Investigation of Escherichia coli Dimethyl Sulfoxide Reductase Assembly and Processing in Strains Defective for the sec-Independent Protein Translocation System Membrane Targeting and Translocation*

Dimethyl sulfoxide reductase is a heterotrimeric enzyme (DmsABC) localized to the cytoplasmic surface of the inner membrane. Targeting of the DmsA and DmsB catalytic subunits to the membrane requires the membrane targeting and translocation (Mtt) system. The DmsAB dimer is a member of a family of extrinsic, cytoplasmic facing membrane subunits that require Mtt in order to assemble on the membrane. We show that the MttA2, MttB, and presumably MttA1 but not the MttC proteins are required for targeting DmsAB to the membrane. Unlike other Mtt substrates such as trimethylamine N-oxide reductase, the soluble cytoplasmic DmsAB dimer that accumulates in the mttdeletions is very labile. Deletion of the mttA 2or mttB genes also prevents anaerobic growth on fumarate even though fumarate reductase does not require Mtt for assembly. This was due to the lethality of membrane insertion of DmsC in the absence of the DmsAB subunits. In the absence of DmsC, DmsAB accumulates in the cytoplasm. A 45-amino acid leader on DmsA is removed during assembly. Processing does not require DmsC but does require Mtt. Translocation of DmsAB to the periplasm is not required for processing. The leader may be cleaved by a novel leader peptidase, or the long DmsA leader may traverse the membrane through the Mtt system resulting in cleavage by the periplasmic leader peptidase I followed by release of DmsA into the cytoplasm.

Until recently, protein translocation in bacteria was thought to take place either by the sec pathway or by specialized translocation systems (1). The translocation of proteins by the sec pathway requires the protein to remain in an open conformation during the process of translocation and extensive information exits for sec-mediated translocation (2, 3). The sec system has also been implicated in the translocation of selected periplasmic membrane-extrinsic and membrane-intrinsic proteins. For example, leader peptidase is targeted by the sec pathway (4), and a newly described protein, YidC (a component of sec-apparatus), was shown to target specifically membraneintrinsic proteins (5)(6)(7). We and others (8 -11) have reported an alternative protein targeting and translocation system termed membrane targeting and translocation (Mtt), 1 also termed twin arginine translocase. The Mtt system is comprised of at least three proteins, MttA 1 A 2 B (also called TatABC) and has been shown to transport fully folded proteins to or across the membrane (8 -10, 12). Unlike the sec system, no known integral membrane protein is targeted by the Mtt pathway (13). The discovery of the Mtt system explained the translocation into the periplasm of cofactor-containing proteins that were assembled in the cytoplasmic compartment. These proteins have a long N-terminal leader with a conserved twin arginine motif ((S/T)RRXF(X/L)K) (11). They have diversity in subunit composition, molecular weight, the nature of the redox cofactors, and cellular localization (13)(14)(15). Bacterial proteins utilizing the Mtt pathway include some proteins without cofactors such as SufI, a member of the copper oxidase family, fusion proteins containing a reporter gene (e.g. ␤-lactamase), and even malfolded proteins (9, 16 -19). Components of the Mtt protein translocation machinery are found in at least half of the complete bacterial genomes available to date as well as the genomes of chloroplasts and plant mitochondria. However, Mtt is totally absent from animal genomes (8,11,13).
Dimethyl sulfoxide (Me 2 SO) reductase is encoded by the dms operon and is expressed under anaerobic growth conditions in the absence of nitrate (20,21). Extensive investigation has shown that the DmsAB subunits form a membrane extrinsic dimer facing the cytoplasm, and DmsC, an integral membrane protein, serves as a membrane anchor (22). Recently, we have confirmed that DmsA has an N-terminal twin arginine leader that functions as a membrane-targeting signal and is also essential for the stability of the holoenzyme (15).
Membrane targeting via the Mtt system is not limited to Me 2 SO reductase. Formate dehydrogenase-O is targeted to the membrane (but not translocated across) by a twin arginine leader (23). More recently, chlorophenol reductive dehalogenase (encoded by cprAB genes) of Desulfitobacterium dehalogenase was shown to be a membrane-bound enzyme, apparently targeted to the membrane by the twin arginine leader. CprA undergoes processing resulting in cleavage of the leader, and the mature form of the enzyme was localized to the cytoplasmic face of the membrane with the CprB subunit presumably serving a membrane anchor function (24). Thus certain membrane-bound enzymes whose active sites face the cytoplasm appear to be targeted to the membrane by the Mtt system (15,23,24).
