INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Besides the well-characterized Sec system, which is used for the translocation of the majority of exported proteins across the cytoplasmic membrane (8, 10, 28), another export pathway is existent in bacteria, the so-called twin-arginine translocation (Tat) pathway (for a recent review, see reference 2). There is strong evidence that, in contrast to the Sec pathway, the twin-arginine translocase exclusively exports across the cytoplasmic membrane folded proteins which, in most cases, contain a bound cofactor (17, 29-31, 41, 46). Precursor proteins that are exported via the Tat pathway possess amino-terminal signal peptides which are substantially longer than typical Sec signal peptides and contain an S/T-R-R-X-F-L-K consensus motif in their amino-terminal region (1, 2). The two arginine residues of the conserved motif are of crucial importance, and mutagenesis of one or both of these residues severely affects membrane translocation of the corresponding mutant precursor proteins (7, 9, 13, 36). Furthermore, the central hydrophobic core (h region) of Tat signal peptides is less hydrophobic than the h region of Sec signal peptides (7). In the more polar carboxy-terminal region that precedes the processing site, basic amino acid residues are frequently observed in Tat signal peptides, whereas signal peptides of the Sec pathway show a strong bias against such residues near the signal peptidase cleavage site (2, 3, 38).

Four integral cytoplasmic membrane proteins, encoded by tatA, tatB, tatC, and tatE, have been shown to be involved in the Tat translocation process in Escherichia coli, although very little is known about their function (2). The bacterial Tat pathway is closely related to the pH-dependent protein import pathway of the plant chloroplast thylakoid membrane (4, 34, 35). Their common phylogenetic origin is stressed by the fact that bacterial Tat substrates can be translocated by the pH pathway and that their signal peptides are interchangeable (14, 24, 42).

The glucose-fructose oxidoreductase (GFOR) of the gram-negative bacterium Z. mobilis exhibits the typical characteristics of a Tat substrate. The homotetrameric protein contains four tightly bound NADP molecules as a cofactor and is found in the periplasm in a soluble form (20, 21). GFOR is synthesized as a cytoplasmic precursor (pre-GFOR) with an extraordinary long signal sequence of 52 amino acid residues containing the typical twin-arginine consensus motif (43). The replacement of one or both of the arginine residues by lysine prevents export of the corresponding pre-GFOR proteins (15). Furthermore, the export kinetics of mutant forms of pre-GFOR which have substantially decreased affinities for the NADP cofactor is significantly slower than that of the wild-type enzyme, suggesting that cytoplasmic cofactor insertion and tight folding are prerequisites for Tat-dependent membrane translocation of GFOR (15). Moreover, it has been shown that pre-GFOR can be translocated in vitro into isolated plant thylakoids in a pH-dependent manner (14).

In previous experiments we have observed that pre-GFOR is not exported to the periplasm of the heterologous host E. coli, although cofactor insertion and formation of correctly folded and enzymatically active pre-GFOR take place in the cytosol (44). These results suggest that the foreign GFOR precursor protein is not recognized by the E. coli Tat machinery. Replacement of the genuine GFOR signal sequence by the OmpA signal peptide, which is a typical Sec signal peptide, results in efficient Sec-dependent export of the corresponding hybrid precursor without its cofactor and in the subsequent degradation of the translocated mature part in the periplasm by proteases (44).

