The twin arginine system rescues the transport of folded Sec substrates

In Gram-negative bacteria two systems are responsible for the post-translational translocation of proteins across the inner membrane. The general secretory (Sec) system transports unfolded proteins, whereas the twin arginine translocation (Tat) system transports folded proteins. Substrates of both systems are recognized via similar N-terminal signal sequences. Co-translational recognition of the signal sequence is thought to be an essential determinant in the sorting of transport substrates en route to the transport machineries. We showed that either delay or block in recognition of the Sec-specific signal sequence in Outer Membrane Protein A (ssOmpA) allowed folding of the passenger proteins Green Fluorescent Protein (GFP) and cytochrome c used in our study. Their folded state rendered them incompatible with the Sec pathway. Surprisingly, transport of these folded Sec substrates was rescued by the Tat pathway. This enabled us to re-evaluate the role of co-translational recognition in making the sorting decision. Introduction Of all proteins that are synthesized in the cytoplasm of Escherichia coli, about 20% are marked for export out of the cell via the presence of N-terminal targeting signals (1). Like most other Gram-negative bacteria, E. coli contains two different machineries for the posttranslational translocation of proteins across the inner membrane into the periplasm (reviewed in (2)). One is the well-studied general secretory (Sec) system. The important intermediate steps between synthesis of a Sec substrate and its translocation into the periplasm are known. The signal sequence is recognized co-translationally by the ribosomal protein L23 and the ribosomebound chaperone, trigger factor (3, 4). Recognition presumably promotes binding of SecB to the nascent chain, which prevents its folding (5). Subsequently, the SecB-substrate complex binds to the motor protein SecA, which docks onto a narrow pore in the membrane formed by SecY, SecE and SecG. While hydrolyzing ATP, SecA threads the substrate through the pore (6). Finally leader peptidase, an enzyme active at the periplasmic side of the inner membrane, cleaves the substrate at a recognition site in between the signal sequence and the mature domain (7). As a result the mature domain is released into the periplasm, where it folds into its active conformation. The second system that transports proteins into the periplasm of Gram-negative bacteria is the Tat system (Twin Arginine Translocation system, reviewed in (8)). The Tat system was first discovered in plant chloroplasts, where it transports fully folded proteins into the thylakoid lumen in a ∆pH-dependent manner (9, 10). Later a related apparatus was discovered in bacteria (11), which transports folded proteins into the periplasm. Compared to the Sec system relatively little is known about the working mechanism of the Tat system. E. coli’s tat operon encodes the proteins TatA, TatB, and TatC. A gene duplicate of tatA located elsewhere on the genome encodes TatE. However, under normal growth conditions TatE is Rescue of folded Sec substrates 57 hardly expressed. An active Tat system in E. coli is therefore considered to consist of TatA, TatB and TatC. TatB and TatC are usually purified in complexes containing several copies of both proteins in a 1:1 ratio (12, 13). Such TatBC complexes were shown to bind specifically to the signal sequences of Tat substrates (9, 14), a process assumed to initiate transport. Some organisms, for example Bacillus subtilis, do not contain a TatB homologue (15). In these organisms the substrate recognition function is probably performed by TatC alone. Substrate binding to TatBC is proposed to induce association of TatA subunits, perhaps in a proton-motive force dependent manner (14, 16). The TatA subunits are supposed to open up a pore in the membrane that is large enough to facilitate the passage of a folded protein. The presence of two systems that both transport proteins into the periplasm raises the question: how do cells sort substrates between the two systems? Signal sequences are considered important in providing pathway specificity to substrates. For example, several mutations in signal sequences have been reported that turn Tat-substrates into Sec substrates (17). Sec and Tat signal sequences are remarkably similar. Both consist of a tripartite structure: a positively charged N-region is followed by a hydrophobic H-region, which in its turn is followed by a short polar C-region that contains the leader peptidase cleavage site motif (18). Despite their overall similarity there are four clear differences between signal sequences of Sec and Tat substrates (18): (i) the N-regions of Tat signals are usually longer; (ii) a consensus S/TRRxFLK motif is found at the boundary between their Nand H-regions, in which the two name-giving arginines are conserved; (iii) the H-regions of Tat signals are less hydrophobic than the ones of Sec signals; (iv) the C-regions of Tat signals usually contain at least one positively charged amino acid, which is sometimes referred to as “Sec-avoidance motif” (17, 19). The question remains