Essential Cytoplasmic Domains in the Escherichia coli TatC Protein*

The twin-arginine translocation (Tat) system mediates the transport of proteins across the bacterial plasma membrane and chloroplast thylakoid membrane. Operating in parallel with Sec-type systems in these membranes, the Tat system is completely different in both structural and mechanistic terms, and is uniquely able to catalyze the translocation of fully folded proteins across coupled membranes. TatC is an essential, multispanning component that has been proposed to form part of the binding site for substrate precursor proteins. In this study we have tested the importance of conserved residues on the periplasmic and cytoplasmic face of the Escherichia coliprotein. We find that many of the mutations on the cytoplasmic face have little or no effect. However, substitution at several positions in the extreme N-terminal cytoplasmic region or the predicted first cytoplasmic loop lead to a significant or complete loss of Tat-dependent export. The mutated strains are unable to grow anaerobically on trimethylamine N-oxide minimal media and are unable to export trimethylamine-N-oxide reductase (TorA). The same mutants are completely unable to export a chimeric protein, comprising the TorA signal peptide linked to green fluorescent protein, indicating that translocation is blocked rather than cofactor insertion into the TorA mature protein. The data point to two essential cytoplasmic domains on the TatC protein that are essential for export.

The twin-arginine translocation (Tat) 1 system transports proteins across the chloroplast thylakoid membrane and the plasma membranes of a wide range of bacteria (reviewed in Refs. 1 and 2). Working in parallel with the well characterized Sec pathway, the Tat system uses an entirely different mechanism and has the unique ability to transport fully folded proteins across coupled membranes (3,4). In general, it appears to be used primarily for the transport of proteins that are either obliged to fold prior to translocation, or which fold too tightly or rapidly for the Sec system to handle. Prime examples of substrates in Escherichia coli are a range of periplasmic proteins that acquire redox cofactors in the cytoplasm before export, such as FeS and molybdopterin centers (5,6). These proteins must fold to a large extent in the cytoplasm, preclud-ing export by the Sec machinery which is known to translocate proteins in a largely unfolded state (reviewed in Ref. 7). In both bacteria and thylakoids, the Tat system recognizes substrates bearing N-terminal signal peptides containing an invariant twin-arginine motif (8,9).
Studies on plant (10 -12) and E. coli mutants (13)(14)(15)(16) have led to the identification of several genes encoding Tat components. Four tat genes have been characterized in E. coli, three of which (tatABC) form an operon while the fourth gene (tatE) is monocistronic. Knockout studies have shown that Tat-dependent export requires minimally TatB (13,14), TatC (15), and either TatA or TatE (14,16). Of the latter two components, TatA is more important than TatE, possibly due to a much greater abundance (16). Additional tat genes have not been identifed to date and a complex of 600 kDa comprising only TatA, -B, and -C has been recently purified from solubilized E. coli membranes (17). However, the activity of this complex has yet to be tested and the existence of further Tat subunits cannot be ruled out at this stage.
Sequence analyses and biochemical analyses of the Tat subunits indicate that TatA, -B, and -E contain a single transmembrane span with a small C-terminal domain exposed on the cis face of the membrane (i.e. the cytoplasm in bacteria, or stroma in chloroplasts (11)), whereas TatC is predicted to contain 6 transmembrane regions with the N and C termini localized in the cytoplasm. There is convincing evidence that the N terminus of TatC is indeed in the cytoplasm since a TatBC fusion protein is active (17) and the N terminus of the TatC domain is linked to the C terminus of TatB in this construct. The Nterminal domain of chloroplastic TatC is, furthermore, exposed on the stromal face of the thylakoid membrane (18).
We presently know little about the mechanism of the Tat system but the first details on the organization of the complex are starting to emerge. In E. coli, it was shown that TatB and TatC form a structural and functional unit in a 1:1 ratio within the Tat complex, and a translational fusion between the proteins was shown to be fully functional (17). Studies on the thylakoidal homologs of these proteins, Hcf106 and cpTatC (19), strongly suggest that these proteins likewise function together and it was also shown that these proteins are very likely to form the initial binding site for the docking of Tat-dependent precursor proteins. However, the important elements of the TatC protein have yet to be defined and in this study we have addressed this point through a mutagenesis study. We show that two cytoplasmic domains are highly important for the function of E. coli TatC.
Site-specific mutagenesis was used to generate a vector that encoded the tat operon within pBAD-ABC with point mutations in the tatC gene, using the QuikChange TM mutagenesis system (Strategene) according to the manufacturer's instructions. For studies on the effects of these mutations on green fluorescent protein (GFP) export, these tatABC sequences were removed from pBAD-ABC using NheI and XbaI, and cloned into the pEXT22 vector (25) using the XbaI site, generating pEXT-ABC and mutant derivatives which are compatible with pJDT1 encoding TorA-GFP (see text below).