In the present work, we addressed the question of why Z. mobilis pre-GFOR is not exported by the E. coli Tat pathway, despite the fact that it is an efficient Tat substrate in its original host. There are several possible explanations for the failure of pre-GFOR to be exported in E. coli: (i) E. coli may lack certain accessory protein factors that are necessary for GFOR export and that are present in pea thylakoids and Z. mobilis, (ii) the folded structure of GFOR may not be compatible with the E. coli Tat machinery, or (iii) the GFOR signal peptide may not be recognized by the E. coli Tat apparatus. Here, we show that a precise replacement of the GFOR signal peptide by an authentic E. coli Tat signal peptide is sufficient to promote the Tat-dependent export of GFOR in E. coli. These results strongly suggest that there exists a recognition event between Tat signal peptides and one or more components of the Tat export apparatus that goes beyond the recognition of the conserved general features found in all Tat signal peptides and that this recognition event seems to have a species-specific component. Our findings imply that the signal peptides of Tat-dependent precursor proteins are optimally adapted only to their cognate export apparatus, which is in marked contrast to the situation that is known for Sec-dependent protein translocation.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Bacterial strains, plasmids, and media. E. coli K-12 strain JM109 (47) was used for standard cloning procedures. E. coli strains MC4100tatAE (JARV15) and MC4100tatC (B1LK0) are derivatives of MC4100 (6) with deletions in the respective tat genes (4, 30). E. coli cells were grown aerobically in Luria-Bertani medium (23) or in mineral salts medium (37) with 0.4% glycerol as a carbon source and ampicillin at a concentration of 100 mg/liter, as required.

View larger version (13K): [in this window] [in a new window]   FIG. 1.   Signal peptide and early mature region of pre-GFOR, pre-TorA, and the TorA-GFOR fusion protein. Processing sites are indicated by inverted triangles; twin arginine residues are in boldface.

Pulse-chase experiments, preparation of spheroplasts, and tryptic digestion. Pulse-chase experiments were performed as described earlier (44). Carbonyl cyanide m-chlorophenylhydrazone (CCCP) was dissolved in ethanol at a concentration of 10 mM and added to samples where indicated to give a final concentration of 0.1 mM. In a control experiment using the same amount of ethanol without CCCP, the processing kinetics were the same as in the absence of ethanol (data not shown).

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

The signal peptide of the E. coli TMAO reductase (TorA) precursor, but not the authentic GFOR signal peptide, allows translocation of the Z. mobilis GFOR protein across the E. coli plasma membrane. Previously, we have observed that the pre-GFOR of Z. mobilis is not translocated across the E. coli plasma membrane despite the fact that the GFOR signal peptide mediates efficient translocation of the enzyme, together with its cofactor, into the periplasm of Z. mobilis (43, 44). To test whether the nature of the GFOR signal peptide is responsible for the lack of export of this protein in E. coli, we constructed a hybrid precursor protein containing the mature GFOR protein fused to an E. coli Tat signal peptide. To do so, the GFOR signal peptide was precisely replaced by the signal prepeptide of the E. coli pre-trimethylamine N-oxide (TMAO) reductase (pre-TorA), an enzyme which is known to be efficiently translocated by the E. coli Tat translocase (29) (Fig. 1). Expression of the gene encoding the TorA-GFOR hybrid protein in E. coli JM109 resulted in GFOR enzyme activities that were similar to the activities which were found when the wild-type gfo gene was expressed in E. coli (data not shown). Hence, correct folding of the GFOR protein and correct insertion of the NADP cofactor do occur in both cases. To analyze whether the signal peptide replacement has an effect on the export behavior of GFOR, pulse-chase experiments were performed. As shown in Fig. 2, lanes 1 to 4, and as described earlier (44), no processing of the wild-type pre-GFOR was observed even after a 60-min chase, confirming that the pre-GFOR was not translocated by the E. coli Tat machinery. In contrast, the TorA-GFOR fusion protein was completely processed to mature GFOR during a 20-min chase (Fig. 2, lanes 5 to 8), and this processing occurred with relatively slow kinetics, similar to the pre-GFOR export kinetics in Z. mobilis (43).

View larger version (35K): [in this window] [in a new window]   FIG. 2.   Processing of pre-GFOR and TorA-GFOR in E. coli JM109. E. coli JM109 carrying either plasmid pZY470, encoding wild-type pre-GFOR (lanes 1 to 4), or plasmid pTW42, encoding the TorA-GFOR fusion protein (lanes 5 to 8), was grown in mineral salts medium to early logarithmic phase and labeled for 1 min with [35S]methionine, after which nonradioactive methionine was added. Samples were withdrawn at chase times of 15 s and 5, 20, and 60 min and subjected to immunoprecipitation with antiserum against GFOR, followed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis and fluorography. p, precursor form; m, mature protein.

