also examined for Tat activity using two phenotypic growth tests: growth on media containing 2% SDS (which requires the export of the Tat-dependent cell wall amidases AmiA and AmiC; see Refs. 53,54), and growth on minimal medium containing TMAO as the sole electron acceptor (which assesses the combined export of TorA and the membrane-bound dimethyl sulfoxide reductase, which can also reduce TMAO). All strains were able to grow on 2% SDS with the exception of those producing TatA variants G33C and T60C. Likewise, all strains could grow with TMAO as the sole electron acceptor except for those producing the ins2C, T22C, G33C, F39C, and K41C TatA variants. Taken together, it would seem that at native expression levels the G33C substitution of TatA completely inactivates Tat function, whereas the remaining substitutions allow detectable transport of at least one Tat substrate protein.

TatA and TatB Have an N-out, C-in Topology
We next confirmed that all of the variant TatA proteins were stably produced (supplemental Fig. S3), and we then repeated the MAL-PEG accessibility assays in EDTA-permeabilized whole cells. It is clear from the data shown in Fig. 5 that Cys residues close to the N terminus of TatA were accessible to reaction with MAL-PEG in the absence of SDS (Fig. 5A) and thus reside at the periplasmic side of the membrane. By contrast, Cys substitutions in the hinge region, the APH or C-tail of TatA could only be PEGylated in the presence of SDS (Fig. 5, A-C), consistent with cytoplasmic localization of these regions. Thus, it is clear that the topology of TatA produced at close to native levels is similar to that of overproduced TatA.
To ascertain whether the topology of TatA was affected by the presence of other Tat components, we constructed a series of analogous strains where mutant tatA genes alone were expressed from the attB site of strain DADE. These cysteinesubstituted TatA variants, which were again shown to be stably produced (supplemental Fig. S3), gave MAL-PEG labeling results that were largely indistinguishable from the labeling results obtained in the presence of TatBC (compare Figs. 5 and 6). We conclude that TatA has a similar N-out, C-in topology in the presence or absence of other Tat components.
Overproduction of a Tat Substrate Does Not Alter the Topology of the TatA APH-The results presented above support an N-out, C-in topology for TatA under all conditions tested and provide no evidence for dual topology of the APH suggested by others (28,31). Although the cells clearly remained viable and presumably also transport competent during the EDTA treatment and PEGylation procedure (supplemental Fig. S1), we maximized the likelihood that the Tat machinery was operational during the labeling experiments by saturating the Tat pathway through the high level overproduction of the plasmidencoded Tat substrate SufI. As shown in Fig. 7A, SufI was massively overproduced from this construct when compared with native level, and the Tat machinery appeared to be saturated because there was significant SufI present in the spheroplast fraction. It can also be seen that some SufI was detected in the periplasmic fractions of strains producing each of the G33C, S35C, F39C, K41C, D45C, and E47C TatA proteins, confirming functionality of the variant Tat machineries, even for those TatA proteins that supported only low levels of Tat transport of TorA. Differential effects of tat point mutations have been noted previously, depending upon the nature of reporter proteins used to assess functionality of the Tat system and whether Tat components or substrate proteins are overproduced (2). Control experiments (Fig. 7A) showed that there was little disruption to the spheroplasts during the procedure because no HA-tagged SoxY could be detected in the periplasmic fractions. It should be noted that the SufI protein present in the spheroplast fractions had approximately the same mobility as that in the periplasmic fraction. It has been shown previously that SufI overproduced from the same construct used here is subjected to proteolytic processing of the N-terminal 17 amino acids, presumably by a cytoplasmic protease (55). Alternatively, these observations may suggest that there is incomplete spheroplast formation in these experiments.
When the 1-h PEGylation reaction was repeated in the presence of overproduced SufI, again none of the Cys residues tested reacted with the MAL-PEG unless SDS was also included in the samples (Fig. 7B). Thus, we conclude that it is extremely unlikely that the TatA APH becomes exposed at the periplasmic side of the membrane during substrate transport.
TatB Has an N-out, C-in Topology Assessed by MAL-PEG Accessibility-To confirm that TatA and TatB show similar overall topology, we applied the same MAL-PEG labeling approach to cysteine-substituted TatB proteins. Cysteine substitutions at positions 2 and 41 of TatB were already available (22), and we constructed a further three substitutions, two more in the extended APH of TatB at positions 45 and 50 and a substitution at the very C terminus of the protein (P171C). Previous studies have shown that TatB is tolerant to mutagenesis (22,56), and indeed each of the Cys-substituted TatB proteins supported a good level of Tat activity (Fig. 8A).