Cell Fractionations-Cells were grown aerobically (in lsLB) or anaerobically (in lsLB-GT) in medium in the presence or absence of IPTG or arabinose (as stated in the text), and periplasm and spheroplasts were prepared by the EDTA/lysozyme/cold osmoshock procedure (26). Spheroplasts were lysed by sonication, and intact cells and cellular debris were removed by centrifugation (5 min at 10,000 ϫ g). Membranes were separated from the cytoplasmic fraction by centrifugation (30 min at 25,0000 ϫ g). Protein concentration was determined using a BCA-linked assay (Pierce). Protein fractions were either separated on a 10% nondenaturing polyacrylamide gel and analyzed for TorA activity (6) or were separated by SDS-polyacrylamide gel electrophoresis and immunoblotted with specific antibodies to GFP (CLONTECH) and horseradish peroxidase anti-rabbit IgG conjugates, using the ECL detection system (Amersham Biosciences Inc.). Fig. 1, with the predicted membranespanning regions shaded and identical residues denoted by asterisks. The figure also illustrates 9 residues (also shaded) that are conserved in at least 30 of the 31 bacterial and plastid TatC sequences available in the data base. The predicted topology of the protein is illustrated in Fig. 2. Given that chloroplast TatC appears to play a role in binding of precursor proteins, we reasoned that a cytoplasmic domain(s) in TatC may mediate this process in E. coli.

Structures of TatC Proteins-An alignment of several bacterial TatC proteins and the chloroplast TatC from Arabidopsis thaliana is shown in
TatC proteins are not particularly highly conserved but analysis of Fig. 1 shows that two cytoplasmically exposed loops contain a number of highly conserved residues, whereas periplasmic loops contain few invariant residues. A few conserved residues are present in the transmembrane spans, but these were not examined in this study. The extreme cytoplasmic N-terminal region and the (predicted) first cytoplasmic loop, in particular, contain a range of conserved charged residues; these are indicated as cytoplasmic domains 1 and 2 (CD1 and CD2) in Fig. 1. These are more highly conserved than is immediately apparent because Arg or Lys is present at the positions corresponding to Arg 104 and Arg 105 in every other bacterial/plastid TatC sequence, with the exception of Odontella sinensis and Staphylococcus aureus, which lack one of the basic residues in these positions. Plant mitochondrial TatC proteins likewise contain only a single basic residue at these positions in some cases. Arg is present at the position corresponding to Arg 17 in all but four of the 31 TatC sequences, where Lys is present instead.
The significance of these residues was tested using sitespecific mutagenesis. As detailed under "Experimental Procedures," we mutated the tatC gene within the tatABC operon in the arabinose-inducible pBAD24 plasmid (plasmid pBAD-ABC), and expressed this vector in a ⌬tatABCDE background. The TatC protein encoded by this plasmid contains a C-terminal Strep II tag which does not affect activity, and expression of pBAD-ABC in this background leads to efficient Tat-dependent export (22). The targeted residues are shown in bold in the alignment in Fig. 1 and their predicted positions are illustrated in Fig. 2. Most of the mutations were to Ala but Glu 15 and Glu 103 were changed to Gln. The reason for this strategy was that these residues were considered as candidates for binding the positively charged twin-arginine signal peptide of TorA, and we sought to remove the negative charge but retain the side chain properties as far as possible.
The mutant TatC proteins were expressed after induction with arabinose and their levels were analyzed by immunoblotting using antibodies to TatA, TatB, and the Strep II tag on TatC. Fig. 3 shows a representative immunoblot of cell extracts from approximately half of the mutants, which illustrates that the various strains all contain similar levels of Tat components,  FIG. 2. Predicted topology of E. coli TatC. The diagram shows the 6-span topology of the E. coli TatC protein predicted using the TopPred II algorithm (32) and the positions of the residues targeted for analysis in this study. Shaded residues represent those that exhibit major defects in Tat-dependent export when mutated as described in this study.
indicating that the tatABC operon is induced to similar extents. Similar findings were made using the other mutants (data not shown). When induced in this manner, the pBAD24 system leads to an about 10-fold overproduction of TatABC (17) which means that the Tat components are effectively present in excess. The effects of the mutations on Tat-dependent export were assessed using several criteria as explained below.
