J Biol Chem, Vol. 274, Issue 51, 36073-36082, December 17, 1999

Sec-independent Protein Translocation in Escherichia coli
A DISTINCT AND PIVOTAL ROLE FOR THE TatB PROTEIN^*

Frank Sargent^§, Nicola R. Stanley^§¶, Ben C. Berks, and Tracy Palmer^§

From the  Centre for Metalloprotein Spectroscopy and Biology, School of Biological Sciences, University of East Anglia, Norwich NR4 7TJ, United Kingdom and the ^§ Department of Molecular Microbiology, John Innes Centre, Norwich NR4 7UH, United Kingdom

	ABSTRACT

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS AND DISCUSSION REFERENCES

In Escherichia coli, transmembrane translocation of proteins can proceed by a number of routes. A subset of periplasmic proteins are exported via the Tat pathway to which proteins are directed by N-terminal "transfer peptides" bearing the consensus (S/T)RRXFLK "twin-arginine" motif. The Tat system involves the integral membrane proteins TatA, TatB, TatC, and TatE. Of these, TatA, TatB, and TatE are homologues of the Hcf106 component of the pH-dependent protein import system of plant thylakoids. Deletion of the tatB gene alone is sufficient to block the export of seven endogenous Tat substrates, including hydrogenase-2. Complementation analysis indicates that while TatA and TatE are functionally interchangeable, the TatB protein is functionally distinct. This conclusion is supported by the observation that Helicobacter pylori tatA will complement an E. coli tatA mutant, but not a tatB mutant. Analysis of Tat component stability in various tat deletion backgrounds shows that TatC is rapidly degraded in the absence of TatB suggesting that TatC complexes, and is stabilized by, TatB.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS AND DISCUSSION REFERENCES

In bacteria, transmembrane translocation of proteins can proceed by a number of routes depending upon the nature of both the targeting signal and the substrate. Most proteins destined for export are synthesized with N-terminal extensions, termed "signal peptides," and the translocation event is catalyzed by the general secretory (Sec) apparatus (1, 2). Signal peptides, which direct export via the Sec pathway, show little identity with each other at the amino acid level, although they share similar structures and physicochemical properties (3).

A subset of periplasmic, and periplasmically oriented, proteins are exported via a pathway which operates independently of the core Sec translocon in Escherichia coli (4, 5). Substrates of this alternative system are synthesized with, or in some cases associate with partner proteins possessing, N-terminal "transfer peptides" bearing the distinctive (S/T)RRXFLK "twin arginine" motif (6). The bacterial twin arginine transfer peptide-dependent protein translocase (Tat system) is apparently both structurally and mechanistically related to the pH-dependent protein import machinery identified in chloroplast thylakoid membranes (7, 8).

Minimally, the Tat machinery comprises four probable membrane proteins: TatA, TatB,¹ TatC, and TatE. The tatC gene product is highly hydrophobic and probably contains six transmembrane helices (10). TatA, TatB, and TatE are sequence-related proteins that are predicted to comprise a single N-terminal transmembrane -helix followed by a water-soluble amphipathic -helix at the cytoplasmic side of the membrane (11-13). TatA and TatE are more than 50% identical in sequence, while the TatB sequence is more divergent (~25% amino acid identity) and has a considerably extended C-terminal region. The TatA/B/E proteins are homologues of maize Hcf106, the first component of the plant thylakoid pH-dependent pathway to be identified (11).

In-frame deletion mutants of tatA or tatE exhibit variable, but never complete, defects in the localization of proteins with twin arginine transfer peptides (12). Targeting to the Tat pathway is, however, completely blocked in a tatAtatE double mutant strain suggesting that TatA and TatE have overlapping essential functions in the Tat system (12).

Two mutant alleles of tatB have been described. Weiner and co-workers (9) isolated a point mutant resulting in a leucine for proline substitution at position 22. This tatB P22L mutant was unable to correctly localize the twin arginine transfer peptide-dependent trimethylamine N-oxide (TMAO)² reductase, Me₂SO reductase, or periplasmic nitrate reductase (9). A strain in which tatB was disrupted by insertion of a kanamycin resistance cassette has also been described (13). As with the tatB P22L mutant, the tatB::kan^R insertion mutant was found to be defective in the export of TMAO reductase. However, in marked contrast to a tatAtatE mutant, the tatB::kan^R mutant was reported to be unaffected in membrane targeting and translocation of the transfer peptide-dependent hydrogenase-2, suggesting that TatB is not required for the translocation of all transfer peptide-bearing proteins (13). Furthermore, while TatB is truncated at amino acid 88 in the tatB::kan^R mutant, the mutant phenotype of the tatB P22L strain can be suppressed by a plasmid encoding only the initial 100 (from a possible 171) amino acid residues of TatB (9). It is therefore conceivable that the N- and C-terminal regions of TatB comprise functionally distinct domains that are differentially affected in the two tatB mutant alleles.

In the current work we have sought to define in more detail the roles of the three homologous TatA, TatB, and TatE proteins. An in-frame deletion mutant of tatB has been constructed allowing determination of the phenotype of an unambiguously null tatB allele, as well as direct phenotypic comparison with previously characterized in-frame deletion mutants in tatA, tatC, and tatE (10, 12). The effects of tatB and tatB::kan^R mutations on the localization of an extensive range of Tat-dependent substrates have been assessed. Our studies show that TatB is an essential component of the Tat pathway for all the substrates tested. No evidence for phenotypic variation between the two tatB alleles was detected. While we confirm a previous report that a portion of hydrogenase-2 activity is associated with the cytoplasmic membrane (13), we go on to demonstrate that this protein is not translocated but remains on the cytoplasmic face of the membrane. Taken together, our results argue against the idea that different Tat components interact with different sets of substrate preproteins. Complementation analysis is used to support the proposal that TatA and TatE are functionally equivalent to one another, but are functionally distinct from the TatB protein. Remarkably, E. coli TatA and TatE function can be substituted by TatA from Helicobacter pylori, suggesting that only weak amino acid sequence constraints are placed on these essential Tat components. Finally, in vivo pulse-chase experiments demonstrate that the TatB protein plays a key role in stabilizing TatC, providing initial evidence for possible complex formation between TatB and TatC.

	EXPERIMENTAL PROCEDURES

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS AND DISCUSSION REFERENCES

Bacterial Strains and Growth Conditions-- A list of the E. coli K-12 strains utilized under "Results and Discussion" of this study are shown in Table I.