where, when and how Sec substrates are sorted from Tat substrates. It has been proposed that this sorting occurs co-translationally (8). The study described here aims to gain more insight into how and where Tat substrates are discriminated from Sec substrates. In order to answer this question we intervened putative co-translational sorting processes by fusing a cleavable intein N-terminally to the signal sequence of model substrates, Green Fluorescent Protein (GFP) and cytochrome c. Gradually, in the cytoplasm intein gets cleaved and the model substrate with intact signal peptide is released. We investigated the translocation of the model substrates as a function of the presence of an intact Tat system. With cytochrome c we also varied the folding state of the substrate. We show here that the Tat system rescues the transport of folded Sec substrates, and we present evidence that the sorting between Sec and Tat substrates can be performed at a late post-translational stage, probably upon interaction with the SecYEG pore. Materials and methods Plasmids and bacterial strains used: The signal sequence of Outer membrane protein A (ssOmpA) was fused to the N-terminus of GFP and to the N-terminus of cytochrome c7A as described (20). The tac promoter region was excised from vector pMal (New England BioLabs) by PCR and cloned upstream of the protein expressing regions. Intein capped versions of both constructs were generated by fusing an engineered intein with a chitin-binding domain (CBD) from pTYB11 (New England BioLabs) to the N-terminus of ssOmpA or ssTorA. The plasmids were transformed into wild type strain MC4100 and into Δtat mutant strain DADE (21). Cell growth and expression of hybrid proteins: Cells expressing GFP hybrids were grown to reach 0.6 OD in YT medium at 37°C. Cells expressing cytochrome c hybrids were grown to reach 0.6 OD in 2YT medium at 30°C. Leaky expression of the fusion proteins was sufficient for their detection and therefore expression was not induced. Cell fractionation: Cells harvested from 25ml culture volume were suspended in 0.5 ml periplasting buffer (200 mM Tris-HCl pH 7.5, 20% sucrose, 1 mM EDTA and 30 U/l lysozyme) for 5 minutes. Upon centrifugation, the supernatant containing the periplasmic extract was separated from the pellet containing spheroplasts. The pellet was resuspended in lysis buffer (10 mM Tris-HCl pH 7.5, 50 mM KCl, 1 mM EDTA and 1% triton) to constitute the cytoplasmic extract, including the cytoplasmic membrane. Fluorescence measurements: Fluorescence intensities of GFP-expressing cells and of cell fractions were recorded at 25 °C in 1 ml cuvettes using a Cary Eclipse fluorescence spectrometer (Varian). Excitation was at 480 nm with a 5 nm slit, emission was recorded between 500 and 600 nm. Immuno-detection of proteins: Freshly made spheroplasmic and periplasmic samples were heat denatured (95°C for 10 min) in loading buffer without any reducing agent and analyzed via 12% SDS-PAGE. The proteins in the gel were transferred to a nitrocellulose membrane, and the cellular localization of GFP or intein was analyzed by immunoblotting, using appropriate antibodies anti-GFP (Biotrend) for GFP and anti-CBD (New England BioLabs) for intein. Heme-staining: Samples were dot-blotted directly onto the membrane. The membranes were stained for peroxydase activity, a characteristic of correctly matured cytochrome c, by applying the ECL kit (Amersham biosciences) as described by Vargas et al. (22). Rescue of folded Sec substrates 59 Results N-terminal masking circumvents co-translational suppression of preprotein folding. The gene encoding GFP was cloned behind the signal sequence of the Sec substrate Outer membrane protein A (OmpA). Upon expression of ssOmpA-GFP in E. coli strain MC4100, GFP fluorescence could hardly be detected (Figure 1). This demonstrates the absence of correctly folded GFP molecules. However, western blot analysis showed that non-fluorescent, i.e. unfolded or misfolded GFP was present in the cytoplasm and in the periplasm (not shown). These results are completely consistent with the ones of Feilmeier et al. (23), who showed that GFP fused behind a Sec-specific signal sequence is transported into the periplasm, where the conditions are such that complete folding and maturation of GFP is prevented. Subsequently, we fused an intein moiety N-terminally to ssOmpA-GFP. Cells expressing intein-ssOmpA-GFP were clearly fluorescent (Figure 1), which implies correct GFP folding (24). This observation shows that in order to be recognized the signal peptide of a Sec substrate needs to be present at the very N-terminus of the nascent polypeptide chain. We assume that in the absence of intein ,the N-terminal signal sequence of ssOmpA-GFP is recognized by ribosomal protein L23 and Trigger Factor, and that this recognition leads to SecB binding. Obviously, N-terminal presence of Intein delayed or blocked co-translational recognition of the adjacent Sec-signal and as a result the folding suppression, which is inherent to the Sec pathway, was circumvented. Figure 1. N-terminal masking enables folding of ssOmpA-GFP. Fluorescence emission spectra of E. coli MC4100 cells expressing ssOmpA-GFP (dashed