Mutations in TatC That Block Anaerobic Growth on TMAO Minimal Media-A standard means of assaying for Tat-dependent export in E. coli is to test for anaerobic growth on TMAO minimal media (11)(12)(13)(14). The periplasmic molybdopterin-containing protein TMAO-reductase (TorA) is a known Tat substrate (6,11) and tat mutants are accordingly unable to grow using TMAO as an electron acceptor during anaerobic respiration. It should be emphasized, however, that this assay selects only for a complete lack of functional Tat system. Null mutations in tatAE, tatB, or tatC are completely unable to grow but we found that the pBAD-ABC plasmid in the ⌬tatABCDE strain supported anaerobic growth even without induction, when levels of TatABC were almost undetectable (17,22). Clearly, even very low levels of the Tat apparatus are sufficient to support growth on TMAO minimal medium. Table I shows the results of this analysis, together with a summary of the export capabilities of these strains (see below). In fact, our results show that very few of these highly conserved residues are essential for Tat function, and the majority of these strains grow at normal rates. However, substitution of either Arg 17 led to a complete loss of growth, suggesting an important function of this (predicted) cytoplasmically exposed region as shown in Fig. 2. In view of the apparent importance of the CD1 region we tested the effects of deleting Leu 20 to Asn 22 , and this mutant (⌬20 -22) was again completely unable to grow under these conditions.
Perhaps surprisingly, a series of other highly conserved residues are not essential for Tat function. His 12 , Glu 15 , and Arg 19 , for example, are all present in cytoplasmic domain 1 and are essentially invariant throughout bacterial and plastid TatC sequences, yet their substitution by Ala (or Gln in the case of Glu 15 ) has no detectable effect. In cytoplasmic domain 2, while substitution of Arg 104 does inhibit growth, substitution of Arg 105 by Ala has no detectable effect on growth even though a basic residue is present at this position in the vast majority of bacterial and plastid TatC sequences (although this mutant is affected in export of torA-GFP; see below). Even more surprisingly, Pro 97 is one of the few invariant residues but its substitution again fails to impair anaerobic growth. Relatively few residues on the periplasmic side were targeted for mutagenesis in this study, but it is notable that substitution of Pro 48 blocks growth entirely, suggesting an important function for this residue.
Several tatC Mutants Are Unable to Export Either TorA or TorA-GFP-To analyze more directly the consequences of the tatC mutations, we next analyzed the distribution of a known Tat substrate, TorA, in cells expressing the various mutant forms. This was achieved using native gels in which the TorA activity is visualized directly in the polyacrylamide gel, using a methyl viologen-linked assay. Fig. 4 shows that the vast majority of active TorA is localized in the periplasm in wild-type cells expressing the pBAD-ABC vector (pBAD-ABC panel) as shown previously (22). The results obtained with the mutant strains, in general, agree with the data from the anaerobic growth analysis. Those tatC mutants that fail to grow anaerobically (R17A, ⌬20 -22, and P48A) are also unable to export TorA and the activity is found almost exclusively in the cytoplasmic fraction. The cytoplasmic TorA has a reduced mobility in this gel system (denoted TorA*), possibly due to the presence of the presequence but more likely due to other effects on folding or binding to other factors, since the presequence is small compared with the 86-kDa mature protein. In the case of R17A, no activity whatsoever is evident in the periplasmic fraction, confirming the central importance of this residue. With ⌬20 -22, a very low proportion of the TorA activity is apparent in the periplasmic fraction but the vast majority of TorA is again present as the cytoplasmic species.
In general, this assay should be more sensitive than the anaerobic growth test and Tat-dependent export defects are indeed more apparent in other cases. The L20A mutant, in particular, exhibited slow growth on TMAO miminal media but Fig. 4 shows that the TorA export is in fact drastically affected with most of the activity present in the cytoplasm. Accumulation of the cytoplasmic TorA* species is also evident with two other mutants, L16A and R105A. Both residues are in the conserved cytoplasmic domains and, in the case of R105A as discussed above, a charged residue is highly conserved among bacterial TatC proteins. These mutations do, therefore, affect TatC function.
As a final test we examined the export of a chimeric protein comprising the presequence of TorA linked to GFP. Previous   Table details the effects of the TatC mutations on anaerobic  growth on TMAO minimal media, and summarizes the data shown in  Figs. 4 and 5 on the localization of TorA and the export of TorA-GFP. Export/growth capabilities are shown on a scale of "none" to ϩϩϩϩ (wild-type levels) and ϩ/ϩϩ/ϩϩϩ represent export abilities that approximate to 25/50/75% of wild-type.