During all genetic manipulations, E. coli strains were grown aerobically in Luria-Bertani (LB) medium (17). Concentrations of antibiotics were as described previously (12). Growth phenotypes of mutants were determined on M9 minimal medium (17). For all biochemical characterizations, cells were cultured in the medium of Cohen and Rickenberg (CR) (18) fortified as described by Sawers et al. (19). For anaerobic respiration, CR was supplemented with glycerol (0.5% w/v) or glucose (0.2% w/v) and TMAO, Me₂SO, sodium nitrate, or sodium fumarate (all at 0.4% w/v) where indicated. For aerobic respiration and anaerobic fermentations, glucose was added to 0.4% (w/v).

Mutant Construction-- The tatB deletion strain BØD was constructed as follows: a 618-base pair (bp) fragment covering downstream sequence and last eight tatB codons was amplified by PCR using primers TATB2 5'-GCGCATCGATCCTTCGTCGAGTGATAAACCGTAA-3' and TATB3 5'-GCGCGGTACCTTGCGTAAGTCTTCTGGCGAG-3' with MC4100 chromosomal DNA template. The product was digested with ClaI and KpnI and cloned into pBluescript (Stratagene, La Jolla, CA) to give plasmid pFAT163. An 828-bp fragment (upstream sequence and start codon of tatB) was amplified using primers TATA1 (12) and TATB1 5'-GCGCGGATCCCACGGATTACACCTGCTCTTTATC-3' with MC4100 DNA template. The product was digested with XbaI and BamHI cloned into pFAT163 to give pFAT164. The tatB in-frame deletion allele was excised by digestion with XbaI and KpnI and cloned into the polylinker of pMAK705 (20). The mutant allele of tatB was transferred to the chromosome of MC4100 as described (20) to give strain BØD.

The tatA,tatE double deletion strain JARV16 was constructed as follows. A 574-bp fragment covering downstream sequence and final six codons of tatA was amplified by PCR using primers TATA6 5'-GCGCATCGATGATAAAGAGCAGGTGTAA-3' and TATA4 (12) with MC4100 DNA template. The product was digested with ClaI and KpnI and cloned into pFAT4 (12) to give plasmid pFAT16. DNA covering the in-frame deletion of tatA was excised by digestion with XbaI and KpnI and cloned into the polylinker of pMAK705. The mutant allele of tatA was transferred to the chromosome of J1M1 (MC4100, tatE) (12) to give strain JARV16. Note that JARV16 was constructed as replacement for the original tatA,tatE double deletion strain, JARV15, described by Sargent et al. (12). The strategy adopted for the construction of JARV15 may have inadvertently altered part of a putative tatB Shine-Dalgarno sequence located within the tatA gene (12). We therefore re-constructed the tatA,tatE strain maintaining all possible tatB Shine-Dalgarno sequences. With respect to Tat-dependent protein export, the JARV16 and JARV15 (12) strains have identical phenotypes (data not shown).

The hydrogenase mutant strain FTD89 was constructed as follows. A 566-bp fragment covering upstream sequence and the first four codons of hyaB (encoding the catalytic subunit of hydrogenase-1 (21)) was amplified from MC4100 DNA by PCR using primers HYA1 5'-GCGCTCTAGACATTATCAAAGTACCTGGCTGCCC-3' and HYA2 5'-GCGCGGATCCCTGAGTGCTCATGCCTGTTTATCC-3' digested with XbaI and BamHI and cloned into pBluescript to give plasmid pFAT420. A 412-bp fragment covering the last five codons and DNA downstream of hyaB was amplified from MC4100 chromosomal DNA using the primers HYA3 5'-GCGCGGATCCGTGCAGGTGCGTTAACAGGAAGG-3' and HYA4 5'-GCGCAAGCTTGCTGTGAATGTCCATTGAGTTGCC-3', digested with BamHI and HindIII and cloned into pFAT420 to yield pFAT421. Following digestion with XbaI and KpnI, the DNA covering the in-frame deletion of hyaB in pFAT421 was transferred to pMAK705 (resulting in plasmid pFAT422), the mutant allele was moved to the chromosome of MC4100 by homologous recombination as described (20) to give strain FTD22 (as MC4100, hyaB). A 534-bp fragment covering the upstream DNA sequence and the first three codons of hybC (encoding the catalytic subunit of hydrogenase-2 (22)) was amplified from MC4100 DNA using primers HYB1 5'-GCGCTCTAGACGTTCATCATGGGCTTCTCGATTG-3' and HYB2 5'-GCGCGGATCCCTGGCTCATGCTTTGCTCGCC-3', digested with XbaI and BamHI and cloned into pBluescript to give plasmid pFAT430. A 566-bp DNA fragment covering the last eight codons of hybC and downstream sequence was amplified from MC4100 DNA using primers HYB3 5'-GCGCGGATCCGTGGTTTCAGTGAAGGTTCTGTAAT-3' and HYB4 5'-GCGCAAGCTTCGCTGCCTGTACTTGCGCCTTCGG-3', digested with BamHI and HindIII and cloned into pFAT430 to give pFAT431. The in-frame deletion of hybC in pFAT431 was excised by digestion with XbaI and KpnI, cloned into pMAK705 (resulting in plasmid pFAT432), and transferred to the chromosome of FTD22 as described (20). The resultant hyaB,hybC strain was named FTD89.

All chromosomal in-frame deletion strains were carefully constructed so to preserve identifiable regulatory elements, coding sequences, stop codons, and Shine-Dalgarno sequences of genes flanking the deletions. Chromosomal deletion strains were confirmed by PCR, and chromosomal PCR products were sequenced to ensure that no point mutations had been introduced.

BØD-P (BØD, pcnB1 zad-981::Tn10d (Kan^R)) and JARV16-P (JARV16, pcnB1 zad-981::Tn10d (Kan^R)) were constructed by P1 transduction (23) of the pcnB1 allele from VJS5833 (kindly provided by Professor V. Stewart). Both tat strains were difficult to transduce with P1. Therefore pFAT205, a temperature-sensitive plasmid carrying tatABCD (below), was introduced into the tat mutants prior to P1 transduction. Transduced strains were selected by virtue of acquired kanamycin resistance, which is linked to the pcnB allele. Kanamycin-resistant tat mutant strains were subsequently cured of pFAT205 by growth overnight at 44 °C. To confirm the linked pcnB allele had been co-transduced into the strains, the yield of plasmid DNA was quantified following transformation of the cured strains with pBluescript and subsequent plasmid purification.