View larger version (49K): [in this window] [in a new window]   FIG. 3.   Trypsin treatment of spheroplasts. E. coli JM109 carrying either plasmid pTW42, encoding the TorA-GFOR fusion protein (A), or plasmid pZY470, encoding the wild-type pre-GFOR (B), was labeled with [35S]methionine for 1 min. After a 5-min chase, cells were converted to spheroplasts and divided into three aliquots. Trypsin was added where indicated (+) to digest periplasmic proteins. As a control, cells in one aliquot were disrupted by ultrasonication after trypsin addition. After trypsin treatment, each aliquot was divided into three parts and subjected to immunoprecipitation with antisera against GFOR (lanes 1 to 3), OmpA (lanes 4 to 6), and DnaK (lanes 7 to 9). p, precursor form of TorA-GFOR (A) or pre-GFOR (B); m, mature GFOR. Positions of tryptic fragments are indicated for GFOR (*), OmpA (+), and DnaK (#). The numbers at the right margin are positions of molecular mass markers.

Export of the TorA-GFOR fusion protein can be blocked by CCCP but not by the SecA inhibitor sodium azide. The protonophore CCCP inhibits both the Tat- and the Sec-dependent translocation pathways, showing that both processes require an intact membrane potential (7, 29, 33). In contrast, sodium azide, which severely inhibits Sec-dependent protein export by interfering with the translocation-ATPase activity of the SecA protein (27), only slightly affects Tat translocation (29). To analyze whether the observed export of the TorA-GFOR hybrid protein occurs via the Tat or the Sec pathway, sodium azide (3 mM) or CCCP (0.1 mM) was added in the pulse-chase experiments prior to the chase. As shown in Fig. 4, the processing of the TorA-GFOR fusion protein was not inhibited by sodium azide. In contrast, the processing was completely blocked by the addition of 0.1 mM CCCP. In a parallel control experiment, processing of chromosomally encoded OmpA, which is exported in a Sec-dependent manner, was inhibited both by sodium azide and by CCCP (data not shown). The insensitivity of the processing of the TorA-GFOR fusion protein to azide is typical for a Tat-dependent precursor protein, indicating that export of TorA-GFOR is mediated by the Tat translocase.

View larger version (24K): [in this window] [in a new window]   FIG. 4.   Effect of CCCP and sodium azide (NaN3) on processing of TorA-GFOR. Pulse-chase experiments were performed with E. coli JM109 carrying plasmid pTW42, encoding the TorA-GFOR fusion protein, as described in the legend to Fig. 2. Sodium azide (3 mM final concentration) (lanes 5 to 8) or CCCP (0.1 mM final concentration) (lanes 9 to 12) was added to the cultures prior to the chase. Samples were taken at the indicated chase times and submitted to immunoprecipitation with GFOR-specific antibodies. p, precursor form; m, mature protein.