To better understand the mechanism of the membrane targeting of Me 2 SO reductase, we have constructed several mutant strains defective in the mtt genes. We provide experimental evidence to show that Mtt catalyzes the targeting of DmsAB to the cytoplasmic face of the membrane, but not its translocation, and that a functional Mtt system is required for processing of DmsA to the mature form and for the stability of the DmsAB dimer.

EXPERIMENTAL PROCEDURES
Materials-Molecular biology reagents and ECL Western blotting detection kit were purchased from Life Technologies, Inc., and Amersham Pharmacia Biotech, respectively. Oligonucleotides were obtained from the Department of Biochemistry DNA core facility, University of Alberta, Edmonton, Canada. DNA polymerases (Taq and Elongase enzyme mix) were purchased from Life Technologies, Inc. Pfu DNA polymerase was purchased from Stratagene. All other reagents used were of the highest purity available commercially.
Methods-Media, growth conditions, anaerobic growth profiles on minimal media composed of glycerol as a carbon source and Me 2 SO, trimethylamine N-oxide (TMAO), or fumarate as the terminal electron acceptor were as described previously (15,25). Procedures for restriction enzyme digestions, agarose gel electrophoresis, elution of DNA fragments from agarose gels, and separation of proteins using SDS-gel electrophoresis, phage P1 transduction, construction of deletion strains using the method of homologous recombination were carried out by standard procedures (25,26). Osmotic shock methods for preparing the periplasmic fraction and preparation of soluble and everted membrane vesicles by French press lysis were as described earlier (15,27). Protein was estimated by a modification of the Lowry procedure (28). Electrophoresis of proteins was carried out in 7.5 or 10% SDS-polyacrylamide gels, and proteins were transferred to nitrocellulose filters for Western blotting analysis using antibodies to the DmsA, DmsB, and ␤-lactamase proteins (22,29). Me 2 SO reductase activity was monitored by the oxidation of reduced benzyl viologen by the substrate TMAO. Fumarate reductase activity was monitored in an assay system similar to the Me 2 SO reductase, except fumarate was used as a substrate (22,30).
Bacterial Strains and Plasmids-Bacterial strains and plasmids used are described in Table I. Construction of plasmids pDMS190 and pDMS160 carrying the entire dms operon under the control of the lac or the native dms promoter, respectively, has been described previously (15,31).
Construction of Plasmids-Procedures for the cloning of the entire mtt operon (mttA 1 A 2 BC) into the vectors pMS119EH (Amp R ), pTZ19R (Amp R ), or pBR322 (Tet R ) and cloning of the mttB gene into pMS119EH were as described below and in the legend to Table II. The relevant mtt gene sequences were amplified using PCR and chromosomal DNA from a wild-type strain (HB101) as a template with the commercially supplied buffers (Life Technologies, Inc.) and Taq, Pfu, or Elongase DNA polymerases in a thermocycler unit programmed for optimum temperatures for annealing and extension for the individual gene amplification reactions (32). The entire mtt operon was amplified using the oligonucleotide pairs RAT20/GZ-01 (see Table II for oligonucleotide sequences) with flanking EcoRI and SalI restriction enzyme sites at the 5Ј and 3Ј ends, respectively. The PCR product (2544 bp) was cloned into the EcoRI and SalI sites of pMS119EH or pTZ19R to yield the plasmids pMTT(Amp R ) and pMTT/19R (Amp R ), respectively. These plasmids carry an 82-bp (5Ј end) and a 32-bp (3Ј end) flanking sequence outside of the mttABC. The plasmid carrying the mttA 1 A 2 BC sequences under control of the native mtt promoter were amplified by PCR using the oligonucleotide pairs GZ-08/GZ-07 with an engineered EcoRI site at the 5Ј end. The PCR DNA (3596 bp) was digested with EcoRI and PstI enzymes, respectively, to yield 3375 bp of the mtt gene sequence. The PstI site is present within the 3Ј end of the amplified DNA, outside of the mtt operon. The EcoRI/ PstI fragment was cloned into EcoRI/PstI-cleaved pBR322 resulting in deletion of the bulk of the ampicillin coding region of the vector to yield the plasmid pMTT-322 (Tet R ). This plasmid also differs from pMTT or pMTT/19R in that it has an extended 5Ј region (588 bp upstream of the ATG start codon of MttA 1 ) and 3Ј sequences (356 bp) downstream of the mtt operon. Plasmid pMTTB gene was amplified by PCR using the oligonucleotides GZ-04/RAT22. The PCR DNA (827 bp) was cut with EcoRI and SalI prior to cloning into pMS119EH to generate the plasmid pMTTB.