Plasmid Construction-- A list of plasmids featured under "Results and Discussion" is shown in Table I. Plasmid pFAT205 carries tatABCD in pMAK705. Initially, tatABCD was excised from pFAT65 (12) by digestion with EcoRI and partial digestion with HindIII and cloned into pQE60 (Qiagen, Crawley, United Kingdom). The tat genes were then excised as a XhoI-PstI fragment and cloned into SalI-PstI-digested pMAK705 to give pFAT205.

Plasmid pFAT222, which carries the tat(A)BCD operon with the tatA in-frame deletion allele in JARV16, was constructed as follows. A 76-bp fragment covering the start codon of tatA and upstream DNA was amplified using primers TATA5 and TATA2 (12) with pFAT65 template. The product was digested with EcoRI and BamHI and cloned into the polylinker of pT7.5 (16) to give plasmid pFAT220. A 2236-bp fragment covering the deletion allele of tatA to the end of tatD was amplified using primers TATA5 and TATD1 with JARV16 DNA template. The product was digested with BamHI and XbaI and cloned into pFAT220 to give plasmid pFAT222.

Plasmid pFAT217, which carries the tatA(B)CD operon with the tatB in-frame deletion allele of BØD, was constructed as follows. A 350-bp PCR fragment covering the start codon of tatB and upstream DNA was amplified using primers TATB1 and TATA5 (12) with pFAT65 template. The product was digested with EcoRI and BamHI and cloned into the polylinker of pT7.5 to give plasmid pFAT206. A 2017-bp PCR fragment covering the deletion allele of tatB plus tatC and tatD was amplified using primers TATA5 and TATD1 (12) with BØD chromosomal DNA template. The product was digested with BamHI and XbaI and cloned into pFAT206 to give plasmid pFAT217.

Plasmid pFAT228, which carries the tatAB(C)D operon with the in-frame deletion allele of tatC present in B1LK0 (10), was constructed as follows. An 840-bp DNA fragment covering tatA,tatB, and initial three codons of tatC was amplified by PCR using primers TATA5 (12) and TATC2 (10) with pFAT65 template. The PCR product was digested with EcoRI and BamHI and cloned into pT7.5 to give plasmid pFAT227. A 1760-bp DNA fragment covering the entire tatAB(C)D operon within the chromosome of B1LK0 was amplified by PCR using primers TATA5 and TATA6 (12) with B1LK0 chromosomal DNA template. The product was digested with BamHI and XbaI and the released 900-bp fragment cloned into pFAT227 to give plasmid pFAT228.

Plasmid pFAT415 carries the tatA gene and more than 500 bp of upstream DNA in pBluescript. It was constructed after PCR amplification of the tatA region using primers TATA1 and TATA4 with MC4100 DNA template. Plasmid pFAT416 carries 500 bp of DNA sequence upstream of tatA (as in pFAT415) but also the tatA allele and the intact tatB gene in pBluescript. pFAT416 was constructed by PCR using primers TATA1 and TATA4 and JARV16 DNA template. Plasmid pFAT45 has tatE as a 1450-bp insert in pBluescript and was cloned following amplification of the tatE region using primers TATE1 and TATE4 (12), with MC4100 chromosomal DNA template. Plasmid pNR42 carries the genes encoding the phage T7 polymerase and the temperature-sensitive  repressor excised from pGP1-2 (16) and cloned as a 4300-bp BamHI-PstI fragment into the polylinker of pSU18 (24).

In the case of plasmids pFAT23Z (tatAB,C::lacZ,D) and pFAT24Z (tatA,B,C::lacZ,D) the tatC gene was replaced from codon 3 to 255 by a complete in-frame lacZ coding sequence within the tat operon and the (tatB) operon as displayed by BØD, respectively. The lacZ gene (minus stop codon) was amplified from pAA182 (25) using primers LACZ1A 5'-GCGCCTCGAGATGACCATGATTACGGATTCACTG-3' and LACZ2A 5'-GCGCGGGCCCTTTTTGACACCAGACCAACTGGTA-3', digested with ApaI and XhoI and cloned into pBluescript to give plasmid pFAT1Z. DNA covering the final 4 codons of tatC and the tatD gene was amplified from pFAT65 with primers TATC3Z 5'-GCGCGGGCCCACTGAAGAATAAATTCAACCGCCCG-3' and TATD1Z 5'-GCGCGGTACCCGATGGTGAGGCTCGCTCC-3', digested with ApaI and KpnI, and cloned into pFAT1Z to give pFAT1ZD (lacZ::tatC'tatD). DNA covering tatAB plus the initial 3 codons of tatC, and tatAB plus the same first 3 codons of tatC, was amplified using primers TATA1 (12) and TATC2Z 5'-GCGCCTCGAGTACAGACATGTTTACGGTTTATCAC-3' with MC4100 and BD DNA template, respectively. PCR products were digested with XbaI and XhoI and cloned into pFAT1ZD to give plasmids pFAT23Z and pFAT24Z.

Plasmids pFATHP1 and pFATHP2, which carry H. pylori 26695 (26) genes HP0320 and HP1060, respectively, were constructed as follows. A 1950-bp region covering HP0320 and a 600-bp region covering HP1060 were amplified by PCR utilizing H. pylori 26695 chromosomal DNA template. Primers for HP0320 amplification were HPTA1 5'-GCGCATCGATCGAAACCCTATAAAACCTATC-3' and HPTA2 5'-GCGCGGATCCGCTTAATCTCTAGCGGAAATTTTGG-3', the product was digested with ClaI and BamHI and cloned into pBluescript to give pFATHP1. Primers for HP1060 amplification were HPTBC1, 5'-GCGCGGATCCGAAAGAAAATTACACTACAATAACG-3', and HPTBC3, 5'-GCGCTCTAGACCTGTAAATGCGGTTTTAAATCTTC-3', the product was digested with BamHI and XbaI and cloned into pBluescript to yield plasmid pFATHP2. All clones obtained from PCR amplified DNA were sequenced to ensure that no mutations had been introduced.

Protein Methods-- Cells were harvested by sedimentation at 7000 × g for 15 min at 4 °C, and washed twice in ice-cold 50 mM Tris-HCl (pH 7.6). Periplasmic fractions were prepared by the lysozyme/EDTA method of Osborn et al. (27) for 0.5 to 2-liter cultures, or by cold mild osmotic shock for smaller (typically 30 ml) cultures. The method of cold mild osmotic shock involved resuspension of the cell pellet from a 30-ml culture in 7.5 ml of 30 mM Tris-HCl (pH 8.0), 20% (w/v) sucrose, 1 mM EDTA. Following a 10-min incubation at 20 °C, cells were resedimented by centrifugation and the pellet taken up in 2 ml of ice-cold 5 mM MgSO₄. After incubation in melting ice for 10 min, the resultant sphaeroplasts were harvested by centrifugation and the supernatant retained as periplasmic fraction. The periplasmic extract was buffered by addition of 100 µl of 1 M Tris-HCl (pH 7.6). Sphaeroplasts were lysed by passage through a French pressure cell and separated into membrane and cytosolic fractions as described (28). Rocket immunoelectrophoresis was performed as described previously (29). Trypsin-catalyzed solubilization of membrane-bound hydrogenase 2 activity from both washed membranes and spheroplasts was performed essentially as described (30, 31). Protein concentration was estimated by the method of Lowry et al. (32).