Export of the TorA-GFOR fusion protein in E. coli is Tat dependent. For direct proof that membrane translocation of the TorA-GFOR fusion protein is in fact mediated by the Tat translocase, pulse-chase experiments were performed with mutant derivatives of E. coli MC4100 harboring a deletion of tatAE or tatC (4, 30) and, as a control, the isogenic wild-type strain. First of all, we noticed that processing of the TorA-GFOR fusion protein in the wild-type strain MC4100 was significantly slower than in E. coli JM109 (compare Fig. 5A, lanes 1 to 4, with Fig. 2, lanes 5 to 8); the reason for this is unknown. Nevertheless, we found that processing of the TorA-GFOR fusion protein was clearly affected in the tatAE (Fig. 5A, lanes 5 to 8) and tatC (Fig. 5A, lanes 9 to 12) mutant strains. However, a mature-sized form also appeared in the tat mutants after a 20-min chase. Quantification of the pulse-chase experiments showed that, after a 60-min chase, about 30% (tatC mutant) or 50% (tatAE mutant) of the immunoprecipitated GFOR protein was in a mature-sized form (Fig. 5B). Since we expected processing of Tat substrates to be completely inhibited in the tatAE and tatC mutants (4, 30), we thought that it might be possible that the mature-sized GFOR protein, which accumulates in the tat mutant strains, is a result of the degradation of the TorA-GFOR precursor by proteases in the cytosol, rather than being caused by export and subsequent processing of the signal peptide. If so, the respective mature-sized GFOR protein should be localized in the cytosol and therefore should be resistant to protease digestion in spheroplasts. To determine the localization of the mature-sized forms of the TorA-GFOR fusion protein in wild-type E. coli MC4100 and in the tatAE and tatC mutant strains, the corresponding cells were converted to spheroplasts after a 60-min chase and treated with trypsin as described above (Fig. 5C). In all experiments, the quality of the spheroplasts and the effectiveness of the tryptic digestions were verified again with OmpA and DnaK as controls (data not shown). Like the situation in E. coli JM109 (Fig. 3A, lanes 1 and 2), the mature-sized GFOR protein was also clearly sensitive to trypsin in spheroplasts derived from the MC4100 wild-type strain (Fig. 5C, upper panel), showing that this mature protein is localized in the periplasm and therefore is a result of membrane translocation and signal peptide processing. In contrast, the mature-sized forms which accumulate in the tatAE (Fig. 5C, middle panel) and the tatC (Fig. 5C, lower panel) mutant strains were completely resistant to tryptic digestion after spheroplast conversion of the corresponding cells, showing that these forms are localized in the cytosol. Cytosolic degradation of a TorA fusion protein in the case of blocked Tat-dependent export was also described in an earlier report; there, a TorA-LepA fusion protein was completely degraded in tat mutant strains or when the RR amino acid residues of the twin-arginine motif were mutated to KK (7). We assume that the TorA signal peptide of the hybrid TorA-GFOR protein is accessible to cytoplasmic proteolysis when interaction with Tat components is disturbed. Taken together, our results clearly show that membrane translocation of the TorA-GFOR hybrid protein is mediated by the Tat export apparatus.

View larger version (31K): [in this window] [in a new window]   FIG. 5.   Processing of TorA-GFOR in E. coli tat mutants and localization of GFOR gene products by trypsin treatment of spheroplasts. (A) Pulse-chase experiments with E. coli strains MC4100, MC4100 tatC, and MC4100 tatAE, containing plasmid pTW42, encoding the TorA-GFOR fusion protein, were performed as described in the legend to Fig. 2. (B) The bands of the gel in panel A were quantified using a PhosphorImager and the FragmeNT Analysis (version 1.1) software (Molecular Dynamics). The percentages of precursor present at the indicated chase times were calculated [p/(p + m) × 100, where p is the amount of precursor and m is the amount of mature form]. Circles, MC4100 wild type; triangles, MC4100 tatAE mutant; squares, MC4100 tatC mutant. (C) To examine the subcellular localization of the different GFOR forms in E. coli MC4100, MC4100 tatAE, and MC4100 tatC, labeled cells were converted to spheroplasts and submitted to tryptic digestion as described in the legend to Fig. 3, with the exception that, due to the slower processing of TorA-GFOR in E. coli MC4100, cells were converted to spheroplasts after a 60-min chase. wt, wild type; p, precursor; m, mature GFOR; m', mature-sized cytosolic GFOR fragment; *, tryptic fragment of GFOR.