Plasmid pDMS193 encodes the DmsAB dimer under control of the dms promoter and was constructed using PCR with primers DS-23/ DS-24 and pDMS160 as template DNA to amplify a 188-bp DNA fragment that covers the 3Ј end of dmsB with an engineered stop codon and flanking SstI and SalI/XhoI restriction enzyme sites at the 5Ј and 3Ј ends, respectively. The plasmid pDMS190 was digested with SstI and SalI to remove part of 3Ј end of dmsB and the entire dmsC (1073 bp). The DNA was gel-eluted, and the larger fragment (7494 bp) containing the entire dmsA and most of dmsB was mixed with the PCR DNA and ligated to yield the plasmid, pDMS193.
Deletion or Insertional Inactivation of the mtt Genes-The mttA 2 gene was inactivated by insertional mutagenesis. The plasmid pMTT/19R was cut with SacII at the unique site within mttA 2 . The recessed ends were filled in to create blunt ends. A blunt-ended chloramphenicol acetyltransferase (Cm R ) gene cartridge, CAT, generated using PCR (see Tet R mttA1A2BC, pBR322, native promoter This study a DSS640 represents a total deletion of mtt operon; DSS641, DSS642, DSS643, and DSS644 represent deletion in the first (mttA 1 ), second (mttA 2 ), third (mttB), and fourth (mttC) genes of the mtt operon, respectively.
b Double deletions are designated as DSS740 (total deletion of mtt and dms operons). DSS743 and DSS744 indicate the deletion of the dms operon plus the third or the fourth genes of mtt operon (mttB or mttC), respectively. below) was inserted at the filled in SacII site of the pMTT/19R plasmid to generate p⌬MTTA 2 /CAT. The Cm R gene sequence (CAT) was amplified using pACYC184 DNA as template and the oligonucleotides CAT5/ CAT3 (Table II). The CAT cartridge has BamHI-flanking ends and was also used to generate various mtt gene deletions as summarized below and in Table II.  Table I. A, TG1 versus DSS641 (⌬mttA 1 ) and DSS642 (⌬mttA 2 ). B, TG1 versus DSS644 (⌬mtt-C). C, TG1 versus DSS640 (⌬mttA 1 A 2 BC) and DSS640/pMTT. D, TG1 versus DSS643 (⌬mttB) and complementation of DSS643/pMTTB.
a Non-homologous sequences are shown in lowercase, and the restriction enzyme sites are underlined. b CAT5/CAT3-amplified DNA was used in combination with the PCR-amplified sequences to generate the plasmids deleted in the individual mtt genes as described.
d The PCR-amplified DNA from GZ-14/GZ-07 (3Ј region of mttC gene and the downstream sequences, 918 bp) was cut with BamHI and cloned along with GZ-08/GZ-09 (upstream region of mttA 1 gene, 768 bp) and a CAT cartridge to generate the p⌬MTT/CAT plasmid in a pTZ19R vector background that has a 1912-bp deletion of the mtt operon, missing in the 3Ј mttA 1 , complete mttA 2 B, and the bulk of the mttC genes, respectively. e The PCR-amplified DNA from GZ-08/GZ-11 (1650-bp DNA with the mtt promoter region, mttA 1 A 2 , and the 230 bp in 5Ј mttB) and GZ-12/GZ-07 (1674 bp, 3Ј region of mttB, and the downstream sequences) were cut with appropriate restriction enzymes, ligated to a CAT cartridge, and cloned in to pTZ19R to generate a plasmid, p⌬MTTB/CAT, with a 274-bp deletion in the mttB gene.
f The PCR-amplified DNA from GZ-08/GZ-13 (2405 bp, is composed of the promoter region along with the mttA 1 A 2 B and the 5Ј sequence of the mttC gene) was cloned with a CAT cartridge into the gel-purified p⌬MTT/CAT vector (deleted for the EcoRI and BamHI-intervening sequences from the previous cloning step) to generate a plasmid, p⌬MTTC/CAT, with a 274-bp deletion in mttC.