In these experiments, the TatB variants were produced from a plasmid along with TatA and a cysteine-less variant of TatC in a strain that was devoid of chromosomally encoded Tat components. MAL-PEG labeling in whole cells demonstrated that the only cysteine accessible to PEGylation in the absence of SDS was at position 2 ( Fig. 8B), confirming that the N terminus of TatB resides at the periplasmic side of the membrane. Cys residues in the APH and C terminus of TatB were not PEGylated unless SDS was also included in the reaction, indicating that these residues most likely reside at the cytoplasmic side of the membrane. We therefore conclude that TatB has the same topological organization as TatA.

DISCUSSION
The development of even the most basic models for Tat pathway function requires knowledge about the location and topology of the essential Tat components. Although the membrane localization of the E. coli TatABC components is not in doubt (e.g. Ref. 57), the orientation and number of membrane-spanning domains present in TatA have been a source of contention. Previously, the only experimental evidence directly addressing the location of the N terminus of TatA led to the conclusion that it resided at the cytoplasmic side of the membrane, contrary to the positive inside rule (28). Furthermore, two separate studies indicated that the APH of TatA showed dual topology, being exposed at both the cytoplasmic and periplasmic side of the membrane. This supported models where the topological re-orientation of the TatA APH accompanied substrate movement across the membrane (1,29,30).
In this study we re-examined the topology of E. coli TatA, as well as addressing for the first time the topological orientation of the related protein TatB. We developed a method that would FIGURE 8. TatB has an N-out, C-in topology. A, relative TMAO reductase activities were determined in the periplasmic fractions of strain DADE-P (⌬tatABCD, ⌬tatE, pcnB) harboring either pQE60 or pQE60 encoding TatA and cysteine-less TatC along with TatB variants containing a single cysteine substitution as indicated. The TMAO reductase activity of strain DADE-P transformed with pUNITATCC4, encoding wild type TatA, TatB, and cysteine-less TatC, was defined as 100% activity and corresponds to 1.9 M benzyl viologen oxidized per min and per mg of protein. The error bars represent the means Ϯ S.E., n ϭ 3. B, cells of strain DADE-P producing plasmid-encoded TatA and cysteine-less TatC along with either wild type TatB (WT) or the indicated TatB Cys-substituted variants in 50 ml of culture were grown to mid-exponential phase (A 600 of ϳ0.4), harvested, and resuspended in 1 ml of buffer. 80-l aliquots of cell suspension were incubated with buffer alone or 5 mM MAL-PEG in the presence or absence of 1% SDS for 1 h at room temperature. Reactions were quenched with 45 mM DTT, and proteins were precipitated with chloroform and methanol. Resolubilized samples were separated by SDS-PAGE and electroblotted, and immunoreactive bands were revealed by incubation with either anti-TatB antiserum or an anti-hemagglutinin HRP conjugate (to detect SoxY). The positions of the molecular mass markers are indicated to the left and the positions of PEGylated and non-PEGylated protein to the right. allow direct detection of labeled cysteine residues introduced at strategic positions throughout each protein by a size shift, while at the same time ensuring that the cytoplasmic membrane remained intact. Our results clearly showed that the N termini of TatA and TatB are exposed at the periplasmic side of the membrane whereas residues along the APH and within the C termini of both proteins were not labeled unless the inner membrane was dispersed with detergent. Topological changes of the TatA APH have been linked to substrate transport (31); however, even when we grossly overproduced the Tat substrate SufI, we did not detect any trace of exposure of any cysteines in the APH at the periplasmic side of the membrane. Because the cells clearly remained viable during these experiments, which were carried out at room temperature, it is reasonable to suppose that the Tat system was still functional during this time. Coupled with the observations that at least some of the TatA variants we tested were not impaired for Tat function, we conclude that Tat transport is not accompanied by exposure of the APH to the periplasm.
We also addressed the accessibility of Cys residues located around the hinge region of TatA, between the TMH and APH.
The results indicate that most of the hinge region is buried within the membrane and is not accessible to PEGylation. This is in agreement with a solid-state NMR spectroscopy study of a truncated B. subtilis TatA d construct comprising just the TMH and APH reconstituted into lipid bicelles (50). It was shown that the N-terminal part of the APH is pulled into the membrane by virtue of the fact that the TMH is unusually short. A solution NMR study of the same protein also showed that the hinge region was unexpectedly rigid rather than flexible with extensive contacts between residues Phe 14 , Leu 18 , Pro 23 , and Leu 26 (58). The rigidity of this region results in quite a steep upward tilt of the APH (50) and would be expected to constrain movement of the APH relative to the TMH. It is not known whether the hinge region of E. coli TatA is similarly rigid, particularly because most of the residues in TatA d involved in this interaction network are not conserved between the two proteins. However, although it would appear that the TMH-APH hinge region plays an essential role in TatA function, it is unlikely to act as a pivot that allows these two domains of TatA to form a helical hairpin. Future models for Tat transport should consider alternative functions for the TatA APH.