Tight binding of the NADP cofactor is essential for membrane translocation of the TorA-GFOR fusion protein. Previously, we have described various mutant derivatives of the GFOR protein which contain an alteration in one or more amino acid residues located in the NADP-binding Rossman fold and which are impaired in tight binding of the NADP cofactor (45). In addition, we demonstrated that these GFOR mutant proteins with reduced affinity for NADP lost the ability to oxidize glucose and reduce fructose but instead were able to exchange reduced cofactor (NADPH) for the oxidized form (NADP), therefore acting as glucose dehydrogenases (45). Since these mutant forms of GFOR still exhibit enzymatic activity, the overall three-dimensional structure should be intact. Interestingly, these mutant forms were not or were only very slowly processed when examined in pulse-chase experiments with Z. mobilis (15). This led us to the hypothesis that the Z. mobilis Tat pathway is able to recognize proper cofactor binding and folding prior to protein export. To test whether the Tat-dependent export of the TorA-GFOR fusion protein in E. coli also requires tight NADP binding, the point mutations S116D, K121A, K123Q, and I124K were introduced into the TorA-GFOR hybrid protein (plasmid pTW43). In pulse-chase experiments with E. coli MC4100, the TorA-GFOR mutant protein was nearly completely degraded during a 60-min chase (Fig. 6). Furthermore, no processed mature form accumulated during the chase, which is in sharp contrast to what was observed with the TorA-GFOR wild-type protein (compare Fig. 6, lanes 1 to 4, with Fig. 5A, lanes 1 to 4). Likewise, in the faster-processing E. coli strain JM109 as well, no processing of the TorA-GFOR cofactor-binding mutant protein to the mature form was detected in pulse-chase experiments and, also in this case, the mutant protein was degraded (data not shown). An identical behavior was found when the TorA-GFOR mutant protein was expressed in the tatC deletion strain (Fig. 6, lanes 5 to 8), showing that the observed degradation most probably had occurred in the cytosol and that point mutations S116D, K121A, K123Q, and I124K in the NADP-binding Rossman fold of GFOR, also in E. coli, severely impairs the Tat-dependent translocation. These results support the current view that correct folding and cofactor insertion are generally an important prerequisite for export of cofactor-containing proteins via the Tat pathway.

View larger version (27K): [in this window] [in a new window]   FIG. 6.   Processing of a cofactor-binding-defective TorA-GFOR fusion protein. Pulse-chase experiments were performed, as described in the legend to Fig. 2, with E. coli strains MC4100 (lanes 1 to 4) and MC4100tatC (lanes 5 to 8) carrying plasmid pTW43, encoding a mutant form of TorA-GFOR with point mutations in the NADP binding site that result in decreased cofactor binding affinity. wt, wild type; p, precursor form.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

In the present work, we have shown that a precise replacement of the authentic signal peptide of the Z. mobilis GFOR precursor protein with a signal peptide derived from the E. coli Tat substrate TorA is sufficient to promote export of GFOR to the E. coli periplasm and that this export is mediated by the Tat translocase. These results clearly demonstrate that the mature part of GFOR is compatible with translocation by the E. coli Tat export apparatus and that the previously described export defect of wild-type pre-GFOR in E. coli is due to an incompatibility of the GFOR signal peptide with Tat-dependent protein translocation in E. coli.

What is the nature of such an incompatibility? For cofactor-containing Tat substrates, it was proposed that their signal peptides might have distinct structural features, allowing specific protein-protein interactions with the mature protein and/or assembly factors (2), which ensure cofactor binding prior to export. In such a model, the signal peptide is sheltered either by the mature part of the apoprotein or by a special accessory protein until cofactor binding takes place (29). Because enzymatically active pre-GFOR is formed in the E. coli cytoplasm, the possibility that the GFOR signal peptide is not accessible to the Tat translocon due to improper folding of the mature part of GFOR or problems with cofactor incorporation can be excluded.

The GFOR signal peptide shows characteristics of a typical twin-arginine signal peptide. Although the twin-arginine motif T-R-R-A-L-V-G does not completely match the motif with the highest consensus of Tat signal peptides (S/T-R-R-X-F-L-K) (1), alignments of Tat signal sequences show that the F-L-K residues are more variable than the invariant R-R residues and that L-V-G residues at respective positions can be found in other Tat signal sequences (1, 7). Moreover, it was most recently shown that the F residue of the consensus (which is L in the GFOR signal peptide) can be functionally replaced by an L and that the K residue in the consensus (which is a G in the GFOR signal peptide) even retards Tat transport (36). In addition, the h region of the GFOR signal peptide, which contains several glycine residues, is less hydrophobic than the corresponding region of typical Sec signal peptides, which is another critical determinant for efficient Tat-dependent protein translocation (7). Thus, the export defect of pre-GFOR in E. coli cannot be explained by differences in the Tat signal peptide consensus features alone, since exactly the same signal peptide mediates efficient Tat-dependent export in the original host, Z. mobilis (43).