Expression and radiolabeling of tat gene products with [³⁵S]methionine (NEN Life Science Products Inc., Boston, MA) from plasmids pFAT65, pFAT217, pFAT222, and pFAT228, under control of the phage T7 10 promoter, was as described previously (12). Radiolabeled cells were chased by addition of non-radioactive methionine (final concentration 750 µg/ml). Tat substrates preSufI and preYacK from plasmids pNR14³ and pNR19³ were expressed under the control of the phage T7 10 promoter: strains MC4100[pNR42], BØD[pNR42], and MCMTA[pNR42] were transformed with either pNR14 (pT7.5, sufI⁺) or pNR19 (pT7.5, yacK⁺) and radiolabeling performed as described (12), except 400 µg/ml rifampicin (final concentration) was used. Radiolabeled cells were chased with nonradioactive methionine as above.

Enzyme Assays-- -Galactosidase activity was determined by the method of Miller (23). TMAO and Me₂SO reductase were assayed as the substrate-dependent oxidation of reduced benzyl viologen (33, 34). Acid phosphatase was assayed as the hydrolysis of p-nitrophenyl phosphate by the method of Atlung et al. (35). Hydrogenase activity in cell fractions was assayed spectrophotomerically as H₂-dependent reduction of oxidized benzyl viologen (28).

In vivo hydrogen oxidation ("uptake") and evolution (formate hydrogenlyase) activities were assayed by a method based on that described by Sawers et al. (19) utilizing a "hydrogen electrode," i.e. a Clark oxygen electrode adapted for the measurement of hydrogen (Hansatech, Cambridge, United Kingdom). Fumarate-dependent hydrogen uptake was measured as follows. The reaction chamber was filled with 2-ml of 100 mM sodium phosphate (pH 6.8). 25 mM glucose and 2.5 µl of a 50 mg/ml glucose oxidase, 5 mg/ml catalase solution was added to scavenge oxygen. After equilibration for 10 min, 250 µl of hydrogen-saturated 100 mM sodium phosphate (pH 6.8) (equivalent to approximately 203 nmol of molecular hydrogen at 20 °C) was added, followed by a small amount of intact cells (typically 50 µl of a dilute 0.05 g of cells (wet weight) per ml of cell suspension). The reaction was initiated by the addition of 20 mM sodium fumarate. The same protocol was used to measure formate hydrogenlyase activity (formate-dependent hydrogen evolution) although addition of exogenous hydrogen was not required. The reaction was initiated by the addition of 20 mM sodium formate. All enzyme assays were performed in duplicate with results not varying by more than 10% from the mean.

	RESULTS AND DISCUSSION

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS AND DISCUSSION REFERENCES

Phenotype of a tatB Mutant Strain-- Previously reported tatB mutants would still synthesize TatB protein: either a truncated tatB gene product in the case of MCMTA (tatB::kan^R) (13) or full-length but amino acid-substituted TatB in the tatB P22L mutant (9). As a result it is possible that these mutant strains may retain some TatB functions. We therefore constructed a strain, BØD, in which the tatB gene was inactivated by an almost complete in-frame deletion. Using this strain we were able to investigate the phenotypic consequences resulting from a complete absence of TatB protein. In addition, the tatB strain allowed direct phenotypic comparison between a tatB mutation and the previously characterized in-frame deletion mutants in tatA, tatC, and tatE (10, 12).

We initially assessed the effect of the tatB mutation on the localization of two of the Tat pathway substrates studied in earlier characterizations of alternative tatB alleles (9, 13). TMAO reductase (TorA) is a water-soluble periplasmic molybdoenzyme that allows E. coli to use TMAO as a respiratory electron acceptor (36), and Me₂SO reductase is a complex membrane-bound metalloenzyme required for Me₂SO respiration (37). The tatB mutant, BØD, is viable under aerobic respiratory or anaerobic fermentative growth conditions. The strain is, however, unable to grow on the non-fermentable carbon source glycerol with either TMAO or Me₂SO as sole respiratory electron acceptor. Subcellular fractionation studies revealed a complete mislocalization of both TMAO and Me₂SO reductase activities to the cytoplasm in the tatB mutant (Table II). These phenotypic effects of the tatB mutation are identical to those reported for the tatB P22L mutant (9). The tatB::kan^R mutant, MCMTA, was also reported to be defective in the translocation of TMAO reductase (13), and we confirm this observation but also demonstrate that Me₂SO reductase is mislocalized in this strain (Table II). Thus, with respect to their effect on TMAO and Me₂SO reductase targeting, the three tatB mutant alleles are equivalent.

We next analyzed the effect of the tatB and tatB::kan^R mutations on the targeting of a further two proteins known to be mislocalized in other tat mutants (10, 12). The nitrate-inducible formate:quinone oxidoreductase (formate dehydrogenase-N) is a complex, membrane-bound, metalloenzyme in which the precursor of the active site-containing formate dehydrogenase-N G subunit bears a twin arginine transfer peptide (38). Both subcellular location and enzymatic activity of formate dehydrogenase-N were assessed by rocket immunoelectrophoresis (Fig. 1, a and b). Formate dehydrogenase-N can be visualized predominantly in the membranes of the parental strain (Fig. 1, a and b, lanes 1 and 5). However, in both the tatB and tatB::kan^R mutants membrane targeting of formate dehydrogenase-N is impaired resulting in the cytoplasmic accumulation of catalytically inactive enzyme (Fig. 1, a and b, lanes 2, 3, 6, and 7). Hydrogenase 1 is a membrane-bound enzyme in which export of both the peripheral membrane catalytic large subunit and electron transferring small subunit is probably directed by a single twin arginine transfer peptide on the small subunit precursor (39, 40). A third, integral membrane, subunit is probably also present in the mature enzyme complex (41). Rocket immunoelectrophoresis shows that hydrogenase-1 activity is mislocalized to the cytoplasm in both the tatB and tatB::kan^R mutants (Fig. 1c, lanes 2, 3, 6, and 7). Thus tatB mutations affect hydrogenase-1 targeting but not the insertion of the active site nickel cofactor.