line) or intein-ssOmpA-GFP (solid line). Tat transports folded GFP with a Sec-signal. An intein is a segment of a protein that is able to excise itself and rejoin the remaining portions (the exteins) with a peptide bond. Here, we employed an intein that has been engineered to be used as a protein purification tag. Upon reduction the intein cleaves itself exactly at its Cterminus, thus releasing the protein of interest without any additional amino acids. Conditions inside E. coli cytoplasm are apparently sufficiently reducing to induce self-cleavage of a significant fraction of intein fusion proteins (25, 26). In this case, spontaneous intein cleavage released GFP with an N-terminal Sec specific signal sequence into the cytoplasm. We fractionated MC4100 cells expressing intein-ssOmpA-GFP and measured GFP fluorescence separately in the respective fractions. Surprisingly, a significant amount of fluorescence was detected in the periplasmic extracts (Figure 2a). A western blot using an anti-GFP antibody showed that GFP was present in the cytoplasm mainly as a 90-kDa protein (Figure 2b), which corresponds to the full-length intein-ssOmpA-GFP construct. In contrast, GFP was found in the periplasm with a size of only 30 kDa, which corresponds to the mass of GFP without intein or signal sequence. Furthermore, a western blot against the intein moiety detected two bands in the cytoplasmic fraction, with sizes of 55 and 90 kDa (Figure 2c).This corresponds to the free intein domain and the full-size fusion protein, respectively. No intein was detectable in the periplasmic fraction. This shows that ssOmpA-GFP was released into the periplasm only after intein cleavage, which most likely happened spontaneously in the cytoplasm as described before (25, 26). It remains possible that the intein was released into the cytoplasm during the transport of ssOmpA-GFP by the action of signal peptidase on the signal peptide region. But this can only happen if the signal peptide, sandwiched between two folded domains can get inserted into inner membrane . To what extent, the N-terminally fused intein affects the events leading to membrane binding of a signal peptide is yet to be determined. Since GFP is unable to fold in the periplasm (23), the observed fluorescence must arise from GFP that was pre-folded in the cytoplasm. This folded GFP was transported into the periplasm either by the Sec system, despite being folded, or by the Tat system, despite the OmpA signal sequence being Sec-specific. In order to discriminate between these two possibilities, intein-ssOmpA-GFP was also expressed in E. coli strain DADE, which lacks all known Tat genes (21). For comparison, we included a construct encoding intein-ssTorA-GFP, providing GFP with the signal sequence of the genuine Tat-substrate trimethyl n-oxide reductase. Fluorescence intensities in the periplasm of DADE cells expressing intein-ssOmpAGFP or intein-ssTorA-GFP were low and indistinguishable (Figure 2a). The observed intensities apparently originate from background fluorescence from other periplasmic components. The low fluorescence intensity in the periplasm of DADE cells shows that in MC4100 cells both GFP variants were transported into the periplasm in a Tat-dependent manner. We estimated the fraction of folded, fluorescent GFP that was transported by the Tat-system from the data presented in Figure 2a. For each construct, the fluorescence intensity at 510 nm of DADE Rescue of folded Sec substrates 61 periplasmic extracts was taken as background, and subtracted from the periplasmic intensity obtained from MC4100 periplasmic extracts. The result was divided by the total fluorescence intensity of the respective MC4100 cells, yielding transported fractions of 0.38 for ssTorA-GFP and 0.23 for ssOmpA-GFP, respectively (Figure 2d). Based on this calculation the Tat system seems to transport ssOmpA-GFP with an efficiency of 60% relative to ssTorA-GFP. Figure 2. Tat transports folded ssOmpA-GFP. a. Fluorescence emission spectra of periplasmic extracts of E. coli cells expressing GFP as intein-capped transport substrates. InteinssTorA-GFP (black lines) or intein-ssOmpA-GFP (grey lines) was expressed in MC4100 (solid lines) and DADE (dashed lines, the two spectra coincide). b. Western blot analysis of the cytoplasm and periplasm of MC4100 cells expressing intein-ssOmpAGFP, using anti-GFP antibody. Both lanes correspond to the same number of cells. c. Western blot of the same samples used in panel b, using antibody against the intein tag. d. The fraction of GFP that was transported in a Tat-dependent manner for E. coli cells expressing GFP fused behind the indicated signal sequence and intein, calculated as (Fperi,MC4100 Fperi,DADE) /Ftotal. To our knowledge, this is the first report of Tat-mediated transport of a protein with a genuine Sec-specific signal sequence that lacks the entire consensus double-arginine motif. Even conservative mutations of the arginine pair (RR) of the TorA signal sequence have been found to slow down (KR and RK) or even block (KK) Tat-mediated transport (27). As a control, we mutated the three consecutive arginines in the RR-motif of intein-ssTorAGFP (SRRRFLA to SKKKFLA) and expressed the