One possible explanation for the lack of export of pre-GFOR in E. coli is that there exist specific recognition events between Tat signal peptides and one or more components of the Tat translocase (or some as-yet-unrecognized factors) that involve more than just the recognition of the generally conserved features in the Tat signal peptides. This would mean that Tat signal peptides optimally interact only with the Tat translocase of the same organism, whereas interactions of a certain Tat precursor with the Tat translocase of a heterologous host organism might not occur or might be nonproductive. In light of this, it is most intriguing that an efficient in vitro translocation of pre-GFOR into isolated plant thylakoids via the pH pathway is possible (14); this may be due to a lesser stringency of the pH translocon for foreign signal peptides.

In contrast to the situation observed with the pre-GFOR protein, it is known that signal peptides are, in general, interchangeable in the Sec protein translocation pathway (18, 39). Signal sequences of gram-negative bacteria are recognized by Sec translocons of gram-positive bacteria and vice versa (22, 25, 32), and even eukaryotic signal peptides of the Sec-related pathway for protein import into the endoplasmatic reticulum and bacterial Sec signal peptides are interchangeable (12, 40). The observed species specificity of pre-GFOR export via the Tat pathway is therefore in marked contrast to the general Sec pathway.

A similar specificity of signal peptide recognition was observed when expression of the thylakoidal Tat substrate pre-23K in E. coli was examined. Whereas the pre-23K protein containing its authentic signal peptide was not exported in E. coli (16), replacement of that signal peptide with the signal peptide of the E. coli TorA protein resulted in Tat-dependent membrane translocation (30). However, it should be mentioned that there are also known cases in which a Tat signal peptide is functional in a heterologous host. Fusion of mature ß-lactamase to the signal peptide of the Tat substrate protein [NiFe]-hydrogenase of Desulfovibrio vulgaris Hildenborough allowed significant translocation of the ß-lactamase into the E. coli periplasm (26).

Two alternative scenarios can be imagined to be responsible for the lack of export of pre-GFOR in E. coli. First, specific recognition of the signal peptide or keeping the GFOR precursor in an export-competent state might require an as-yet-unidentified factor which is present in Z. mobilis but lacking in E. coli. Second, it cannot be excluded that an unknown E. coli protein or nonproteinaceous component unspecifically interacts with the GFOR signal peptide and, despite correct folding of the mature protein and successful cofactor binding, prevents the pre-GFOR from entering the Tat secretion pathway.

Binding of NADP is necessary to stabilize the GFOR quaternary structure in which the NADP binding pocket of one subunit is covered by the amino-terminal arm of another subunit (19). When directed to the Sec pathway by replacement of the GFOR signal peptide with the OmpA signal sequence, the OmpA-GFOR fusion protein is exported, processed to mature GFOR, and then very quickly degraded by periplasmic proteases (44). In contrast, we have now shown that GFOR, which is translocated by the Tat pathway with its bound cofactor, is completely stable during a 60-min chase period. Therefore, the previously observed proteolytic instability of GFOR exported by the Sec pathway is due to the lack of NADP cofactor, which is not present in the E. coli periplasm and which cannot be cotranslocated with the apoprotein, which, in the case of translocation by the Sec machinery, is threaded through the membrane in a more or less unfolded form.

In contrast to the cofactor-binding-proficient TorA-GFOR fusion protein, the TorA-GFOR S116D/K121A/K123Q/I124K fusion protein with mutated amino acid residues in the NADP binding pocket was not processed to a mature form but instead was almost completely degraded during a 60-min chase period in a wild-type strain as well as in a tatC mutant. These results further strengthen the view that correct folding, which in the case of cofactor-containing proteins requires cofactor insertion or, in other words, the absence of unfolded structures, seems to be an absolute requirement for Tat-dependent protein translocation.