View larger version (75K): [in this window] [in a new window]   Fig. 1.   Formate dehydrogenase-N and hydrogenases-1 and -2 accumulate in the cytosol in tatB mutant strains. Cells were grown anaerobically with either glycerol plus nitrate to induce expression of formate dehydrogenase-N, or glycerol plus fumarate for maximal expression of the uptake hydrogenases. Sphaeroplasts were formed (lysozyme/EDTA method), further fractionated and rocket immunoelectrophoresis performed as described under "Experimental Procedures." All plates are as follows: lanes 1, MC4100 (parental strain), membrane fraction; lanes 2, BØD (tatB) membrane fraction; lanes 3, MCMTA (tatB), membrane fraction; lanes 4, DADE (tatABCD, tatE), membrane fraction; lanes 5, MC4100 cytosolic fraction; lanes 6, BØD, cytosolic fraction; lane 7, MCMTA, cytosolic fraction; lane 8, DADE, cytosolic fraction. Periplasmic fractions are not shown. All samples represent the same proportion (0.2%) of total protein present in the two fractions. a, anti-formate dehydrogenase-N, activity stained; b, anti-formate dehydrogenase-N, protein stained; c, anti-hydrogenase 1, activity stained; d, anti-hydrogenase 2, activity stained.

The tatB and tatB::kan^R mutations were further characterized by assessing the effects of the mutations on the export kinetics of two additional possible Tat pathway substrates, SufI and YacK. Sequence analysis indicates that both these proteins belong to the multicopper oxidase superfamily and are synthesized with twin arginine transfer peptides (6). Yet while YacK is predicted to bind four copper ions per protein monomer, SufI may in fact be devoid of copper cofactors.³ Pulse-chase analysis of preSufI and preYacK in the parental strain MC4100 shows a time-dependent processing of the precursor proteins into a smaller mature form (Figs. 2a and 3a). Subcellular fractionation demonstrated that this mature form corresponds to material that has been transported to the periplasmic compartment (data not shown). No processing of either precursor protein was observed when the experiments were repeated in the tatB or tatB::kan^R mutant backgrounds (Figs. 2, b and c, 3, b and c), and subsequent subcellular fractionation confirmed that the radiolabeled precursor proteins remain in the cytoplasm (data not shown). Thus export of SufI and YacK is also completely blocked in the two tatB mutants.

View larger version (21K): [in this window] [in a new window]   Fig. 2.   Export of SufI is blocked in the tatB mutants. In vivo processing of pre-SufI in: (a) MC4100, parental strain; b, BD, tatB; and c, MCMTA, tatB::kanR. In each case precursors were radiolabeled as described under "Experimental Procedures." Samples of whole cells were taken at the time points indicated, proteins separated by SDS-PAGE (12.5% (w/v)) and exposed to photographic film overnight. WT, processing of pre-SufI in MC4100 background after 5-min chase.

View larger version (22K): [in this window] [in a new window]   Fig. 3.   Export of YacK is blocked in the tatB mutants. In vivo processing of pre-YacK in: a, MC4100, parental strain; b, BØD, tatB; and c, MCMTA, tatB::kanR. In each case precursors were radiolabeled as described under "Experimental Procedures," samples of whole cells were taken at the time points indicated, proteins separated by SDS-PAGE (12.5% (w/v)) and exposed to photographic film overnight. WT, processing of pre-Yack in MC4100 background after 5-min chase.

The complete mislocalization of all six tested Tat substrates seen in both the tatB and tatB::kan^R mutants suggests not only that TatB is an essential component of the Tat apparatus but also that the two tatB alleles studied are phenotypically equivalent. A complete block in the targeting of TMAO reductase, Me₂SO reductase, formate dehydrogenase-N, and hydrogenase-1 is also exhibited by tatC (10) and tatAtatE strains (12), as well as a strain DADE deleted in all currently known tat genes (Table II; Fig. 1, a, b, and c, lanes 4 and 8). Clearly, this is in marked contrast to single tatA or tatE mutations that only give rise to partial defects in Tat protein localization (12).

Next, we tested the effect of the tatB mutations on protein export by the Sec system. Acid phosphatase is one of six phosphatase isoenzymes expressed in E. coli, and its targeting to the periplasm is directed by a classic Sec-type leader sequence (42). The acid phosphatase has a pH optimum of 2.5, thus its activity can be easily distinguished from that of the other phosphatases. Moreover, although expressed at a low level aerobically, the expression of the appA gene encoding acid phosphatase is maximal in cells growing anaerobically which correlates well with the expression of many of the Tat substrates described in this work (42).

The tatB and tatB::kan^R mutations had no effect on periplasmic targeting of the Sec-dependent acid phosphatase enzyme (Table II) confirming that TatB is not required for export via the Sec pathway. Furthermore, neither mutation affects the localization of the predominant nitrate reductase and fumarate reductase enzyme activities, both of which are reliant on metallocofactor-binding proteins located at the cytoplasmic face of the membrane (Table II). These observations, together with the cytoplasmic accumulation of some catalytically active Tat substrates, suggest that TatB is not required for metallocofactor insertion.

Export of Hydrogenase-2 Requires a Functional TatB Protein-- Hydrogenase-2 is a structurally related isoenzyme of hydrogenase-1 in that the core enzyme comprises a nickel-iron cofactor-binding large subunit and a twin arginine transfer peptide-bearing small subunit (41). Nevertheless, Chanal and co-workers (13) have reported that biosynthesis of this enzyme is unaffected in their tatB::kan^R mutant and as a consequence suggested that TatB is not involved in the translocation of all proteins with twin arginine transfer peptides. We have now investigated this observation in more detail. Initially we used rocket immunoelectrophoresis to monitor the subcellular location of the catalytically active peripheral membrane subunits of hydrogenase-2 in various genetic backgrounds. In both the tatB and tatB::kan^R mutants, while there is considerable mislocalization of hydrogenase-2 activity to the cytoplasm, a small proportion of the activity remains tightly associated with the isolated membrane fractions (Fig. 1d). The proportion of membrane-bound hydrogenase-2 activity in the tatB mutants was seen to increase in a growth phase-dependent manner, typically reaching 25% of the parental strain in late stationary phase (data not shown). A proportion of hydrogenase-2 activity is also detectable in membranes from a strain deleted in all known tat genes (Fig. 1d, lanes 4 and 8), and a small amount is present in membranes from tatC (40) and tatAtatE mutants (12). We therefore conclude that, while hydrogenase-2 targeting is severely affected in all the tat mutants, a small proportion of hydrogenase-2 activity associates with the membranes in all such strains and not just in the tatB mutants.