mutated construct in MC4100 cells. As expected, no GFP fluorescence was observed in the periplasm of these cells (not shown). Apparently, a mutated twin-arginine sequence is worse than none at all. This suggests that the lack of a double arginine motif can somehow be compensated for by other characteristics of the signal sequence. In an earlier study we found that Tat signals directly interact with biological membranes, and we hypothesized that such interactions might play an important role in the transport of Tat substrates (25). Diminished or structurally altered membrane interactions of signal peptides without a double arginine motif may be partially responsible for their low transport competence. Changing the characteristics of the Hand C-regions can perhaps restore the necessary membrane interactions. The Tat system also transports folded cytochrome c with a Sec-specific signal It could be possible that the described Tat-dependent transport of a folded Secsubstrate was specific for GFP. In order to address this option we used cytochrome c as an alternative substrate. Cytochrome c is a 14-kDa protein that folds around a heme cofactor, and in the absence of heme the apoprotein is largely unfolded (28). The two covalent bonds between the cofactor and the protein need to be formed enzymatically. We previously engineered a cytochrome c that can either be matured by E. coli’s own over-expressed enzyme system, which is active in the periplasm, or by the heterologously expressed mitochondrial one, which is active in the cytoplasm (29). Cytochrome c with a Sec-specific signal sequence was only transported in an unfolded conformation, whereas cytochrome c with a Tat-specific signal sequence is only transported when fully folded and matured (Figure 3a) (20). Transport of Tat substrates is thus subject to a tight quality control: unfolded or immature proteins are not transported (20, 28). The nature of this quality control is still unclear. We investigated the effect of N-terminal masking on the translocation of cytochrome c. Plasmids have been constructed encoding intein-ssOmpA-cytochrome c and intein-ssTorAcytochrome c. These plasmids were transformed into E. coli strains MC4100 and DADE, also expressing the cytoplasmic maturation system. The presence of this maturation system is expected to result in folded cytochrome c in the cytoplasm. The cells were grown to the same OD and the cytoplasmic and periplasmic fractions were isolated. The samples were tested for the presence of matured and thus folded cytochrome c by heme staining as described in Materials and Methods. When expressed as an intein fusion, ssOmpA-cytochrome c was indeed matured in the cytoplasm of MC4100 and DADE cells (Figure 3b, 1 and 3 lane). Maturation is made Rescue of folded Sec substrates 63 possible by N-terminal masking, since ssOmpA-cytochrome c is not matured in the cytoplasm when expressed without an intein (20). Interestingly, folded cytochrome c was found in the periplasm of wild-type cells, but not in the periplasm of DADE cells (Figure 3b, 2 and 4 lane). The results on cytochrome c translocation are only interpreted qualitatively, since dotblots hardly contain quantitative information. Folded ssOmpA cytochrome c is clearly transported into the periplasm in a Tat-dependent manner, just as described above for folded ssOmpA-GFP. Intein-ssTorA-cytochrome c behaves as expected. Mature cytochrome c was found in the cytoplasmic fractions of MC4100 and DADE cells and also in the periplasm of MC4100 cells, but not in the periplasm of DADE cells (Figure 3b, 5th to 8th lane). This shows that after removal of intein folded ssTorA-cytochrome c is released into periplasm. In conclusion, the results on cytochrome c translocation confirm that the Tat system transports folded Sec substrates. N-terminal masking does not lead to translocation of unfolded Tat substrates N-terminal masking was also used to investigate the influence of co-translational recognition on the folding and translocation of Tat substrates. In absence of the cytoplasmic maturation system, cytochrome c remains unfolded in the cytoplasm. Previously, we showed that unfolded cytochrome c is not transported by the Tat system (20). We now investigated whether N-terminal masking results in translocation of this protein, either via the Sec or via the Tat pathway. We co-expressed intein-ssTorA-cytochrome c with E. coli’s own periplasmic maturation system, in order to stabilize and detect transported cytochrome c. However, to test the periplasmic maturation system we first co-expressed it with intein-ssOmpA-cytochrome c. Mature cytochrome c was observed in the periplasm of both MC4100 and DADE cells, but not in their cytoplasm (Figure 3b, 9 to 12 lane). This proves that unfolded ssOmpA-cytochrome c was released into periplasm following transport via the Sec pathway, since presence or absence of Tat translocons did not make any difference. In case of inteinssTorACytochrome c, the periplasmic maturation system failed to produce detectable levels of mature cytochrome c (Figure 3b, 13 to 16 lane), which shows that unfolded ssTorA-Cytochrome c was not translocated into the periplasm. An antibody against the abundant