Characterization of mtt Deletion
Strains-Deletions in mttA 2 , mttB, and mttA 1 A 2 BC failed to support anaerobic growth on GD (15,25) (Fig. 1, A, C, and D) or GT (15,25) minimal medium (data not shown). Deletion of mttA 1 showed moderate inhibition of growth compared with the wild-type strain, TG1 (Fig. 1A). This was shown to be due to a compensating effect of a functional homologue of MttA 1 , TatE, present on Escherichia coli chromosome (9). Deletion of mttC alone had very little effect on the growth profiles measured under our experimental conditions (Fig. 1B). Growth on glycerol/Me 2 SO medium was restored when the Mtt polypeptides were expressed from a multicopy plasmid pMTT carrying the entire mtt operon in the ⌬mtt strain (Fig. 1C). Similarly, growth of the ⌬mttB strain was corrected by a plasmid (pMTTB) expressing the MttB polypeptide (Fig. 1D). The defective growth phenotype observed for the strain deleted in mttA 2 was also corrected by a plasmid expressing the Mtt polypeptides (data not shown).
These results clearly demonstrate that the mttA 2 and mttB genes are critical for anaerobic respiration on Me 2 SO.
As a control for the growth experiments, we investigated the growth of our mtt mutants on GF (15, 25) minimal medium. None of the fumarate reductase polypeptides bear a twin arginine leader, and this enzyme is not targeted to the membrane by the Mtt system, thus growth on GF medium should be unaffected in these mutants. Surprisingly, ⌬mttA 1 and ⌬mttA 2 showed limited growth, whereas ⌬mttB and a total deletion of the mtt operon (⌬mtt) nearly abolished growth on GF medium (Fig. 2, A and B). Anaerobic growth on GF medium was restored by in vivo complementation of the ⌬mttB strain by pMTTB (Fig. 2B); similarly the ⌬mtt and ⌬mttA 2 strains were complemented by expression plasmids carrying the entire mtt operon (data not shown). The mtt deletions grew normally under aerobic conditions and anaerobically with nitrate (data not shown).
We hypothesized that the lack of growth on GF medium resulted from the insertion of DmsC into the membrane without the DmsAB subunits. Previously, we showed that expression of DmsC in the absence of DmsAB is lethal (34). In confirmation of this hypothesis growth was observed on GF medium for the double mutants DSS740 and DSS743 (⌬mtt ⌬dms and ⌬mttB ⌬dms, respectively) (Fig. 2C) which was similar to the parent strain TG1 (Fig. 2A) or a ⌬dms (DSS301) strain (Fig. 2C). Predictably, introduction of the DmsABC expression plasmid (pDMS160) into the double mutant, ⌬mtt ⌬dms (DSS740), suppressed growth on fumarate (Fig. 2C), whereas the expression of the DmsAB subunits without DmsC (pDMS193) did not suppress growth (data not shown).
Comparison of Me 2 SO Reductase Activities in the Control and the mtt Deletion Strains-To correlate growth to the mem- FIG. 2. Anaerobic growth of E. coli on glycerol/fumarate minimal medium was measured as described in Fig. 1. A, TG1 versus DSS641 (⌬mttA 1 ) and DSS642 (⌬mttA 2 ). B, TG1 versus DSS640 (⌬mttA 1 A 2 BC), DSS643 (⌬mttB), and DSS643/pMTTB. C, DSS301 (⌬dms) versus the double mutants, DSS740 (⌬dms, ⌬mtt), DSS743 (⌬dms, ⌬mttB), and DSS740/pDMS160. brane association of the reductase activity, we measured enzyme activities with the artificial electron donor, benzyl viologen with TMAO as a substrate. As expected, the control strain TG1 showed up to 94% of the activity in the membrane fraction (Table III). The reductase activity in the mutants DSS640 (⌬mtt), DSS642 (⌬mttA 2 ), and DSS643 (⌬mttB) was predominantly soluble. Complementation of the deletions with appropriate plasmids resulted in membrane-bound reductase activity. ⌬mttA 1 (DSS641) and ⌬mttC (DSS644) exhibited profiles close to the control strain. The distribution and the specific activity of fumarate reductase remained unaffected indicating that this anaerobic enzyme was not influenced by the Mtt pathway (data not shown).