Mutant
Anaerobic growth TorA export GFP export ϩϩϩϩ ϩϩϩϩ ϩϩϩ studies (23,28) have shown this construct to be exported exclusively by the Tat pathway and GFP represents a much simpler export substrate than TorA, because it does not acquire an additional cofactor before export. Another important point is that the export of this protein is examined over a much shorter time scale. One disadvantage of the TorA assay is that only steady-state levels can be analyzed, and this can mean that defects in the TatC operation can be masked because the cells have sufficient time to export the protein even when the Tat system is functioning suboptimally. In contrast, the TorA-GFP is rapidly induced at fairly high levels, after which the Tat system is expressed and the cells are fractionated within a few hours. Under these conditions, effects on export rates are more likely to be apparent. (Ideally, the GFP would be analyzed using pulse-chase techniques which give a much better picture of the export kinetics. However, we have been unable to apply this technique to GFP export because the immunoprecipitation is ineffective, for unknown reasons.) To examine the export of this construct, we used a growth regime in which expression of the TorA-GFP was induced by arabinose, after which the ⌬tatAB-CDE cells were washed and incubated with IPTG to induce expression of the tatABC operon from the pEXT22 plasmid. The localization of the GFP was monitored by immunoblotting. Fig. 5 shows that expression of the wild-type tatABC operon (pEXT-ABC panel) results in export of the TorA-GFP and the vast majority of the GFP is found in the periplasmic fraction (lane P) as the mature form. Essentially no TorA-GFP is apparent in the membrane fraction (M) and very little maturesize GFP is apparent in the cytoplasmic fraction (C). Previous work on the export of this construct (23) also found that the cytoplasmic GFP was primarily mature-size, presumably due to proteolysis of the signal peptide. Otherwise, the results obtained using the tatC mutants closely resemble those obtained from the TorA analysis. Export of TorA-GFP in the R17A, ⌬20 -22, and P48A mutants is drastically affected and the majority of GFP is found in the cytoplasmic and membrane fractions. Export is also affected in the R104A and R105A mutants, confirming the importance of this conserved domain. The one surprise is L20A, which exports torA-GFP with high efficiency, yet which fails to export TorA except at low levels. The reason for this finding is currently under study. These data are summarized in Table I. SUMMARY In this report we have aimed to provide a first dissection of the E. coli TatC protein to begin mapping the important regions. The TatC family is in fact remarkably poorly conserved in terms of primary sequence, and the actual sequence data provide very few clues as to its role in Tat-dependent protein export. This lack of sequence conservation is all the more surprising given the highly conserved nature of the translocation mechanism as well as the targeting signals that are recognized by this system. For example, chloroplast thylakoids can recognize and import E. coli Tat substrates as efficiently as those of their cognate substrates (29,30). While the precise role of TatC still requires detailed study, the available data indicate that (a) it functions together with TatB (or Hcf106 in chloroplasts) and (b) it may form part of the initial binding site for precursor proteins (17,19).
Relatively few TatC residues are invariant even among eubacteria and in this study we have mutated the majority of conserved residues located in either the cytoplasmic or periplasmic loop regions. In fact, many of these mutations have no drastic effect on TatC function, although it must be stressed FIG. 4. Localization of TMAO reductase activity in cells expressing TatC mutants. E. coli MC4100 cells containing the pBAD24 vector, or ⌬tatABCDE cells expressing mutations in the tatC within pBAD-ABC were grown anaerobically and then fractionated to generate cytoplasmic and periplasmic fractions (C and P) as detailed under "Experimental Procedures." The samples were analyzed by native gel electrophoresis and TMAO reductase activity was visualized in the gel. Mature-size TorA is indicated and TorA* denotes lower-mobility cytoplasmic form.
FIG. 5. Effects of tatC mutations on export of TorA-GFP. All strains contained plasmid pJDT1 encoding TorA-GFP in the pBAD24 vector (23). As a control, wild-type MC4100 cells contained in addition plasmid pEXT-ABC (encoding the wild-type tatABC operon; panel pEXT-ABC) and the remaining panels represent ⌬tatABCDE cells containing the same vector with mutated tatC genes. Expression of TorA-GFP was induced using arabinose, after which expression of TatABC was induced using IPTG as detailed under "Experimental Procedures." Cells were fractionated to yield cytoplasmic (C), membrane (M), and periplasmic (P) samples and immunoblotted using antibodies to GFP. Mature-size GFP and the TorA-GFP precursor protein are indicated. that kinetic analyses on export rates may reveal minor effects that are not evident from the types of analysis used here. However, it is notable that a high proportion of these highly conserved residues are located in the two cytoplasmic domains described in this report, and we have now shown that mutations in several of these residues lead to an absolute block in Tat function. None of the mutations affect Tat subunit stability to any detectable extent, and we therefore propose that these data point to the presence of two essential cytoplasmic domains in the TatC protein. Of course, it is possible that these two areas of the protein function together as a single functional domain. We emphasize that the data do not indicate how these domains function and, while they may well form a critical binding site for incoming precursor proteins, it is also possible that these mutations destabilize the TatABC complex and hence affect translocation in a more indirect manner.
These data now pave the way for more detailed studies on the roles of the individual domains in TatC function. A recent report (31) has described an in vitro assay for Tat-dependent import into inverted E. coli inner membrane vesicles, and studies on the thylakoid system (19) have shown that Tat substrates can bind to the Tat complex under appropriate conditions. It should therefore be possible to use such techniques to determine in detail the function of individual residues and domains in TatC, in terms of either substrate binding or the subsequent translocation mechanism.