We next investigated whether the membrane-associated hydrogenase-2 activity detected in the tat mutants has been translocated to the periplasmic side of the cytoplasmic membrane. The hydrogenase-2 activities detected by rocket immunoelectrophoresis were assayed using the nonphysiological electron acceptor benzyl viologen (Fig. 1d). If the membrane-bound hydrogenase-2 activity represents enzyme that has been correctly translocated to the periplasmic side of the membrane then it might be expected that the enzyme would be able to transfer electrons from hydrogen to its physiological electron acceptor menaquinone. To ascertain whether this was the case we measured whole cell hydrogen oxidation rates with fumarate as electron acceptor. In these experiments the menaquinone reduced to menaquinol by hydrogenases-1 and -2 is reoxidized via fumarate reductase, an enzyme that does not require the Tat system for biosynthesis (Table II). Fumarate-dependent hydrogen uptake was found to be completely abolished in the tatB and tatB::kan^R mutants, as well as in the complete tat deletion mutant (DADE) and in a control strain (FTD89) which has in-frame deletions in the genes coding for the catalytic subunits of both hydrogenases-1 and -2 (Table III). During fermentative growth E. coli synthesizes a third hydrogenase isoenzyme (hydrogenase-3) as part of the cytoplasmically oriented formate hydrogenlyase complex in which electrons derived from formate oxidation are used to reduce protons to hydrogen (41). Control experiments show that tat mutations have no effect on hydrogen evolution by this cytoplasmically located hydrogenase indicating that Hyc biosynthesis is Tat-independent (Table III).

Clearly the membrane-localized hydrogenase-2 activity in the tat mutants is not physiologically active. This would be consistent with a defect in the export of the benzyl viologen-reactive peripheral membrane subunits, but could also conceivably be caused by inactivity of electron transfer components located between the core hydrogenase and the menaquinone pool. In order to distinguish between these two possibilities we attempted to determine directly whether any of the membrane-associated hydrogenase-2 was located on the periplasmic side of the cytoplasmic membrane. In the parental strain, the peripheral membrane subunits of hydrogenase-2 are anchored to the periplasmic face of the plasma membrane by a short hydrophobic domain at the extreme C terminus of the small subunit (31). Limited trypsinolysis results in specific proteolytic cleavage of the C-terminal domain of the small subunit at its exposed N-region. As a result, the peripheral membrane subunits are released as a stable, water-soluble fragment that retains catalytic activity with benzyl viologen as electron acceptor (30, 31, 43). Trypsin treatment of sphaeroplasts formed from the parental strain MC4100 resulted in the release of benzyl viologen-linked hydrogenase activity into the sample medium over time (Fig. 4a, ). No hydrogenase release from these sphaeroplasts could be detected in the absence of protease (Fig. 4b, ). In contrast, only a small amount of hydrogenase activity could be detected in the trypsin-treated soluble protein fractions of sphaeroplasts prepared from the tatB and tatB::kan^R mutants (Fig. 4a,  and ). Furthermore, the levels of soluble hydrogenase activities released from the tatB mutant sphaeroplasts were almost identical to those measured in the control experiment in the absence of trypsin (Fig. 4, b, , and b, ). Thus the small amount of hydrogenase release detected was due to leakage from the cytoplasm of the soluble, active hydrogenase precursors identified in the tatB mutants (Fig. 1, a and b) as opposed to enzyme-catalyzed release. It was, however, possible to release the membranous hydrogenase-2 activity found in the tatB mutants by trypsin treating isolated membranes rather than sphaeroplasts (data not shown). Taken together, these data demonstrate that hydrogenase-2 is not translocated in the tatB mutants. However, a proportion of the active cytoplasmic precursor binds to the cytoplasmic face of the plasma membrane probably, given the trypsin sensitivity of this material, via the same hydrophobic C-terminal domain that would normally anchor these subunits to the periplasmic face of the membrane.

View larger version (20K): [in this window] [in a new window]   Fig. 4.   Hydrogenase 2 is not translocated in the tatB mutants. Cells were harvested in stationary phase and sphaeroplasts formed by the lysozyme/EDTA method described under "Experimental Procedures." Sphaeroplasts were incubated with: a, 0.25% (w/v) trypsin; or b, without trypsin, and soluble hydrogen:benzyl viologen oxidoreductase activities measured at the time points indicated. Hydrogenase activity was released from: () MC4100, parental strain; , BØD, tatB, and , MCMTA, tatB::kanR derived spheroplasts. Enzyme activities are expressed relative to the total cellular hydrogenase activities determined after disruption of sphaeroplasts.

It is notable that, in contrast to hydrogenase-2, the closely related E. coli hydrogenase-1 isoenzyme displays no membrane association in the tat mutant strains even though all enzymes of this type have weakly homologous membrane anchor regions at the C terminus of the small subunit (44). In the case of integral membrane proteins, the orientation of transmembrane regions is normally fixed by the distribution of basic amino acids close to the transmembrane helix ends according to the "positive-inside rule" (45). In the absence of an active translocation system, it is therefore conceivable that the hydrogenase-2 small subunit possesses sufficient basic residues within the N-region of the C-terminal anchor domain (including those that are the target for trypsin proteolysis) such that this transmembrane helix region can insert into the membrane at low efficiency, either in an inverted orientation or as a helical hairpin. In contrast, hydrogenase-1 lacks basic residues in this region. Furthermore, the ability of small subunits to maintain hydrogenase attachment at the periplasmic face of the membrane in mutants devoid of a third, hydrogenase-specific, integral membrane subunit varies between enzymes from different biological systems (46, 47), indicative of variable membrane affinities between related proteins. The reason for this variable behavior is unclear but may also underlie the differing behavior of E. coli hydrogenases-1 and -2 in tat mutants.

Transcomplementation Analyses of tatB Mutants-- With a well characterized tatB allele in hand we were able to examine the functional interrelationships between the homologous TatA, TatB, and TatE proteins by means of complementation analysis. In these experiments the tat genes under test were carried on a plasmid and expressed under the control of the tatA promoter. Complementation was assessed as the ability of the tested genes to restore periplasmically located TMAO reductase activity (i.e. a TorA⁺ phenotype) to the mutant strains.