Stability of Me 2 SO and TMAO Reductases in the mtt Deletions-The total Me 2 SO reductase activity in ⌬mttA 2 and ⌬mttB was greatly reduced, compared with the control, DSS641 and DSS644 (Fig. 3A). By comparison the periplasmic enzyme TMAO reductase was stable in these mutants even though it accumulated in the cytoplasm (Fig. 3B). We used DSS301 as the strain for comparison of TMAO reductase activity because this strain has a deletion in the dms operon and thus Me 2 SO reductase does not interfere with the benzyl viologen-TMAO assay. Upon in vivo complementation, the specific activity as well as the total activity of Me 2 SO reductase approached the control values.
Localization of DmsAB-As a precursor to examining the processing of the DmsA leader, we felt that it was essential to re-examine the cytoplasmic localization of DmsAB. Eliminating the anchor DmsC should simplify interpretation as expression of DmsAB closely mirrors a typical dual subunit Mtt substrate such as the E. coli hydrogenase-2. Such an experiment was reported earlier, with a construct that encoded DmsAB and the first two transmembrane loops of DmsC (pDMSC59X) (8). Expression from this construct showed some accumulation of soluble DmsAB dimer in the periplasmic compartment (8) leading us to propose that DmsC served a "stoptransfer" role. However, as the experiment lacked adequate controls for lysis and the quality of the periplasmic fraction and to eliminate potential problems arising from expression of a partial DmsC subunit, we constructed a plasmid encoding only DmsAB (pDMS193). Localization was studied in the control strain (DSS301) and the mtt deletions by immunoblotting of appropriate fractions (Fig. 4). As an internal control, the plasmid encoded ␤-lactamase was also monitored in these same fractions loaded identically on a separate gel and transferred onto a blot for probing with ␤-lactamase-specific antibody (Fig.  4). ␤-Lactamase is a known sec-dependent and periplasmically localized enzyme. The DmsAB subunits encoded by the plasmid pDMS193 in the control strain (DSS301) were predominantly cytoplasmic. The membrane fraction had a very small amount of DmsAB, presumably due to the contamination from occluded soluble reductase in the membrane pellet. The periplasmic fraction was almost devoid of DmsAB. In these samples ␤-lactamase was predominantly in the periplasmic fraction. A small amount of the ␤-lactamase was noted in the cytoplasmic fraction and none in the membrane fraction. These findings clearly indicate that the DmsAB subunits are present only in the cytoplasm. A soluble cytoplasmic enzyme marker (glucose-6phosphate dehydrogenase) was also monitored in these fractions (35). Glucose-6-phosphate dehydrogenase was present only in the cytoplasmic fraction, also validating the quality of the extracts prepared (data not shown). As expected, the DmsAB subunits were localized to the cytoplasmic fraction in the ⌬mtt strain. The ␤-lactamase (Fig. 4) and the glucose-6phosphate dehydrogenase (data not shown) were present in the expected cell compartments. ␤-Lactamase export was not affected by deletion of the mtt genes. These experiments were carried out in a strain totally lacking DmsC (DSS301/ pDMS193) and a strain with sub-stoichiometric levels of DmsC relative to DmsAB (DSS640/pDMS193). However, it is expected that in strain DSS640/pDMS193 the DmsAB dimer would accumulate in the cytoplasm, regardless of the level of  3. Comparison of the stabilities of TMAO and Me 2 SO reductases in the mtt deletions. A, bacteria were grown in peptone/ fumarate medium (TG1, DSS640, DSS641, DSS642, DSS643, and DSS644) or B, peptone/fumarate/TMAO medium (DSS301, DSS740, and DSS744). Membrane and supernatant (periplasm ϩ cytoplasm) fractions were prepared and assayed for enzyme activity as described under "Experimental Procedures," using TMAO as substrate for both TMAO and Me 2 SO reductases. The combined total activity units in the membrane and supernatant fractions were used for the histogram.
the DmsC expression from the chromosomal copy of the dms operon.