cytoplasmic chaperone DnaJ was used as a control of cell fractionation (Figure 3b). In all cases tested, DnaJ was only found in the cytoplasmic extracts, not in any periplasmic ones. In two cases the control was not performed. In both cases however cytochrome c was found in only one compartment, which is impossible when fractions are contaminated. Furthermore, also intein was found exclusively in all cytoplasmic samples (not shown). The absence of intein from all periplasmic fractions confirmed once more the quality of the fractionation, and in addition it shows that cytochrome c is not transported as long as the intein moiety is bound to its N-terminus. The results obtained on cytochrome c translocation are summarized schematically in the bottom panel of Figure 3b. Unfolded ssOmpA-cytochrome c was transported by the Sec machinery after intein cleavage. It is likely that co-translational recognition is not essential for the transport process itself. In other words, a polypeptide is competent with Sec-mediated translocation as long as it remains unfolded. Usually, co-translational interference is needed to prevent folding. This principle is often referred to as kinetic competition (30). N-terminally masked unfolded ssTorA Cytochrome c was not translocated into the periplasm, neither by the Sec nor by the Tat pathway. Cotranslational recognition may not be essential for folding quality control of the Tat pathway, and the Sec pathway does not accept unfolded proteins bearing a Tat specific signal. Figure 3. Cytochrome c as a marker for protein transport. a. Schematic overview of the cytochrome c-based folding/translocation assay as described by Sanders et al. (2001). Heme (shown in red) can be incorporated into apocytochrome (shown as a black line) in the cytoplasm by a plasmid-encoded maturation enzyme (light blue). Mature cytochrome c (black line + red square) is transported across the inner membrane (gray) by the Tat system (dark blue), provided that it has the correct signal sequence (cyan). Unfolded cytochrome c with the appropriate signal sequence is transported by the Sec system (yellow), where it can be matured by E. coli’s over expressed maturation system (green). b. Localization of mature cytochrome c expressed as intein-signal-cytochrome c. The top panel gives the location of the cytochrome c maturation system (cytoplasm or periplasm), the signal sequence of the respective construct (ssOmpA or ssTorA), the strain (wild-type or DADE) and the cell fraction (c for cytoplasm or p for periplasm). The second panel gives the results of the heme-stain (folded cytochrome c) and the western-blot against DnaJ. The bottom panel gives a schematic summary of the results: the compartments in which mature cytochrome c is found are colored red. Rescue of folded Sec substrates 65 Discussion Sec and Tat substrates are sorted at a late post-translational stage Signal sequences play a central role in targeting transport substrates to the respective translocation systems. A “Sec avoidance motif” on Tat signal peptides has been proposed (17, 19). More recently, also the N-terminus of the mature domain was shown to be involved in Sec avoidance of Tat substrates (31). It is a long-standing question how and where the two transport paths split. One hypothesis is that already on the ribosome Sec-specific and Tat-specific signals are recognized and transport substrates are funneled into the respective routes (8). However, unfolded cytochrome c bearing a Tat specific signal peptide was not translocated out of the cell by the Sec system (Figure 3). This result does not depend on co-translational sorting, since Nterminal masking has no effect on the outcome of the experiment. Apparently, the Sec system discriminates between Sec and Tat-specific signals at a later stage. On the otherhand, it remains that the masking of the relatively long Tat signal sequence by intein may be not as effective as masking of the shorter Sec signal sequence. If co-translational processes are not essential for Sec avoidance of Tat substrates then the decision made on the ribosome could perhaps be reduced to whether a protein constitutes a transport substrate or not. The next interaction partner of all positively identified transport substrates then could be SecB. SecB is by nature a general chaperone which in vitro and in absence of trigger factor binds to all nascent polypeptides (32). The presence of trigger factor at the ribosome exit pore is needed to restrict SecB binding to transport substrates (32). Furthermore, it has been shown that the export rate of a protein destined for Tat transport is increased in the absence of SecB, which strongly suggests that SecB binds to Tat substrates in vivo (19). For the following step on the Sec-route, in vitro experiments have shown that SecA binds signal peptides of the Sec and the Tat pathway equally well, and both binding events lead to ATP hydrolysis (33). Thus also SecA seems to be unable to discriminate between Sec and Tat signals. It may thus be possible that the decision of routing transport substrates into two different paths is not made on the ribosome but further down the route, perhaps as far as the Sec translocation complex.