The TorA phenotype of all mutant strains tested (tatAtatE, tatB, tatB::kan^R, and tatABCtatE) could be suppressed by pBR322-based plasmids harboring the entire tatABCD operon (data not shown). However, we found it impossible to complement the tatB and tatB::kan^R mutants in trans with high (pBluescript) or moderate (pSU18) copy number plasmids expressing the tatB gene alone (data not shown). Indeed, such plasmids even rendered the parental strain MC4100 TorA suggesting that increasing the cellular levels of TatB relative to the other Tat proteins prevents the Tat system from functioning. In an attempt to circumvent this stoichiometry problem we introduced the pcnB1 mutation, an allele that restricts the copy number of ColE1-type plasmids such as pBluescript (48), into the tatB strain BØD by P1 transduction resulting in the strain BØD-P. The pcnB1 allele was also introduced to the tatAtatE mutant JARV16 to yield the strain JARV16-P, although it should be noted that this was not strictly necessary for successful plasmid complementation of this strain (data not shown).

In the presence of the pcnB1 mutation the block on TorA export in a tatB strain can now be corrected by the provision of tatB in trans (Table IV) but not by additional plasmid-borne copies of either the tatA or tatE genes (Table IV). The TorA export defect in JARV16-P (tatAtatE, pcnB1) can be fully corrected by supplying only one of the tatA or the tatE genes in trans, but not by providing the cells with extra copies of tatB on a plasmid (Table IV). These results support the idea that the TatA and TatE proteins are functionally interchangeable, but that the role of the homologous TatB protein in Tat-dependent export is distinct from that of TatA and TatE. Moreover, the data presented earlier in this work demonstrate that this functional division does not correspond to interactions with two different pools of precursor proteins.

TatB Stabilizes the tatC Gene Product-- It is likely that at least some of the known tat gene products interact directly with each other and may even form complexes. We considered it feasible that the loss of such interactions in single tat gene mutants might manifest itself as a decrease in the stability of the Tat proteins that normally interact with the missing gene product.

In order to assess the relative stabilities of the tat gene products, the plasmid pFAT65, harboring the tatABCD operon under the control of an inducible T7 promoter, was introduced into the tatABCDtatE strain DADE. Following induction of plasmid-borne tat gene expression the cells were labeled with [³⁵S]methionine for 5 min followed by a 60-min chase with excess non-radioactive methionine. No significant degradation of the tat gene products was observed (Fig. 5, a, b, and c). Note that by a similar approach we have been able to establish that the radiolabeled Tat proteins produced by pFAT65 are targeted to their predicted cellular compartments (membrane for TatA, -B, and -C, and cytoplasm for TatD).⁴ This suggests that a functional Tat apparatus would be produced in these experiments and pFAT65 can indeed suppress the mutant phenotype of the tatABCDtatE strain (data not shown).

View larger version (56K): [in this window] [in a new window]   Fig. 5.   The tatC gene product is destabilized in the absence of an active tatB gene. Pulse-chase analysis of tat gene products expressed in DADE (tatABCD, tatE) were performed as described under "Experimental Procedures." After a 5-min pulse with [35S]methionine the cells were chased with non-radioactive methionine for a total of 60 min. Whole cell aliquots were taken at the time points indicated and reactions stopped by flash-freezing in liquid nitrogen. Proteins were separated by SDS-PAGE (12.5% (w/v)), and exposed to photographic film overnight. "Control" samples in each experiment were a pulse labeling of the products of the tatABCD operon harbored on pFAT65. a, pulse labeling of the products of tat(A)BCD (pFAT222). b, pulse labeling of the products of tatA(B)CD (pFAT217). c, pulse labeling of the products of tatAB(C)D (pFAT228).

The tatB deletion found on the chromosome of BØD was introduced into pFAT65 producing the tatA(B)CD plasmid pFAT217. Pulse-chase analysis using this plasmid in the tatABCDtatE strain, DADE, allows an estimate of the relative stabilities of the tat operon products in a tatB background (Fig. 5b). It is clear that after a 5-min pulse essentially no radiolabeled tatC gene product can be visualized (Fig. 5b). In contrast, the radiolabeled polypeptides corresponding to the tatA and tatD gene products remain stable even after a 60-min chase (Fig. 5b). Thus the tatC gene product appears to be rapidly degraded in the absence of an active tatB gene.

In order to eliminate any possibility that the tatB allele had a polar effect on the transcription or translation of the downstream tatC gene an in vivo expression study was undertaken. A translational tatC::lacZ fusion was constructed using a "gene-replacement" strategy as described under "Experimental Procedures" in which the tatC gene was replaced between codons 3 and 255 by a complete in-frame lacZ gene (minus stop codon) encoding -galactosidase. The tatC::lacZ fusion was placed downstream of the native tat promoter and the intact tatA and tatB genes on plasmid pFAT23Z, and also on a similar plasmid harboring the tatB allele present on the chromosome of BØD, pFAT24Z. When expressed in multicopy in aerobically grown MC4100 (Tat⁺), values of 480 and 330 Miller units of -galactosidase activity were recorded for pFAT23Z and pFAT24Z, respectively (compared with a basal level of 1 unit in the strains without the tatC::lacZ fusion). When tatC translation was measured in a Tat strain (DADE) similar values were obtained (data not shown). These data strongly suggest that transcription and translation of tatC would not be significantly impaired by the presence of the upstream tatB allele.

We are therefore confident that the TatC phenotype displayed in a tatB-null background (Fig. 5b) is not a consequence of transcriptional or translational polarity of the tatB deletion on tatC. Furthermore, a second downstream gene, tatD, is still transcribed and translated in these experiments, and what is more, the identical chromosomal tatB mutation can be complemented by tatB alone, thus demonstrating that the essential TatC protein must be synthesized in this strain (Table IV).

A similar experiment using pFAT65 mutagenized with the tatC allele present on the chromosome of B1LK0 (10), shows that the remaining tat operon products are synthesized and remain stable even after a 60-min chase with non-radioactive methionine (Fig. 5c). Thus, while TatB is required for TatC stability the converse does not seem to apply. Finally, introduction of the tatA allele present on the chromosome of strain JARV16 into pFAT65 does not affect the stability of the remaining tat operon products (Fig. 5a). Note also that none of the strains tested in these pulse-chase experiments has an intact tatE gene, thus TatE has no significant effect on the stability of the other Tat proteins.