Processing of the Twin Arginine Leader Sequence of DmsA in the mtt Deletions-The DmsA leader sequence is critical for expression, membrane targeting, and stability of the reductase (15). We examined the processing of DmsA by immunoblot analysis in whole cell lysates derived from the mtt deletions (Fig. 5). Whole cell lysates closely reflect the levels of the reductase under various conditions of growth and expression and minimize artifactual degradation resulting from preparation of membrane and soluble fractions. DSS301/pDMS190 and DSS644/pDMS190 showed good processing with the majority of DmsA migrating at a position corresponding to the mature form based on the mobility of the purified mature protein (lane with Me 2 SO reductase standard). A very faint slower migrating band was also observed in these strains corresponding to the precursor form of DmsA. On the other hand, DSS640/pDMS190 and DSS643/pDMS190 showed predominantly the precursor form of DmsA. The expression levels of DmsA were similar or slightly higher in all the strains tested, indicating that the mtt deletions did not impair the expression of the reductase. The presence of only the precursor form of the reductase in DSS640 and DSS643 confirms that at a minimum MttA 2 and MttB are required for processing of the reductase.

DISCUSSION
Considerable information has accumulated on the role of Mtt in the translocation of periplasmic proteins. The necessity for MttA 1 , MttA 2 , and MttB but not MttC has been documented for TorA, Fdn-N, Hya 1, SufI, and YacK proteins (9,19,36). Me 2 SO reductase is a membrane-bound protein with the extrinsic DmsAB subunits facing the cytoplasm (22). Assembly of Me 2 SO reductase requires Mtt, and DmsA has a typical twin arginine leader that is essential (15). It was of interest to see if Me 2 SO reductase had the same requirements for Mtt proteins. We constructed a series of mtt chromosomal deletions and confirmed that Me 2 SO reductase requires MttA 2 and MttB using anaerobic growth measurements (Fig. 1). Deletion of MttA 1 had only a marginal effect on growth. This is due to the redundancy of MttA 1 resulting from the presence of the homologous gene, ybec (tatE), on the E. coli chromosome. A strain deleted for tatA (mttA 1 ) and tatE failed to direct the Me 2 SO reductase to the membrane implying a role for these proteins (9). MttC was not required, and MttC has been shown to be a protein exhibiting nuclease activity and is apparently unrelated to the Mtt system (37). We have noted that amplification of Me 2 SO reductase in an mttC deletion strain showed only 50% of the membrane-bound reductase compared with the wild type. We assume that in the mttC deletion, the mtt mRNA may be less stable resulting in the lower activity measurements. Complementation of the mutants with appropriate mtt genes on plasmids corrected the phenotype. However, we have observed a significant lag in the complemented strains (Fig. 1). Overexpression of the Mtt proteins from multicopy plasmids hinders growth and in some instances may totally inhibit cell growth (36). These studies confirm and extend the observations of others (9,36) who have shown that the growth defects on Me 2 SO and TMAO are due to a generalized defect in the Mtt system.
As a control for the growth studies, we examined the ability of the mtt deletions to grow on TMAO, fumarate, and nitrate. TMAO reductase is a soluble periplasmic reductase, with a twin arginine leader, that utilizes Mtt for translocation. Growth and activity results with this enzyme reflected the Me 2 SO reductase data (data not shown). Fumarate and nitrate reductase are membrane-bound enzymes that lack twin arginine leaders and do not utilize Mtt. We were surprised to find that the mtt deletions (⌬mtt and ⌬mttB) failed to grow on fumarate (Fig. 2) but grew normally on nitrate (data not shown). By use of a double deletion for both dms and mtt, we traced the phenotype to the expression of incorrectly assembled Me 2 SO reductase in the mtt deletions. Expression of DmsC, the Me 2 SO reductase anchor subunit in the absence of correctly assembled DmsAB catalytic dimer is lethal (34). As expected, deletion of both the dms operon and the mtt operon restored growth on fumarate. As nitrate represses the expression of dmsABC, growth on nitrate was unaffected.  The enzyme activity distribution of Me 2 SO reductase (Table  III) and TMAO reductase (9) also confirms that the proteins are mislocalized in strains with deletions of the mtt operon. Me 2 SO reductase was very labile compared with TMAO reductase in these mutants (Fig. 3) suggesting rapid proteolytic degradation of mistargeted DmsAB.