Taken together, these results suggest that TatB, but not the other Hcf106 homologues TatA and TatE, has a function in stabilizing TatC, possibly because the two proteins form a complex. In the context of this hypothesis, it is likely that TatB forms a scaffold onto which TatC assembles. Also, we suggest that the deleterious effect of overproducing the TatB protein could be explained if the TatA/E and TatC proteins interact only when bound to a central TatB polypeptide. In this case, such an increase in TatB levels would make it unlikely that the other Tat proteins are bound to the same TatB molecule with a concomitant reduction in translocase activity. While we suspect that the TatB protein itself has a vital role in Tat-dependent transport, it is not advisable at this stage to exclude the possibility that the phenotype of a tatB null allele may be mediated entirely by the effect of this mutation on the stability of the TatC protein. However, it has not proven possible to demonstrate any suppression of the tatB strain mutant phenotype by increasing the levels of tatC transcript via a high copy number plasmid bearing only tatC (data not shown).

Distinct "TatA-like" and "TatB-like" Proteins on the Tat Protein Export Pathway?-- At least two genes coding for proteins of the Hcf106 family are coded in the complete genomes of all prokaryotes possessing the Tat system, with the exception of Rickettsia prowazekii which appears to encode but a single such protein. The bipartite division of function between Hcf106-like proteins found in E. coli is thus likely to be a general feature of Tat systems. We have tested this idea by heterologous complementation experiments using H. pylori 26695 Hcf106 homologues. In H. pylori, genes for two such proteins, HP0320 and HP1060, are found at separate chromosomal loci, with HP1060 apparently co-transcribed with a gene, HP1061, encoding a putative TatC homologue (26). The HP0320 and HP1060 genes, together with their putative promoter regions, were independently cloned into pBluescript and tested for their ability to restore the E. coli tatAtatE, pcnB1 (JARV16-P) or tatB, pcnB1 (BØD-P) strains to a TorA⁺ phenotype. The HP0320 gene was found to complement the tatAtatE, but not the tatB strain (Table IV), strongly suggesting that the HP0320 gene product is functionally "TatA-like" and further supporting our proposal that there are two classes of Hcf106 homologues with distinct roles in Tat systems.

Primary sequence analysis would appear to predict that the HP1060 gene should encode a polypeptide more closely related to E. coli TatB rather than TatA/E. However, in contrast to the results presented for the HP0320 gene, HP1060 was unable to complement either E. coli mutant (Table IV). Thus if HP1060 does indeed code for a TatB-like protein, then the sequence constraints placed on such proteins are apparently more stringent than those for the TatA type.

In the present study we suggest that there are two functionally distinct classes of Hcf106-like proteins: "TatA class" and "TatB class." The exact roles of these two classes in Tat-dependent transport are unclear, but we have been unable to obtain any evidence that would support the idea that the two classes of Hcf106-like proteins interact with different sets of substrate molecules. Our data suggest that TatB class proteins may form a specific complex with TatC proteins, and that the sequence constraints on the TatA class proteins are considerably more relaxed than those on the TatB class. Therefore is it possible, using the large amount of sequence data available for putative Hcf106 homologues encoded by bacterial genomes, to define amino acid sequence features that can distinguish the two functional classes?

Such features can indeed be identified for the obviously TatA class and TatB class proteins from E. coli and other Proteobacteria (for example Hemophilus influenzae, Azotobacter chroococcum, and H. pylori) in which the gene encoding the TatB class protein is directly linked to, and probably co-transcribed with, the gene coding for TatC. In these organisms, only the TatB class proteins have the essential proline residue (Pro²² in E. coli TatB) identified by Weiner et al. (9). This residue, together with an invariant preceding glycine residue (Gly²¹ in E. coli TatB), may well form a flexible hinge region between the predicted N-terminal transmembrane helix and the following predicted amphipathic helix. Interestingly, the bacterial TatA class proteins also harbor this invariant glycine residue (Gly²¹ in E. coli TatA and Gly²¹ in E. coli TatE). In the case of TatA class proteins, however, this glycine is not followed by a proline but instead preceded by a conserved phenylalanine residue (Phe²⁰ in E. coli TatA) and then followed by a second conserved phenylalanine in the predicted amphipathic helix (Phe³⁹ of E. coli TatA). Neither phenylalanine residue is conserved in the TatB class protein sequences. Most intriguingly, the predicted N-terminal transmembrane helix of TatB class proteins contains a conserved charged residue, Glu⁸ of E. coli TatB, which we speculate may be involved, for example, in proton gating during the translocation event or perhaps in maintaining a strong salt-bridge interaction with TatC. Finally, the C-terminal domain of TatB class proteins would appear to be far longer than that of TatA class proteins and encompasses a greatly extended amphipathic region with a string of glutamic acid residues on the polar face. Clearly extensive site-directed mutagenic studies will be required to define further the biological roles of TatA class and TatB class proteins.

In conclusion, however, we concede that none of the above criteria is consistently able to divide the multiple Hcf106 family proteins of each non-Proteobacterial organism into two separate classes. It is therefore conceivable that more than one type of sequence, or combination of sequences, can be used to form each of the two Hcf106 structures required for the operation of the Tat apparatus.

	ACKNOWLEDGEMENTS

We thank Dr. Gary Sawers for valuable discussions and advice, and critical reading of the manuscript. We are also grateful to Dr. Long-Fei Wu for comments on the data presented in this work and for providing strain MCMTA. We acknowledge Prof. David H. Boxer for providing antisera to formate dehydrogenase-N and to hydrogenase-2 and we thank Dr. Erik H. Manting for the cold mild osmotic shock protocol for recovery of periplasmic proteins. Finally, we acknowledge Dr. Kim Hardie, Dr. Nicky Hughes, and Dr. Dave Kelly for the kind gifts of H. pylori chromosomal DNA.

	FOOTNOTES

^* This research was supported by a Royal Society University Research Fellowship (to T. P.) and Biotechnology and Biological Sciences Research Council Grants B04897 and 88/P09634.The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

^¶ Recipient of Norwich Research Park Studentship.

To whom correspondence should be addressed: Dept. of Molecular Microbiology, John Innes Centre, Norwich NR4 7UH, United Kingdom. Tel.: 44-1603-456-900 (ext. 2726); Fax: 44-1603-454-970; E-mail: tracy.palmer@bbsrc.ac.uk.

¹ Also termed MttA2 (revised nomenclature in GenBank accession number AF067848). Note that this reading frame was incorrectly assigned in the original report (9).

³ N. R. Stanley, T. Palmer, and B. C. Berks, manuscript in preparation.

⁴ M. Wexler, F. Sargent, R. L. Jack, N. R. Stanley, E. G. Bogsch, C. Robinson, B. C. Berks, and T. Palmer, submitted for publication.

	ABBREVIATIONS

The abbreviations used are: TMAO, trimethylamine N-oxide; PCR, polymerase chain reaction; bp, base pair(s); CR, medium of Cohen and Rickenberg.

	REFERENCES

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS AND DISCUSSION REFERENCES