A large amount of experimental evidence utilizing immunoblotting of periplasmic fractions obtained by osmotic shock and chloroform washing, protease susceptibility, lactoperoxidasecatalyzed iodination, immunogold electron microscopy of everted membrane preparations, TnphoA fusions, and electron paramagnetic resonance monitoring the effects of the probe dysprosium(III) on an engineered 3Fe4S cluster in DmsB indicated that DmsAB faces the cytoplasm (22,38). Others (11) have suggested that DmsAB faces the periplasm based solely on the presence of a twin arginine leader sequence in DmsA. In this study we have re-examined the question of the localization of DmsAB in both wild-type and mtt deletions (Fig. 4). Our results once again confirm that DmsAB is cytoplasmic.
In a recent study, the localization of DmsAB was compared in a strain expressing a partial DmsC subunit (pDMSC59X) with full-length DmsC originating from the chromosomal copy of the dms operon. In those experiments we reported (8) some DmsAB in the periplasm in the absence of full-length DmsC, leading us to suggest that DmsC might serve a stop-transfer role. In the current experiments, with a complete DmsC deletion, DmsAB subunits were only observed in the cytoplasm. It is possible that the absence of DmsC, or sub-stoichiometric levels of this polypeptide, could alter the localization of the DmsAB dimer. However, the large body of experimental evidence gathered to date supports the cytoplasmic localization of DmsAB in the wild-type and in the reductase-amplified strains (with stoichiometric amounts of DmsC) (22,38) as well as DmsC-deficient strains (Fig. 5). The earlier studies lacked compartment-specific marker enzyme controls for the fractionation protocols, and as a result a small amount of cell lysis and/or an increase in membrane leakiness due to expression of the partial DmsC subunit could have contributed to the observed DmsAB in the periplasm (8).
In the absence of the DmsC anchor subunit, one might expect that DmsAB should be translocated to the periplasm directed by the twin arginine leader. Why is DmsAB not translocated even in the absence of DmsC? It could be that the DmsA leader contains some stop-transfer information. This seems unlikely because in a previous study we showed that replacement of the dms leader with the tor leader did not affect the topology of DmsAB (15). It appears that DmsAB contains information in the mature polypeptides that prevents translocation to the periplasm. This conundrum is not limited to DmsAB. The localizations of FdoGH in E. coli (23) and CprAB in D. dehalogenase (24) were also shown to be cytoplasmic despite the presence of twin arginine leader sequences. Furthermore, N-acetylmuramyl-L-alanine amidase (AmiA), a twin arginine leader sequence containing protein, was shown to be targeted and translocated by an Mtt-independent mechanism (39). These studies imply that information in the total polypeptide and not solely the signal sequence determines the localization of a given protein. Elegant information on the importance of information in the mature protein for membrane targeting came from studies of a TorA-leader peptidase (TorA-Lep) fusion protein. The full-length TorA-Lep fusion protein was directed to the membrane via the Sec pathway and not by the Mtt system, even though a twin arginine leader was present (4). Conversely, the DmsA twin arginine leader was shown to mediate the export of a fusion protein (yeast cytochrome c) via the Mtt system (40). Taken together, these results favor the view that both the twin arginine leader and the mature polypeptide contribute to the final localization of a specific protein.
The results of whole cell immunoblotting (Fig. 5) clearly indicate that the MttA 1 , MttA 2 , and MttB proteins are required for processing of the DmsA leader. This was seen with both the catalytic DmsAB dimer and the holoenzyme indicating that the DmsC anchor does not play a role in processing of pre-DmsA to mature DmsA. It is not clear how the DmsA leader is processed. There is no evidence to date that any of the known Mtt subunits are directly involved in the processing of the leader sequence. It has been assumed that the leader peptide is cleaved by leader peptidase I which faces the periplasm (4). Attempts to study the processing of the DmsA leader using a conditional lethal leader peptidase mutant were inconclusive, due to the poor growth of this strain and the difficulty in obtaining total peptidase deficiency under our experimental conditions (data not shown). Since none of the substrates of the Mtt system studied to date are true soluble cytoplasmic proteins, the observation of apparent DmsA processing in the cytoplasm should be interpreted with caution. The DmsAB dimer attached to the membrane-bound Mtt system possesses a leader sequence of 45 amino acids that is sufficient to traverse the bilayer and mediate processing on the periplasmic side, without the bulk of the DmsAB ever crossing the bilayer. The processed DmsAB would then lack affinity for the Mtt system and could be released into the cytoplasm. Alternatively, the DmsA leader could be cleaved by a novel cytoplasmic protease. Interestingly, mutation of the two conserved alanines to asparagines at the DmsA leader cleavage site did not affect the membrane targeting or the processing (15).